Isoflurane Action in the Spinal Cord Blunts Electroencephalographic and Thalamic–Reticular Formation Responses to Noxious Stimulation in Goats

The local institutional animal care committee approved this study. Seven goats (55.0 ± 12.4 kg, mean ± SD) were anesthetized by mask with isoflurane and their tracheas were intubated. End-tidal isoflurane was determined using a calibrated agent analyzer. A tube was passed through the esophagus to drain rumen contents. A peripheral intravenous catheter was placed for administration of lactated Ringer’s solution. Bilateral neck dissections were performed to isolate the jugular veins and carotid arteries on both sides. 1,5A craniotomy was performed to access the brain. Intravenous heparin (3 or 4 mg/kg initially, 1 or 2 mg/kg every 2 or 3 h) was administered. Y cannulae were placed into each jugular vein, and a cannula directed toward the head was placed into a carotid artery. A small catheter placed in the same carotid artery was directed toward the heart, thereby permitting measurement of systemic blood pressure and withdrawal of blood for analysis of glucose, hematocrit, and blood gases.

Anesthesia was maintained with isoflurane and, after administration of pancuronium (0.1–0.15 mg/kg and 0.1 mg/kg every 1 or 2 h), the head was placed in a stereotaxic frame and secured with ear bars and a mouth piece. In each of six goats, six stainless steel electrodes (28-gauge, covered with polyurethane except for the distal 2 mm) were passed into the midbrain region, with three on either side of midline. 4The targets were the caudal midbrain reticular formation (MRF; 5 mm lateral to midline, 5 mm superior to the horizontal interaural line, and 3 mm rostral to the vertical interaural line), the rostral MRF–caudal thalamus interface (5 mm lateral, 5 mm superior, 10 mm rostral), and the rostral thalamus (5 mm lateral, 5 mm superior, 22 mm rostral). 6No specific thalamic–MRF nuclei were targeted. For EEG recordings, four platinum needle electrodes (E-2; Grass Instruments, West Warwick, RI) were placed into the periosteum overlying the skull in the occipital and frontal areas bilaterally, with a ground electrode placed in the scalp. Impedances of the bifrontal–hemispheric scalp electrodes were less than 2 kOhm. The depth electrodes had slightly higher impedances, and were generally less than 5 kOhm. Ten channels were recorded: ipsilateral and contralateral cerebral cortex, bifrontal cortex, ipsilateral and contralateral MRF, ipsilateral and contralateral thalamus, bilateral caudal–rostral MRF and thalamus. The electrodes were connected to a Grass Model 8-10E EEG machine (Grass Instruments, Braintree, MA), and the signals were amplified and entered into a personal computer where they were digitized (12-bit resolution at 250 Hz) using a commercial program (PolyViewPro; Grass Instruments, Braintree, MA). The amplifiers filtered the signals (low: 1 Hz, high: 70 Hz; the data were further filtered off-line at 35 Hz; see Statistical Analysis section).

In each of five animals, a tungsten microelectrode (Frederick Haer, Inc., Brunswick, ME) was passed into the caudal MRF to measure single-unit activity. The coordinates were 5–7 mm lateral to midline and 3–4 mm rostral and 5 mm superior to the interaural line. 6The microelectrode had a tip diameter of approximately 25 μ and an impedance of approximately 10 MΩ. A hydraulic microdrive (David Kopf Instruments, Tujunga, CA) was used to adjust electrode depth. The signal was amplified, filtered (0.3–3 kHz), and entered into a personal computer where the signal was digitized and stored for later analysis and construction of peristimulus-time histograms. 7The signal-to-noise ratio usually exceeded 10:1. Single units were discriminated based on amplitude and waveform using a custom computer program (SPIKE, University of Erlangen, Erlangen, Germany). 7Units were sought that increased activity in response to the noxious stimulus applied to the forelimb.

Blood (500 ml) was drained from the animal and used to prime a bubble oxygenator (Bentley B-10Plus; Baxter, Irvine, CA). Gas flow to the oxygenator was 95% O²–5% CO². An isoflurane vaporizer was placed in line with the gas flow. Isoflurane concentration in the arterial blood of the bypass unit was estimated from the isoflurane concentration in the exhaust of the oxygenator, which was measured using a calibrated agent analyzer. 1,5Cranial bypass was initiated by draining cranial venous blood into the oxygenator and infusing it into the cranial circulation via  a roller pump (cranial flows 200–500 ml/min). The remaining carotid artery that transmitted systemic blood to the cranial circulation was temporarily ligated to establish complete bypass, thereby separating the cranial and systemic circulations at the level of the upper cervical spinal cord and caudal medulla. 5The vertebral arteries do not contribute to the goat cerebral circulation. 5Mean cranial blood pressure was 61 ± 14 mmHg and mean systemic blood pressure was 122 ± 22 mmHg. Glucose (10–20 mg/min) was infused into the oxygenator. Torso temperature was measured from the rectum and maintained at 38.6 ± 0.8°C (mean ± SD) using a heating lamp. During bypass, nasopharyngeal temperature was maintained at 38.2 ± 0.7°C using the heat exchanger of the bypass unit.

The cranial isoflurane concentration was kept at 1.1 ± 0.3% (1.2 ± 0.2% in the five goats for which recording of both the EEG and the MRF single-unit activity was recorded). Although the cranial isoflurane was slightly different among the goats, in each individual goat the cranial isoflurane was always maintained at the same concentration. The torso (end-tidal) isoflurane concentration was alternated between 1.2 ± 0.3% and 0.3 ± 0.1%, with at least 15 min elapsing at a new concentration, to permit equilibration before data collection. We did not seek lesser torso isoflurane concentrations because of the additional time necessary to remove residual isoflurane. The order of the torso isoflurane concentrations was alternated experiment to experiment. At each of the two torso isoflurane concentrations, the MRF single-unit and EEG and thalamic–reticular formation responses were recorded 1 min before and during a 1-min application of a noxious mechanical clamp (10-in hemostat) to the dewclaw of a forelimb. In several animals, the MRF responses were recorded 2 or 3 times using an interstimulus interval of at least 5 min.

At the conclusion of the experiment, electrolytic lesions were made (6–8 V DC for 30–40 s). The animals were killed with potassium chloride and the brains were removed and fixed in 10% formalin and 1% potassium ferricyanide. The brains were frozen and sectioned with a microtome (50 μm) and the recording sites were determined using a light microscope.

The data are presented as mean ± SD. The raw EEG and depth macroelectrode signals were visually inspected and artifacts were deleted before analysis (< 1% of data). The electrical activity data for each channel were analyzed using the PolyViewPro software (Grass Instruments, Braintree, MA). The raw data were filtered at 1–35Hz. Total power and power in the δ (1–3 Hz), θ (3–8 Hz), α (8–13 Hz), and β (13–35 Hz) ranges were determined. The EEG data were analyzed for 1 min before and 1 min during the application of the noxious stimulus, using 4-s consecutive epochs of time. The power in these epochs was averaged over each 1-min period. Because the data appeared to be normally distributed, the means of the prestimulus and poststimulus data (for cranial and torso isoflurane concentration combinations) were compared using repeated-measures analysis of variance to detect overall differences, and a Student-Newman-Keuls test was used to determine differences between the pre- and poststimulus data at each cranial or torso anesthetic combination. The percent change in power after the noxious stimulus application for each frequency range (all channels combined) was analyzed using analysis of variance. The single-unit activity was analyzed by counting the total number of action potentials in 1-min intervals before and after application of the noxious stimulus. The single-unit activity was not normally distributed and was first subjected to log transformation, 8followed by repeated-measures analysis of variance and the Student-Newman-Keuls test; a P < 0.05 was considered significant.

When the torso isoflurane concentration was 1.2 ± 0.3%, the noxious stimulus had no effect on the EEG or subcortical electrical activity, which exhibited a predominantly low-frequency, high-amplitude pattern. When the torso isoflurane was decreased to 0.3%, the noxious stimulus caused the EEG and subcortical electrical activity pattern to shift from a slow-frequency, large-amplitude pattern to a fast-frequency, low-amplitude pattern. An individual example is shown in figure 1. Virtually all sites showed this pattern, with decreased total, δ, θ, or α power; β power was unchanged (figs. 2 and 3). The data for all channels combined showed that, when torso isoflurane was 0.3 ± 0.1%, there was decreased total, δ, θ, and α power after the noxious stimulus was applied (range, 26 ± 16% to −51 ± 22%;P < 0.05).

The MRF single-unit activity paralleled the EEG data. The MRF cells characteristically had large receptive fields that included two or more extremities and sometimes the face. Figure 4provides an averaged peristimulus-time histogram of all MRF units (N = 5) during the two anesthetic conditions and shows the significant increase in evoked activity when the torso isoflurane was low. At the 1.2%–1.2% combination the spontaneous and noxious-evoked activity were not significantly different (370 ± 381 to 489 ± 437 impulses/min). When the torso isoflurane was decreased to 0.3%, the spontaneous and noxious-evoked activity were significantly different from each other (712 ± 892 to 1,286 ± 1,317 impulses/min;P < 0.05).

Histologic analysis confirmed that the depth electrodes were located in the region of the caudal MRF, the MRF–thalamic interface and the rostral thalamus (fig. 5); the microelectrodes were located in the caudal MRF. Blood gases, glucose, and hematocrit were normal, with the exception of a mild acidosis in the oxygenator blood (table 1).

The results of the current study are consistent with the hypothesis that isoflurane exerts an indirect effect on the brain via  an action within the spinal cord. The desynchronization pattern evoked by the noxious stimulus when the torso isoflurane concentration was low is consistent with a transition toward arousal. This pattern was observed at cortical sites and in deeper structures (thalamus and reticular formation). The MRF single-unit activity paralleled the EEG data. The MRF and thalamus are considered to be critical to consciousness, and activity in these areas is assumed to correlate with consciousness. 9In general, anesthetics depress MRF cells. 10,11Angel 12,13documented the importance of the thalamus as a site at which anesthetics blunt the centripetal transmission of ascending neural information. The current study underscores the importance of the spinal cord as well. Anesthetic action in what parts of the spinal cord might account for our current findings?

The dorsal horn is the first relay in the transmission of peripheral stimuli to the brain. Numerous studies over several decades documented the depressant effect of various anesthetics on dorsal horn cellular responses to noxious and non-noxious stimuli. 3,14–16To our knowledge, none of these studies determined whether the recorded dorsal horn cells had any ascending projections, in particular to the MRF, thalamus, and midbrain. Spinal cells can project to supraspinal structures via  the spinomesencephalic, spinoreticular, spinothalamic, postsynaptic dorsal columns, and spinocervicalthalamic tracts. 17Thus, it is possible that isoflurane could act at any or all of these different types of ascending spinal cells to depress sensory transmission to the brain. Additional work is necessary to determine whether these different classes of spinal cells have differing sensitivities to anesthetics. It is possible that some depressant effects might be the result of anesthetic action on conduction along the axon, as opposed to the synapse. In clinical concentrations, however, anesthetics appear to have minimal depressant effects on nerve conduction 18; therefore, this is unlikely to be the cause of our findings. Last, we cannot rule out a peripheral anesthetic effect as a factor in the data. However, peripheral anesthetic actions of isoflurane and halothane appear to be minimal or paradoxically may sensitize peripheral nociceptors. 19,20We previously determined that a peripheral anesthetic action of isoflurane, if any, does not affect minimum alveolar concentration (MAC). 21Therefore, we do not believe that peripheral actions can explain our findings.

Clinical cases that involve spinal cord injury also are consistent with our findings. Patients with cervical spinal cord injuries have decreased global cerebral metabolism compared with volunteers without spinal cord disease. 22These data suggest that the diminished peripheral sensory input results in decreased cerebral activation. Also, patients with high spinal cord injury have altered sleep and EEG patterns. 23 

Our technique of recording from subcortical structures (thalamus and MRF) is limited because the macroelectrodes located in these sites probably also measured electrical activity from structures farther away, such as the cerebral cortex. 4Nonetheless, we believe that the activity recorded at each site represents activity primarily from local, and not distant, structures. 4 

We did not thoroughly investigate the spinal cord isoflurane concentration (e.g. , dose–response) that would have prevented the desynchronization response. However, in our previous study, we noted that isoflurane blunted the desynchronization response at concentrations between 0.9 and 1.1 MAC. 4We expected that isoflurane would have blunted the response at lower concentrations inasmuch as the brain is more sensitive to anesthetics, compared with the spinal cord, at least when considering two anesthetic end points (amnesia and unconsciousness). 24There is at least a theoretical, if not physiologic, link between these end points and the EEG activity. The EEG activity can be used to investigate memory and consciousness. 25Furthermore, analysis of processed EEG data (such as bispectral analysis) permits close monitoring of anesthetic effects vis-à-vis memory and consciousness. 26,27Iselin-Chaves et al.  found that a trapezius squeeze, which is noxious but not supramaximal, increased the bispectral index number, consistent with cerebral activation. 26Studies that evaluated anesthetic requirements for amnesia and unconsciousness usually have been performed in volunteers who were not subjected to noxious stimulation. 24We believe that, in the presence of noxious stimulation, additional anesthesia would be necessary to achieve amnesia and unconsciousness. However, clinical experience alone would suggest that amnesia and unconsciousness occur at anesthetic concentrations less than 1 MAC. Several studies that evaluated the effect of central neural blockade on anesthetic and hypnotic requirements are consistent with the current results. In humans, subarachnoid local anesthetics appear to lower hypnotic requirements. 28In rats, intrathecal bupivacaine decreases the hypnotic and anesthetic requirements for thiopental, suggesting that the brain is indirectly affected by anesthetic blockade in the spinal cord. 29Cole et al.  found that intrathecal tetracaine in rats decreased the cerebral metabolic rate during somatosensory stimulation, suggesting that ascending afferent information activated the brain, and that local anesthetic block diminished this effect. 30 

The spinal cord has assumed an important role as a site of anesthetic action. Our work 1and that of Rampil 2strongly suggest that immobility is produced by anesthetics via  action in the spinal cord. As pointed out by Kendig, 31however, anesthetic action in the spinal cord could indirectly affect consciousness and memory by blunting the orderly transmission of stimuli to the brain, and our current results are certainly consistent with this possibility.

The authors thank Marla Piercy and Mirela Iodi Carstens for assistance.