Thiopental Uncouples Hippocampal and Cortical Synchronized Electroencephalograpbic Activity

The experimental protocols were approved by the Institutional Animal Care Committee at Stanford University and adhered to published guidelines of the National Institutes of Health, Society for Neuroscience, and American Physiological Society. One week before the thiopental infusion studies, four male Wistar rats (weighing 250–300 g) were anesthetized with isoflurane, and scalp tissue was locally infused with lidocaine. Rats were placed in a stereotaxic apparatus and cortical dural electrodes and hippocampal microelectrodes were chronically implanted. Stainless steel dural electrodes (approximately 1.0 K omega) were positioned to record occipitofrontal activity in a bipolar differential recording configuration. The occipital electrode was placed 1.8 mm posterior to bregma and 2.5 mm lateral to the mid-line. The frontal electrode was placed 1.4 mm anterior to bregma and 2.1 mm from the midline. Each cortical electrode was implanted through a 1-mm burr hole to contact the dura mater, and affixed to jeweler's screws placed in the skull. The hippocampal bipolar microelectrode was fabricated from (25-G) Teflon-insulated (du Pont, Wilmington, DE) Nichrome wire (Fredrik Hear Co., Brunswick, ME) with tips staggered by 600 micro meter and insulated to approximately 10 micro meter of the recording tip (1–2 M omega). They were implanted through a burr hole and cemented using acrylic adhesive affixed to jeweler's screws placed in the skull. Hippocampal electrodes were located on the contralateral side to cortical electrodes at the following stereotaxic coordinates (Figure 1): 3.5 mm posterior to bregma, 2.75 mm lateral to the midline, and the lowest tip placed at a depth of 2.4 mm ventral to the dural surface. These coordinates placed recording tips in the oriens layer of hippocampal CA 1 neurons and molecular layer of the dentate gyrus, and positions were optimized for EEG amplitude during implantation. [10].

After allowing 7–10 days for recovery from the electrode implants and a day before each animal was used for a thiopental study, rats were anesthetized with isoflurane to place intrajugular and femoral artery catheters. Catheters were flushed with a heparin solution to prevent coagulation. The next day, rats were connected to instrumentation via flexible wires, which permitted free movement within a 12 x 10 inch cage. The intrajugular catheter was connected via an infusion tubing to a calibrated syringe pump (Model 22, Harvard Apparatus, South Natick, MA) for administration of thiopental (10 mg *symbol* kg sup -1 *symbol* min sup -1). The femoral artery catheter was connected to a blood pressure transducer (MS 20, Electromedics, Englewood, CO) using a low dead volume 22-G T-tube. Heart rate and blood pressure were monitored and recorded on a polygraph (Model 7, Grass Instruments, Quincy, MA). The T-tube side arm was used to sample arterial blood for thiopental concentration determinations and blood gas analyses. Arterial samples (250 micro liter) were taken every 60 s during the thiopental infusion, and at longer time intervals after the infusion. Samples were replaced with 2.5 volumes of normal saline. Blood gases were measured at five times after the infusion and mild hypercapnia occurred (P^CO2increased from an average of 30 in control to 55 at the peak drug effect) but apnea was not observed, only a slowing in respiration. Oxygen saturation remained > 96% throughout, and was maintained with supplemental oxygen by mask after stage 2 burst suppression. Hematocrit values showed no change over time. Thiopental concentrations were determined using chromatography, as previously described [11]and blood gases were analyzed using a Corning Model 178 Analyzer (Medfield, MA). Core temperature was monitored and a heating pad and lamp were used to maintain normal body temperature (37 degrees C) throughout the experiment.

Two cortical and two hippocampal leads were connected using braided and shielded 30-G wire to suppress movement artifacts. They terminated, together with catheter tubing, on a connector mounted to the skull (Amphenol, Palo Alto, CA). Fine braided wire and thin catheter lines were used to allow free movement of each animal before and after the thiopental-induced anesthesia. Occipitofrontal cortical and hippocampal leads were differentially amplified (x10,000, Beckman Accutrace, San Diego, CA) to limit background noise. Signals were filtered (0.5–50 Hz, band-pass) and simultaneously recorded on FM magnetic tape (Digital PCM Recorder, Vetter, Phillipsburg, PA), amplified (10–20x, Brownlee Precision, San Jose, CA), and digitized at 200 Hz using DataWave Systems (Longmont, CO) software to control a Data Translation (DT2821, San Carlos, CA) analog to digital converter connected to a 486, 33 MHz computer (Gateway 2000, North Sioux City, SD).

During each experiment, 4-s epochs of cortical and hippocampal EEG were sampled continuously and used to construct on-line total power versus time graphs from fast Fourier transform (FFT) analysis (DataWave Systems). These data and graphs were used to determine when a 30-min period of stable baseline activity had occurred before beginning a thiopental infusion. Recordings were considered to be stable if animal movement was not associated with transient voltage artifacts and cortical power remained within a+/-0.05 micro Volt²range. Sensory stimuli, sample and infusion times were time-locked to the EEG data using time-stamped key strokes (approximately 10 s resolution). All data were stored on disk for subsequent analysis. For off-line analysis, DataWave Systems and Axum software (Trimetrix, Seattle, WA) were used to measure EEG amplitude, total FFT power, frequency at maximum FFT power, and FFT power for each of four frequency bands from 0.5–3.5 (delta), 3.5–7.5 (theta), and 7.5–11.5 (alpha) to 11.5–30 (beta) Hz. Single-epoch or 10-epoch averaged FFTs were used to construct power/frequency versus concentration/time profiles. Single-epoch FFTs were used when EEG signals were undergoing rapid changes in frequency during the thiopental infusion.

Event-time histograms were used to examine the synchronization of burst suppression events between cortex and hippocampus. Burst discharge events in time-locked cortical and hippocampal signals were detected using amplitude-time window discriminators in DataWave and the time of occurrence for each burst was calculated and stored as DataWave time-stamps. Cortical and hippocampal time-stamps were compared to determine whether a hippocampal event(s) had been detected within 500 ms of a cortical event. If so, the shortest time difference between a cortical and hippocampal event was calculated. If not, the cortical event was classified as uncorrelated. Time differences were then plotted using a peri-event histogram function in Axum with the cortical event serving as the reference (0 time). This analysis worked well when burst events were separated by at least 50 ms but was unable to correlate higher frequency events, especially during large amplitude delta activity and the transition from delta to burst suppression activity, because the window discriminators could not distinguish between some delta and burst events riding on delta oscillations. For this reason, analysis began after clear burst suppression activity was observed, with each burst separated by > 50 ms. Level 1 burst suppression was defined as the earliest 15-s period with bursts separated by 50 ms, and level 2 activity was defined as the last 30-s period before a 5-s period of isoelectric activity was detected. Thiopental infusions were terminated once this 5-s period of isoelectric activity was observed.

At the beginning and end of each experiment, a 200-micro Volt calibration signal at 10 Hz was applied to both EEG channels to ensure that amplification and filtering had remained stable.

Each rat was used for only one experiment and received a single infusion (10 mg *symbol* kg sup -1 *symbol* min sup -1 for approximately 5 min) of thiopental, to minimize acute tachyphylaxis and avoid chronic tolerance. Studies were initiated within a 1 h time interval on each day and rats were maintained on a constant 12/12 h light/dark cycle to ensure animals had experienced comparable baseline sleeping/waking cycles for control conditions before each experiment. Food and water were provided ad libitum. Before each experiment at least 30 min of baseline activity was acquired under freely moving conditions to ensure stable EEG signals were recorded, exhibiting minimal movement artifacts, and to examine behavioral-EEG relationships during movement and awake immobility. Behavioral-EEG correlates were not systematically investigated in this study, but were noted for comparison with the literature on rat theta EEG responses. [10,12,13]Once stable recordings were achieved, arterial blood sampling began and an intravenous infusion of thiopental was initiated and continued until a 5-s period of isoelectric activity had occurred. Startle responses (orienting head movements) to auditory (hand clap) and manual vibrissa stimulation using a cotton swab were monitored by visual inspection. Overt behavior (loss of righting reflex); spinal (tail pinch using a hemostat) and corneal reflexes were tested at 30–40-s intervals. Loss of righting reflex occurred when rats were unable to make postural adjustments after being manually placed on their backs. Tail pinch and corneal reflexes were measured as a movement or eye blink after stimuli within 3 s. Corneal stimulation consisted of a brief stroke with a handheld moistened cotton swab, alternating between right and left eyes. Neither stimulus intensity nor response amplitude were quantitatively measured. Noxious stimuli were tested after loss of righting reflex and began with the least noxious stimulus (tail pinch) determined from previous studies. [14,15]Response endpoints were time-locked to EEG data using computer key strokes to time-stamp their occurrence (+/- 2 s variability). Response time-stamps were measured off-line and time from the start of a thiopental infusion was calculated. Rats were monitored until recovery of EEG signals had occurred (approximately 8 h postinfusion), evidenced in a return to baseline levels in the total FFT power profile.

Thiopental plasma concentration-time profiles were characterized by a biexponential model defined by Equation 1: where Cp is the predicted plasma concentrations, R(t) is the infusion rate as a function of time,* denotes convolution, and Aⁱand alphaⁱare the coefficients and exponents of the unit impulse disposition function. The pharmacokinetic model was fitted to the data of each animal separately using the nonlinear least squares regression program NONMEM. A constant coefficient of variation model was used to characterize the residual error of the pharmacokinetic model fitted to the data.

Steady-state concentration-EEG effect relationships were derived for the total FFT power, and the power in the delta (0.5–3.5 Hz), theta (3.5–7.5), alpha (7.5–11.5 Hz), and beta (11.5–30 Hz) frequency bands for both the cortical and hippocampal EEG signals. The equilibration delay between plasma concentrations and EEG effect was modeled using the effect compartment approach. [16]The hypothetical effect site concentrations Ce (equal to steady-state plasma concentrations) were calculated from the following equation:Equation 2where k^eois first order rate constant describing the rate of equilibration between plasma and effect site concentrations. The value of k^eowas estimated by minimizing the hysteresis loop for effect site concentration versus effect curve. [17]A mean concentration-effect relationship was derived for each frequency band from the individual concentration-effect relationship for each animal. Unbound aqueous concentrations at the effect site were estimated based on an unbound plasma fraction of 10–25%. [18].

Hippocampal and cortical EEG signals showed a consistent relationship to the overt and startle response behaviors in freely moving rats. The theta rhythm (3.5–7.5 Hz) always was observed during awake voluntary movements and was seen whenever a rat was making postural adjustments, walking, or making exploratory head movements. Theta frequency activity was most prominent in the hippocampal signal (Figure 2) but also was the dominant rhythm in cortical signals when rats were exhibiting voluntary movement. Awake movement and startle responses to auditory stimuli also were characterized by increases in cortical fast activity (particularly in the 40–50 Hz band) and occurred concomitantly with a marked decrease in hippocampal low frequency activity (< 5.0 Hz, Figure 2). In general, this activation of the cortical EEG was observed after sensory stimuli, whereas hippocampal activation occurred during both sensory and movement related activity, in agreement with previous reports. [12,13].

During awake immobility and sensory arousal, cortical EEG signals exhibited a mixture of delta (0.5–3.5 Hz), theta (3.5–7.5 Hz), and alpha (7.5–11.5 Hz) frequency activity. Beta activity (11.5–30 Hz) was reduced during awake immobility. The average amplitude of the EEG was higher during awake immobility (122+/-8.3% compared with movement, P < 0.01 analysis of variance) and high amplitude sharp waves were evident in both cortical and hippocampal signals (Figure 2) during immobility. Sharp wave activity occurred independently in the hippocampus and cortex, because no temporal correlation between signals was obvious from event histograms (not shown). Theta activity seen during awake movement, in contrast, occurred simultaneously in cortical and hippocampal signals and was evident in averaged FFTs (Figure 2).

A continuum of concentration-dependent effects was observed during an infusion of thiopental. Plasma thiopental concentrations of 15–20 micro gram/ml were achieved within 1 min and the cortical EEG showed a pronounced increase in amplitude (240+/-9.5% of control; mean +/-SD; n = 4) across all frequency bands. The hippocampal signal also showed an increase in amplitude, but to a lesser extent (178 +/-8.6%) and lacking in higher frequency components compared with the cortical EEG (Figure 3). After 2 min (plasma thiopental concentration [nearly equal] 40 micro gram/ml), both the cortex and hippocampus exhibited high amplitude low frequency activity and short trains of high amplitude (364+/-12.5% above control) spindles were observed. Spindles occurred synchronously in the hippocampus and cortex (Figure 3) and appeared on a background of slow wave delta activity. Burst suppression activity was evident in both EEG signals within 4 min after the start of the thiopental infusion (plasma thiopental concentration greater or equal to 50 micro gram/ml). The cortical and hippocampal burst activity occurred synchronously and exhibited higher amplitudes (approximately 250%) compared to the largest sharp waves seen in control recordings (Figure 2and Figure 3). During the last minute of the infusion, burst suppression activity became more intermittent with longer periods of isoelectric activity separating each burst. By 5 min (plasma thiopental concentration greater or equal to 70 micro gram/ml), isoelectric periods of 5–6 s were observed in both the hippocampus and cortex. The cortical signal was more strongly depressed and small amplitude events could still be detected in the hippocampal signal. The progression of effects: activation > slowing > burst suppression > isoelectric activity occurred with smooth transitions in Fourier analysis of total power (see later).

Fast Fourier transform analysis of EEG signals revealed marked and consistent changes in total power and frequencies at maximum power, which were associated with increasing thiopental concentrations and anesthetic depth. Plasma concentrations of thiopental in the range of 15–20 micro gram/ml produced a 343+/-60% and 85+/-7.3% increase in cortical and hippocampal EEG total power, respectively (P < 0.01, analysis of variance;Figure 4). Rats lost the righting reflex at the peak of this total power increase, at approximately 2 min into the infusion, and after the cortical frequency increase had peaked and was beginning to slow to delta frequencies (Figure 4). The hippocampal EEG, although showing a marked increase in total power, did not exhibit frequency increases comparable to those seen in cortical signals. Three minutes into the thiopental infusion, rats were no longer responsive to whistles or loud claps and had lost EEG and motor responses to moderate tactile stimuli. Tail pinch, corneal, and vascular reflexes remained intact and both cortical and hippocampal signals exhibited further slowing to delta activity and a gradual decline in total power measures. Anesthesia was apparent by 4 min, evidenced by loss of tail pinch and corneal reflexes and by pronounced burst suppression activity in both cortical and hippocampal EEG signals. Arterial blood concentrations of 60–70 micro gram/ml were achieved and FFT total power measures were reduced below control levels in cortex and hippocampus (Figure 4). Heart rate and respiration also were beginning to slow and short periods (2–10 s) of arrhythmic heartbeats were noted. Apnea was observed only rarely (briefly in one rat) and oxygen was administered within 1 min after loss of righting reflex. Vascular reflexes remained intact, evidenced as blood pressure increases in response to tail pinch stimuli. At no time did blood gasses or core temperature change more than 10% compared with prethiopental measures, and oxygen saturation remained > 96% throughout the infusion and postinfusion periods.

Rats were anesthetized for 5–6 min (lack of tail pinch and corneal reflexes) and 4–5-s periods of isoelectric activity were seen during the final 30 s of the thiopental infusion. Burst suppression activity persisted throughout the period of anesthesia, lasting for several minutes (4–6) after the infusion. Total FFT power began to recover within 1 min after the infusion, but remained below control levels until tail pinch and corneal reflexes had recovered. Frequency at maximum power measures showed a more complex relationship with higher frequency spikes evident in cortex, associated with burst suppression activity during anesthesia. Frequency spikes also were also seen in the hippocampus during anesthesia, again reflecting burst suppression activity, but these spikes occurred on a background of lower frequency activity that persisted for > 30 min postinfusion.

Tail pinch and corneal reflexes returned within 5 min after infusion of thiopental and coincided with an increase in total power levels and recovery from burst suppression activity in both hippocampus and cortex. Recovery of the righting reflex required 1–1.5 h and rats remained sedate for an additional 2–3 h, evidenced in decreased exploration and motor behavior. Full recovery of EEG signals to preinfusion levels of total power and frequencies at maximum power required 8–10 h.

Although some similarities were evident, individual EEG frequency bands were differentially altered by thiopental. Low frequency delta activity contributed some of the power increase (approximatley 0.2 of approximately 1.3 micro Volt²total power) to the early EEG activation seen in the first minute of thiopental infusions. delta Frequency total power increased from an average baseline level of 0.09 micro Volt²and peaked at 0.62 micro Volt², for the rat shown in Figure 5(A). More modest increases were seen in other animals with an increase of 209+/-52%(mean+/-SD, P < 0.001, n = 4) evident across experiments (Figure 5(B)). alpha activity increased from 0.04 to 0.38 micro Volt²(809+/-212%), showing the greatest overall change in power for a given frequency band during the peak activation. beta activity contributed to the early activation, increasing from baseline levels of 0.04–0.26 micro Volt²(743+/-215%), and theta activity also contributed to the power increase (from 0.01 to 0.16 micro Volt²; 268+/-53%). All four frequency bands showed a significant (P < 0.01, analysis of variance) increase in power compared with preinfusion baseline activity. Relative to the total power increase seen during thiopental-induced peak activation, each frequency band contributed the following amounts on a percentage basis: delta approximately 10%, theta approximately 15%, alpha approximately 40%, and beta approximately 35%.*.

During behaviorally defined anesthesia (loss of tail pinch and corneal reflexes), cortical EEG power levels were depressed in all frequency bands (Figure 5). xi activity was depressed most on a percentage basis (7.6% of baseline; P < 0.005, analysis of variance), but significant reductions (P < 0.01) also were evident in delta (32%), alpha (12.3%), and beta (18%) activity for total power averaged across the period of anesthesia observed in each animal. Similar decreases were observed for hippocampal EEG signals, except that theta activity showed a greater overall depression (2.1% of baseline), reflecting the larger contribution of theta to baseline total power in the hippocampus. Because burst suppression activity was the dominant waveform recorded in both cortex and hippocampus during anesthesia, these dramatic power reductions in the major EEG frequency bands were expected.

Marked differences were observed in individual frequency bands during recovery after thiopental infusion (Figure 5(A)). Delta activity was first to recover and showed a trend of increasing power before maximal depression was observed in theta, alpha, and beta activity. Delta activity was evident during periods preceding and after burst suppression and was strongly depressed for only a brief time (approximately 1 min) when periods of isoelectric EEG activity were seen. Alpha depression correlated best with the time course of anesthesia, because recovery corresponded to a return of corneal and tail pinch reflexes. Xi and beta activity remained depressed for at least 1 h after the thiopental infusion, and recovered with a time course comparable to recovery of the righting reflex.

Steady-state plasma concentration-EEG effect relationships for thiopental are shown in Figure 5(B). Cortical and hippocampal EEG effects were normalized with respect to baseline values in each frequency band. No significant differences in k^eovalues were found between the cortex and hippocampus. The mean k^eovalue for total power in cortex was 0.99+/-0.33 min sup -1 (mean+/-SD, n = 4) whereas the mean k^eovalue for total power in hippocampus was 1.09+/- 0.51 min sup -1. Small differences were observed between k^eovalues for specific frequency bands, however, the biphasic shape of concentration-EEG effect relationships in cortex and hippocampus was similar. In fact, for the delta frequency band the concentration-effect relationships were superimposable (Figure 5(B)). The main difference between cortical and hippocampal concentration-effect relationships was the magnitude of initial EEG activation produced by thiopental, cortical effects were 2–3 times greater than hippocampal effects (Figure 5(B)). Alpha, beta, and delta activity contributed most of the absolute power increase to the early EEG activation observed at low thiopental concentrations. Loss of righting reflex occurred at the peak of alpha frequency activation. Both alpha and theta activation occurred at thiopental concentrations that were slightly less than those required for peak delta frequency activation.

The thiopental concentration--effect relationship could be divided into four stages. The first stage, activation, occurred at 10–20 micro gram/ml thiopental, which corresponds to free (unbound) aqueous levels of 5–15 micro Meter (Figure 5(B)). Delta Activity was evident at concentrations of 20–45 micro gram/ml (approximately 15–35 micro Meter). Burst suppression activity was associated with loss of tail pinch and corneal reflexes and occurred at concentrations of 45–70 micro gram/ml (approximately 35–65 micro Meter), and isoelectric activity was observed above 70 micro gram/ml (approximately 80 micro Meter).

Burst suppression activity was clearly associated with the period of surgical anesthesia measured behaviorally, but also showed a concentration-dependent continuum ranging from > 10 bursts/s in the early stages to < 1 burst/s immediately preceding a 5-s period of isoelectric EEG activity. Cortical burst events were synchronized with events in the hippocampal signal (Figure 6), although differences in timing, morphology, and amplitude indicated that polysynaptic interactions were involved. For example, cortical bursts had a faster rise time (20–80 ms) and sometimes (36+/-7%, n = 180) started with a negative voltage deflection, compared with hippocampal bursts (120–150 ms), which usually (94+/-4%) started with a positive deflection. In addition, cortical events were multiphasic (negative--positive--negative--positive), whereas hippocampal bursts were biphasic (positive--negative). When cortical bursts were aligned with respect to the peak positivity of the hippocampal signal (Figure 6) a clear relationship was not evident, because some events were aligned with the early negativity (42%), late negative component (34%), early positivity (11%), late positivity (9%), or not at all (4%).

Event-time histograms comparing peak positivity of cortical and hippocampal bursts indicated that most hippocampal events (> 95%) occurred approximately 60 ms before a cortical burst. Histograms comparing cortical and hippocampal burst synchronization in the early stages of burst suppression activity showed a high degree of synchrony (Figure 6), evidenced by a sharp Gaussian distribution with a chi-square value of 72.2 and a peak of > 16/200 (8%) events centered at -58 ms. This synchrony was markedly reduced in the late stages of burst suppression (level 2), although the majority of hippocampal events (> 70%) still occurred within +/-200 ms of a cortical event. The level 2 data showed a Gaussian distribution with a chi-square value of 123.8, but with < 5/200 (2.5%) events contributing to a peak centered at -9 ms. Data for level 1 burst suppression were acquired during the first 15 s of burst activity and level 2 data from the 30-s period preceding a 5-s interval of isoelectric activity. A total of 200 burst events were analyzed for both levels 1 and 2 burst suppression.

It has long been recognized that certain EEG rhythms show a good correlation to behavioral and/or psychological states. This is particularly true for the relationship of cortical and hippocampal theta rhythms associated with arousal in rodents, higher mammals, and primates, [19]including humans. [20]Arousal being defined either physiologically, as an EEG response to sensory stimulation, [21]or more recently as sensory motor processing defined behaviorally by simultaneous monitoring of EEG and spontaneous or learned movements. [10,12,13,22,23]Results from the current study confirm previous reports showing that awake immobility was associated with high frequency low amplitude activity in the cortical EEG and with high amplitude irregular activity (including sharp waves) in hippocampal signals. Awake voluntary movements were consistently associated with a prominent increase in theta frequency activity, particularly in the hippocampus (Figure 2).

These results add to a growing body of literature relating EEG alterations to sleep/wake patterns, [24]psychological-behavioral states, [25]pathologic conditions, [26]cognitive processing, [27]and learning/memory. [28]Cortical EEG has been shown to provide a measure of brain electrical activity with a high degree of consistency across individuals and between species, particularly when using higher resolution techniques to monitor spatial-temporal function over a well-defined region (< 0.5 cm²) of cortex. [27]There also has been an increasing understanding of the neuronal and synaptic basis of EEG signals, especially with the advent of microelectrode recording techniques in vivo [10,12,29]and brain slice preparations capable of generating EEG activity. [12,28,30,31]Several ascending synaptic systems contribute to synchronization of cortical neurons underlying alpha, beta, delta, and theta rhythms, and these systems are highly divergent with individual brain stem neurons forming synapses with several million cortical cells. [29,32]These synchronizing systems have important functional aspects; for example, theta rhythms are involved in long-term potentiation of synaptic transmission and memory formation in hippocampus, [28]40-Hz rhythms are associated with input processing in primary sensory cortex, [33]and there is an ever expanding literature on mechanisms and function associated with the best known brain waves: delta (slow wave sleep) and alpha rhythms. Even some longstanding paradoxes regarding the diversity of EEG signals recorded from comatose patients have been elucidated, as it is now clear that coma can come about via several mechanisms, some of which spare higher cortical function to varying degrees. [34]It is likely that some of the diverse EEG patterns produced by different general anesthetics also can be explained by a diversity of cellular mechanisms contributing to different anesthetic states. [7,8,35].

Thiopental produced a concentration-dependent continuum of effects, including activation (increased amplitude and frequency), EEG slowing (increased delta rhythms), burst suppression, and isoelectric activity. This continuum of effects was mapped to anesthetic depth with burst suppression EEG activity associated with surgical levels of anesthesia, similar to previous findings for barbiturates, propofol and volatile agents. [4,5,8]The same continuum of effects has been observed across human subjects [36]and between species. [5,14,37]Burst suppression activity occurred at concentrations from 50 to 80 micro gram/ml (approximately 230 micro Meter), which corresponds to a free aqueous effect site concentration of approximately 50 micro Meter based on pharmacokinetic models. [11,18]These concentrations are comparable to thiopental levels needed to produce burst suppression activity in cats (approximately 25 micro gram/ml, reduced by the presence of 0.5 vol% halothane)[5]and humans (approximately 30 micro gram/ml), [38]and agree with previous findings in rats. [14,37].

Loss of righting reflex occurred at the peak of EEG activation at steady-state thiopental concentrations of 20 micro gram/ml. Similar results are found in humans, peak EEG activation occurs at steady-state plasma concentrations of approximately 20 micro gram/ml. [38]In humans, peak activation was associated with a loss of response to verbal stimuli. [38]Studies in humans have shown that achieving burst suppression EEG activity is required to block memory storage and recall after anesthesia. For example, Rampil and coworkers demonstrated that subjects taken only to a level of delta EEG slowing showed a statistically significant ability to recall spoken words during anesthesia with isoflurane. [39]Similarly, Ghoneim and Block [40]recommended studies to establish minimum alveolar concentration values for various agents using EEG measures linked to psychophysical testing to monitor learning and consciousness during general anesthesia. Burst suppression activity in the cortical EEG could be indicative of a loss of consciousness and recall when receiving barbiturate or volatile agents alone, [40]but little is known about EEG effects required for amnesia when mixtures of anesthetics are used. [7].

Anesthetic-induced burst suppression activity may come about via a depression of cortical activity with brief periods of excitation superimposed on increasing levels of inhibition. It is thought that the excitation arises from ascending synaptic inputs [5,29,41]and this could explain the high degree of synchrony seen for burst events in cortex and hippocampus in the current study. As thiopental concentrations increased, the synchrony was reduced (Figure 6), suggesting a progressive depression of ascending inputs until isoelectric activity was observed. This is consistent with the view that general anesthetics depress forebrain structures first [1,8,42]sparing midbrain and spinal reflexes when isoelectric activity is seen in cortex [7]or even after forebrain structures have been removed. [43]Further studies are required to determine where the excitatory drive originates during burst suppression activity, but it is evident that reticular formation neurons do not generate this activity because they were also depressed by thiopental at concentrations that produced burst suppression in cortex. [5,41]A possible source of ascending excitation is the brain stem lemniscal pathway and cuneate nucleus, because barbiturates (and other anesthetics) have been shown to increase synaptic transmission in this pathway. [44].

It also is possible that intrinsic excitation within the cortex could trigger burst discharges. Early EEG studies demonstrated that barbiturate-induced burst suppression and isoelectric activity were evident in isolated (undercut) regions of neocortex in humans [45]and canines. [46]This, together with depressant effects produced by barbiturates at the brain stem and thalamic levels, lead Rosner and Clark [47]to suggest that burst suppression activity could originate in neocortex. The recent demonstration of thiopental-induced burst suppression activity in isolated cortical brain slices lends strong support to this argument. [31]The synchronization of bursts between hippocampus and cortex could result from reciprocal excitatory connections that are known to exist between these brain structures. It is likely that both intrinsic and ascending excitatory drive contributes to burst suppression activity in the cortex and hippocampus.

The altered electrical activity observed in hippocampus and cortex could account for some of the higher nervous system effects of thiopental. For example, the hippocampus is required for memory formation [48,49]and theta rhythms have been linked to memory formation in recent studies showing that long-term potentiation of hippocampal synapses is enhanced when afferent activity is synchronized to CA 1 neuron theta frequency discharge. [28]The block of recall produced by thiopental could come about via disruption of memory formation in the hippocampus, secondary to the marked depression of theta activity observed (Figure 5). Depressed rhythmic activity and burst suppression EEG patterns reflect alterations in higher nervous system function, which may account for the loss of consciousness, blocked perception, and amnesia that are essential components of anesthesia.

The results support previous findings that cortical EEG provides a good measure of the concentration-dependent continuum of sedation > hypnosis > anesthesia seen behaviorally. [3,7,8]Fast Fourier transform analysis of EEG signals revealed a profound depression of alpha, beta, delta and theta rhythms, replaced by burst suppression and isoelectric activity during anesthesia. The time course of recovery of rhythmic EEG activity was correlated with recovery of corneal, motor, and righting reflexes. It is likely that a progressive depression of EEG rhythms comes about via a cumulative recruitment of effects on excitatory and inhibitory synaptic transmission. [31]The continuum of behavioral effects was reflected in a concentration-dependent progression of depressed synaptic responses, evident in the slowing and uncoupling of synchronous hippocampal and cortical discharge during burst suppression EEG activity.