Dopamine D2-receptor Antagonist Droperidol Deepens Sevoflurane Anesthesia

MIDBRAIN dopaminergic neurons are implicated in arousal-promoting pathways, as well as in the cholinergic, histaminergic, noradrenergic, and orexin networks.¹  Dopamine release by neurons in the ventral tegmental area was recently reported to induce emergence from general anesthesia via the dopamine D1 receptor.^2–4  The ventral periaqueductal gray matter also provides wake-active neurons, a major ascending dopaminergic pathway involved in arousal.^5,6  Dopamine is thus of particular interest when considering arousal responses and/or emergence from general anesthesia. However, few reports have investigated the effects of dopaminergic systems on maintenance or deepening of general anesthesia.

Several studies have reported that droperidol, a dopamine D2 receptor antagonist, reduces Bispectral Index (BIS) values during sevoflurane anesthesia in humans at lower doses than those required for neuroleptanesthesia. This has also been reported with spinal anesthesia combined with propofol sedation, although detailed information regarding the electroencephalogram changes are currently unknown.^7,8 

The majority of D2 receptors are found on postsynaptic nondopaminergic neurons, including γ-aminobutyric acid–mediated (GABAergic) neurons as heteroreceptors.⁹  These include dopamine receptors in the ventral mesencephalon, found on γ-aminobutyric acid (GABA)–secreting nondopaminergic neurons. Generally, dopamine acts on the D2 heteroreceptors of these GABAergic neurons, causing hyperpolarization, resulting in increased activity within the neural network by decreasing inhibitory GABA activity.^10,11  Since the dopaminergic pathway is linked to various regions of the brain, such as the thalamus, dorsal raphe, locus coeruleus, basal forebrain, ventrolateral preoptic nucleus, and prefrontal cortex, D2 receptor antagonism may directly influence general anesthesia via these neural networks.

GABAergic anesthetic agents promote α and δ-θ activity in the electroencephalogram, associated with neural network regulation and resonance of the thalamocortical and corticothalamic axons.^12–16  In the nonlinear reverberating system between the cortex and thalamus, the output signal from the reverberating circuit should self-modulate by reentering the system as an input signal. This self-modulation comprises intermodulation products (the signal component produced by multiplication of the reentrant signals), resulting in quadratic phase coupling between the wave components in a signal. We previously indicated that bicoherence (the normalized bispectrum) can reveal the reverberating components of the thalamocortical reverberant network, representing massive synchronizations by detecting the intermodulation products and quadratic phase coupling.^17,18  The GABAergic anesthetics propofol, sevoflurane, desflurane, and isoflurane all show a certain unified pattern in the α and δ-θ area in the bicoherence spectra depending on the level of anesthesia by manifesting thalamocortical oscillatory activity via GABAergic pathways.^19–22  Accordingly, bicoherence analysis is suitable for understanding the level of anesthesia related to GABAergic mechanisms.

This study examined the effect of a dosage of the D2 receptor antagonist droperidol, lower than that typically required for neuroleptanesthesia, on sevoflurane anesthesia by analyzing continuous electroencephalogram bicoherence before and after droperidol injection. We hypothesized that droperidol deepens sevoflurane anesthesia, and that it alters the electroencephalogram and bicoherence pattern by enhancing reverberating oscillations characteristic of a GABAergic mechanism.

This study was approved by the Institutional Review Board for Human Experiments at Kyoto Chubu Medical Center (Kyoto, Japan). We enrolled patients scheduled for general anesthesia in Kyoto Chubu Medical Center from January 2017 to March 2017, between the ages of 20 and 60 yr. Patients were excluded if they had a history of neurologic disease and/or QT interval prolongation (QTc ≥ 0.45). Two cases were excluded because surgery commenced before depth of anesthesia could be assessed in accordance with our protocol. We were able to analyze data from 45 patients (22 men, 23 women). According to our previous studies concerning the similar anesthesia level,^17,21–23  the SD of α bicoherence was at most 15; a sample size of 45 subjects per group is therefore adequate for statistical significance with a power of 0.8 and a significance level of 0.05. Mean age (SD) was 39 (3) yr, with age ranging between 20 and 58 yr, and weight (SD) was 63 (15) kg. Written, informed consent was obtained from all patients before participation.

Patients did not receive premedication. Anesthesia was induced using propofol (1 mg/kg) combined with sevoflurane (end-tidal concentration, 2%), remifentanil (0.3 µg · kg^–1 · min^–1), and fentanyl (2 μg/kg). Rocuronium was administered to obtain appropriate muscle relaxation. All patients were intubated and ventilated. Immediately after tracheal intubation, the sevoflurane end-tidal concentration and remifentanil infusion rate were changed to 1% and to 0.1 μg · kg^–1 · min^–1, respectively. Anesthesia was maintained with sevoflurane and remifentanil at the same dosages. A BIS sensor (Quatro sensor, Aspect Medical Systems, USA) was mounted on the forehead and connected to an A-2000 BIS monitor (version 4.0, Aspect Medical Systems). Raw electroencephalograms (converted from analog to digital at 128 Hz) were continuously sampled from an A-2000 BIS monitor for offline analysis using Bispectrum Analyzer (BSA) version 3.22B2 software via an RS232 interface on a personal computer (CF-MX3, Panasonic, Japan).²⁴ 

Approximately 15 min later, after the end-tidal sevoflurane concentration had been deliberately kept constant (1%), intravenous droperidol (0.05 mg/kg) was administered by bolus. The dose of droperidol (0.05 mg/kg) in this protocol was determined by our previous clinical experience. In our hospital, we have conventionally used a fixed dose of 2.5 mg droperidol in patients without QT prolongation, mainly for antiemetic purposes. We noticed in these instances that the BIS index decreased after 2.5 mg of droperidol injection, which sparked our interest in this phenomenon. Accordingly, we used an approximately equivalent dosage of droperidol in this protocol. Electroencephalogram analyses (power spectrum analysis and bicoherence analysis) were retrospectively performed at just before and then 10 min after droperidol injection, from the collected electroencephalogram data. Systolic arterial blood pressure and heart rate were maintained at greater than 90 mmHg and greater than 44 beats/min, respectively, using phenylephrine and atropine as required. Data collection for analysis was completed before the start of surgery.

Electrode impedance was checked every 10 min and was maintained at 5 kΩ or lower throughout the study using the A-2000 BIS monitor. Signals less than 0.5 Hz or greater than 50 Hz were excluded. The spectral edge frequency 95 (SEF95) and BIS index values were simultaneously collected with the A-2000 every 60 s. The power spectrum was calculated from the preceding electroencephalogram over 1-min periods at 0.5 Hz intervals using BSA. Average power and bicoherence in the δ-θ area were calculated in the respective spectra under 6 Hz.

Bicoherence values were computed in all pairs of frequencies between 0.5 and 20 Hz at 0.5 Hz intervals from the preceding 3-min period of artifact-free signals, and were represented as two-dimensional moving averages. These were calculated using nine points of bicoherence every 0.5 Hz, from 1.5 to 20.0 Hz, similar to the calculation method described in our previous studies.^17,21,23  The 3-min-long electroencephalogram signals were divided into a series of 2-s epochs, with each epoch overlapping by 75%, resulting in 360 epochs. After applying the Blackman window function to the respective epoch, bicoherence values were calculated using the following equations:

sum of absolute triple product:

j: epoch number

: complex value calculated by Fourier transformation of the jth epoch

: conjugate of

Computations were performed using MATLAB software (version 8.4.0.150421 [R2014b], MathWorks, USA). The bicoherence spectra were then presented around the diagonal lines (the same pairs of the frequencies), as shown in figure 1.

Comparison was made between the data collected before and then at 10 min after injection of droperidol from electroencephalogram parameters (BIS, SEF95; both frequencies causing the α power peak and the α bicoherence peak; α power peak and the α bicoherence peak). Nonparametric analysis (Wilcoxon signed-rank test) was performed as the distribution indices were not necessarily regarded as normal according to the Shapiro–Wilk normality test. Shifts of averaged δ-θ in the power (logarithmic form) and bicoherence spectra were analyzed using a paired t test, as the distributions were regarded as normal. Predictive Analysis Software (IBM SPSS Statistics version 23.0, SPSS, Japan) was used for statistical procedures. The results were presented as median (25th, 75th percentiles), or means (95% CI). The difference values of P < 0.05 were considered significant. The difference in medians with 95% CI was examined by the Hodges-Lehman method using R: A Language and Environment for Statistical Computing (R Core Team [2017], R version 3.4.2, R Foundation for Statistical Computing, Austria, http://www.R-project.org/).

Two representative time courses of power and bicoherence spectra before and after droperidol injection in a 47-yr-old woman and a 50-yr-old woman are shown in figures 2 and 3. The raw electroencephalograms and the corresponding power and bicoherence spectra in the respective patients at just before and about 10 min after injection of droperidol are also shown. Several minutes after droperidol injection, the electroencephalograms changed gradually, then reached approximately maximum effect: the α-bicoherence peak was augmented, the α peaks in both the power and bicoherence spectra shifted to the lower frequency, the δ power increased, and the peak bicoherence in the δ-θ area increased. The continuous augmentation of α-bicoherence and continuous shift of the α-bicoherence peak to the lower frequency induced by droperidol (figs. 2 and 3) are noteworthy, because these results suggest the changes are due to the functional modulation of the same reverberant source.

Figures 4 and 5 show the summary of electroencephalogram results in 45 cases. Approximately 10 min after droperidol injection, the electroencephalogram variables BIS and SEF95 were significantly decreased from 59 (57 to 60) and 18.0 (17.3 to 18.7) Hz to 42 (40 to 44) and 13.5 (13.0 to 14.1) Hz (mean [95% CI], P < 0.0001 for each), mean difference (95% CI), –17 (–18 to –15) and –4.5 (–4.8 to –4.1) Hz. Both α peak values in power and bicoherence were significantly augmented from 1.97 (1.26, 3.68) μV² and 30.3% (24.2%, 42.4%) to 3.07 (1.53, 5.26) μV² and 50.8% (41.7%, 55.2%) (median [25th, 75th percentiles], P = 0.002 for power, P < 0.0001 for bicoherence), Hodges-Lehman median difference (95% CI), 0.83 (–0.02 to 1.75) μV² and 15.8% (11.3 to 21.4%). Simultaneously, the frequency causing α peaks in the power and bicoherence spectra decreased significantly from 12.0 (11.0, 12.5) and 11.5 (11.0, 13.0) Hz to 10.5 (9.5, 11.0) and 10.5 (10.0, 11.0) Hz, respectively (P < 0.0001 for each), Hodges-Lehman median difference, –1.5 (–2.0 to –1.0) and –1.5 (–2.0 to –1.0) Hz. Averaged power and bicoherence in the δ-θ area increased conspicuously from 0.252 (0.167 to 0.336) log(μV²) and 17.2% (15.6 to 18.7%) to 0.680 (0.600 to 0.760) log(μV²) and 25.1% (23.0 to 27.3%) (P < 0.0001 for each), mean difference, 0.428 (0.363 to 0.494) log(μV²) and 8.0% (6.0 to 9.9%), respectively.

Figure 6 shows the 45 superimposed power and bicoherence spectra around diagonal lines at just before and approximately 10 min after droperidol injection, as the average with 95% CI. Again, α peaks in the bicoherence spectra increased. Simultaneously, both α peaks in the power and bicoherence spectra shifted to the lower frequencies. The averaged power and bicoherence in the δ-θ area increased.

The findings of this study indicate that droperidol increases α bicoherence while simultaneously decreasing the basal frequency of α oscillation. The power and bicoherence in the δ-θ area are also significantly increased by droperidol. In general, along with deepening GABAergic general anesthesia from light (defined as BIS of about 60, and/or 1% sevoflurane) to moderate levels (defined as BIS of about 40, and/or 2% sevoflurane), α-bicoherence is augmented and the frequency causing the α-bicoherence peak decreases. Slow wave activity increases simultaneously. At deep levels of anesthesia (defined as BIS of about 30, and/or 3% sevoflurane), α bicoherence finally decreases with a predominance of slow waves and further increases in δ-θ bicoherence.^17–22  The current findings are similar to the electroencephalogram changes induced by deepening light 1% sevoflurane anesthesia to a moderate 2% level, which mainly affects the GABAergic pathway.¹⁷  Although sevoflurane is known to bind to various other targets in the central nervous system, such as N-methyl-d-aspartate, glycine, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid, acetylcholine, potassium channels, and serotonin receptors, GABAergic inhibition is recognized as the dominant mechanism of action of sevoflurane anesthesia, because the electroencephalogram during sevoflurane anesthesia shows δ-θ and α oscillations similar to propofol.^25–28 

These coherent δ-θ and α oscillations are known to occur in thalamocortical oscillations chiefly via the GABAergic pathway. Bicoherence, the normalized bispectral analysis, is capable of tracking changes in these reentry systems by quantifying the quadratic phase coupling between signals using a one-channel cortical electroencephalogram.^17–22  Because droperidol augmented α bicoherence and δ-θ bicoherence while simultaneously shifting the α bicoherence peak to a lower frequency in a sustained and graded manner, the resultant electroencephalogram changes were recognized as being induced by modification of the same subcortical reverberant source, namely thalamic reticular and thalamocortical neurons. Thus, droperidol, a dopamine D2 receptor antagonist, appeared to enhance sevoflurane anesthesia, partly through the same neural pathway of GABAergic mechanisms.

Five dopamine receptor subtypes (D1 to D5) have been identified in mammals. They are G protein–coupled receptors divided into two major groups: the D1-like receptors (D1, D5) and the D2-like receptors (D2, D3, D4). While activation of D1-like receptors increases excitability through activating adenylyl cyclase and increasing cyclic adenosine monophosphate levels, D2-like receptors are inhibitory via suppression of calcium ion influx with coupling of inhibitory G protein coupled receptors and inhibition of adenylyl cyclase.^9,29 

D2 receptors exist as autoreceptors on the terminals of dopamine neurons themselves, playing a key role in regulating the dopaminergic system by providing feedback inhibition. However, the majority of D2 receptors are found as heteroreceptors on postsynaptic nondopaminergic neurons, including GABA neurons.⁹  Accordingly, we think that these D2 heteroreceptors may play a crucial role in the modulation of anesthesia induced by droperidol. Although the precise mechanism of droperidol’s anesthetic effect has not been well studied, D2 receptor antagonism can increase inhibitory tone in the neural network of midbrain dopaminergic neurons via facilitation of inhibitory GABAergic interneurons.^10,11  This may explain how droperidol increases the depth of sevoflurane anesthesia.

Besides its dopaminergic actions, droperidol may affect anesthesia via its antagonistic effect on α1-adrenergic receptors. The main noradrenergic nerve nucleus, the locus coeruleus, and other noradrenergic nuclei such as the medial septal area, the medial preoptic area, and the lateral hypothalamus, are known to exert potent arousal-promoting actions, where α1- and β-adrenergic receptors are known to have important roles with conjugation of complex subcortical network and cortical innervations.^30–32  Additionally, α1 as well as α2 adrenoceptors in the dorsal raphe nucleus modulate the release of serotonin, which also influences anesthesia.³³  Thus, α1 blockade by droperidol also modulates sleep, anesthesia, and arousal responses, which may be included in the current results as an indirect influence on the GABAergic pathway.

On the other hand, droperidol has been reported to induce adverse effects via its actions on the GABA-type A receptor itself, causing inhibition of the GABAergic response. Unwanted side effects, such as anxiety and dysphoria, were reported on emergence from droperidol neuroleptanesthesia, due to submaximal GABAergic inhibition.³⁴  Conversely, GABA has also recently been found to be released directly from dopaminergic axons, and may function as a cotransmitter in monoaminergic neurons in basal ganglia circuits.^35,36  Droperidol reportedly also has preferential nicotinic effects on neuronal nicotinic acetylcholine receptors, mediating neuroleptanesthesia.^34,37  Dopamine’s important role in learning and reinforcement has also been revealed in midbrain dopaminergic neurons through glutamatergic signaling via an isoform of the vesicular glutamate transporter, although the function of glutamatergic synaptic transmission by dopaminergic neurons remains unclear.^36,38,39  As such, dopamine acts on complex neural network systems via a large number of mechanisms. In rats, Tanifuji et al. reported that droperidol does not alter minimum alveolar concentration of general anesthetics, although minimum alveolar concentration expresses the immobility produced by inhaled anesthetics, and does not necessarily equate with brain activity.⁴⁰  Since droperidol has various influences on arousal, sedation, consciousness, emergence, and depth of anesthesia, we consider that the current results reflect a primary, complex, and integrated effect that may be easily influenced by environmental and other factors. The current study was conducted using a relatively light plane of sevoflurane anesthesia of about 1%, where the BIS was about 60.

The dose of droperidol used was considerably higher than the current recommended dosages for antiemetic purposes (0.625 to 1.25 mg), which are not considered to pose a significant clinical risk for arrhythmias.⁴¹  We used 2.5 mg of droperidol in a patient of 50 kg, as we have commonly used this dose for chiefly antiemetic purposes in our institution in patients without QT prolongation. At the time of writing, no arrhythmic complications related to droperidol have been documented at our hospital. Although our dose was much lower than that used for neuroleptanesthesia and sedation, we should have employed the lower antiemetic dosage if we wished only to provide prophylaxis against postoperative nausea and vomiting.

However, the remarkable electroencephalogram changes observed in this study were surprising, because lower doses of droperidol than those used for neuroleptanesthesia displayed a strong anesthetic effect. The simultaneous use of droperidol and GABAergic anesthetics such as sevoflurane and propofol may result in lower doses of GABAergic agents being needed to achieve the appropriate depth of anesthesia. This may have useful economic implications, as has been reported previously.⁴²  However, we were unable to obtain data later than 10 min after droperidol injection, because the initial depth of anesthesia was not maintained after surgery commenced. As such, we are unable to comment on how long the enhanced anesthetic effect of droperidol on sevoflurane anesthesia is maintained.

Droperidol, a dopamine D2 receptor antagonist, augmented both α and δ-θ bicoherences while shifting the α-bicoherence peaks to lower frequencies. There was a simultaneous increase in slow wave oscillation, suggesting that droperidol enhances the effect of sevoflurane anesthesia on the electroencephalogram, partly through GABAergic subcortical reverberating networks.

The authors thank Satoshi Hagihira, M.D., Ph.D. (Clinical Professor, Department of Anesthesiology, Kansai Medical University, Osaka, Japan), for supplying Bispectrum Analyzer (BSA) software.

Supported by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (JSPS KAKENHI No.15K10518), Tokyo, Japan.

The authors declare no competing interests.