Temperature Changes of greater or equal to 1 degree Celsius Alter Functional Neurologic Outcome and Histopathology in a Canine Model of Complete Cerebral Ischemia

After Institutional Animal Care and Use Committee approval, the study was conducted in 24 purpose-bred dogs, weighing 17.5-23.0 kg. All dogs were fasted, but had free access to water, for a minimum of 8 h before initiating the study. Anesthesia was induced with halothane 1.5-3.5% inspired in oxygen in an induction box. Once the dogs were anesthetized, the tracheas were intubated and the lungs were mechanically ventilated with a Harvard pump (South Natick, MA). A tidal volume of 15-20 ml/kg was used, and the respiratory rate was adjusted to maintain arterial carbon dioxide tension (Pa^CO2) near 35 mmHg. Anesthesia was maintained with halothane 0.87-1.13% end-expired (1.0-1.3 minimum alveolar concentration in dogs) in nitrogen and oxygen during the preparatory period, and the inspired oxygen fraction was adjusted to maintain arterial oxygen tension (Pa^O2) near 150 mmHg. A cannula was inserted into the forelimb vein for fluid and drug administration. Neuromuscular block was induced and maintained with intravenous pancuronium.

Temperatures within the right atrium were measured with a catheter-type thermistor inserted via the external jugular vein, and cranial temperatures (beneath the temporalis muscles; thermistor tip in the calvarium's periosteum) were measured bilaterally with percutaneous needle thermistors (model 73A, Yellow Springs Instruments, Yellow Springs, OH). Temperatures were maintained at target values (see below) with a forced-air heating and cooling blanket (Polar Bair prototype, Augustine Medical, Eden Prairie, MN) placed over the dorsal and lateral portions of the dog. This device was supplemented with a ventral surface heating pad and overhead heating lamps to produce small changes in and optimally stabilize regional body temperatures.

Cannulae were inserted into both femoral arteries using PE-190 and PE-320 polyethylene catheters (Becton Dickinson, Parsippany, NJ). The former was used for continuous blood pressure monitoring; the latter was used for blood withdrawal and infusion. Arterial blood gases were measured with electrodes at 37 degrees Celsius (model 1304, Instrumentation Laboratories, Lexington, MA). A two-channel, bifrontal and biparietal, electroencephalogram (EEG) was monitored with periosteal needle electrodes and a polygraph (model 8-10, Grass, Quincy, MA). Inspired and end-expired oxygen, carbon dioxide, and halothane were measured by Raman scattering analysis (RASCAL II, Albion Instruments, Salt Lake City, UT).

The animals were prepared for the production of complete cerebral ischemia using a previously described compression technique. [13]In brief, using a Hustead introduction needle, a polyethylene catheter (PE-190) was inserted into the lumbar subarachnoid space. Two Hustead 18-G needles were inserted into the cisterna magna. The lumbar catheter and one cisternal needle were used to infuse 0.9% saline solution, warmed to the target temperature (model 4493 water bath, Haake, Berlin, Germany) to increase intracranial pressure. The second cisternal needle was used to continuously monitor subarachnoid pressure.

Halothane was decreased to 0.87% end-expired, and arterial blood gases, acid-base status, blood glucose concentrations, and temperatures were maintained within protocol limits for a minimum of 20 min before obtaining control physiologic measurements. Because of previous studies reporting significant differences in neurologic outcome resulting from minor differences in the preischemic blood glucose concentrations, [14,15]blood glucose concentrations were strictly maintained between 110-130 mg/dl before control and study measurements (model 23A, Yellow Springs Instruments). Thus, if the blood glucose concentration was < 110 mg/dl, an intravenous infusion of 10% glucose was titrated to a steady-state blood concentration that was within the protocol criteria.

The dogs were randomly assigned to one of three groups: (1) a reference group maintained at 37.0 plus/minus 0.3 degree Celsius (target temperature plus/minus range); (2) a 38.0 plus/minus 0.3 degree Celsius group; or (3) a 39 plus/minus 0.3 degree Celsius group (n = 7 per group). Both the right atrial and cranial temperatures were maintained at the target temperature for approximately 20 min before, during, and 1 h after the ischemic insult.

Complete cerebral ischemia was produced by modifying a previously described "compression" method. [13]In brief, blood was withdrawn from the larger femoral artery cannula into a warmed heparinized reservoir to rapidly reduce the systolic blood pressure to less or equal to 60 mmHg. Upon achieving this target blood pressure, warmed (to target temperature) saline solution was infused into the lumbar and one of the cisterna magna subarachnoid conduits until the subarachnoid pressure was at least 20 mmHg greater than the systolic blood pressure. This procedure resulted in complete ischemia, as confirmed by an isoelectric EEG, within seconds. Once the EEG became isoelectric, the inspired halothane and nitrogen were discontinued. During this period, right atrial and subtemporalis temperatures were maintained at the target value, and systolic blood pressure was maintained at less or equal to 60 mmHg. Any tendency for blood pressure increases during the ischemic period were treated with additional blood withdrawal. After 11.5 min of ischemia, the mean arterial pressure (MAP) was increased to 60-80 mmHg by reinfusing the withdrawn blood and, if needed, infusing phenylephrine (40 micro gram/ml intravenously). After exactly 12.5 min of ischemia (determined from the onset of an isoelectric EEG), the subarachnoid conduits were opened to air, resulting in restoration of cerebral perfusion pressure (defined as MAP - intracranial pressure) to greater or equal to 60 mmHg within seconds.

During the first 5 min of reperfusion, the MAP was maintained at a target value of 100 mmHg by blood withdrawal or reinfusion. Thereafter, the MAP was maintained near 150 mmHg. In all dogs, the previously withdrawn blood was reinfused before discharge from the operating room.

When the dogs were judged to be hemodynamically stable, but before the return of EEG activity, they were sedated with 50% nitrous oxide in oxygen in an attempt to prevent cerebral hypermetabolism from the stress of immobilization.* The dogs were maintained at their target temperatures for 1 h postischemia, at which time the temperatures of the 38 degrees Celsius and 39 degrees Celsius study groups were returned to 37 degrees Celsius. This was achieved by 2.5 h post-ischemia Figure 1. The reference group was maintained at 37 degrees Celsius throughout the study period. At 3 h postischemia, neuromuscular block was reversed with intravenous neostigmine 0.07 mg/kg and intravenous glycopyrrolate 0.014 mg/kg, and the nitrous oxide was discontinued. When judged to have adequate airway reflexes and the arterial blood gas demonstrated Pa^Osub 2 > 60 mmHg and Pa^CO2< 45 mmHg, while spontaneously breathing room air, the tracheas were extubated. All animals received intramuscular procaine-benzathine penicillin 450,000 U (Flo-cillin, Fort Dodge Laboratories, Fort Dodge, 1A) and intramuscular gentamycin 1.7 mg/kg before extubation of the trachea.

For the ensuing 24 h, the dogs were cared for in padded recovery cages in which the floor temperature of the cages was controlled to 37 degrees Celsius by servomechanism (thermal pad and servomechanism controller, Shor-Line, Kansas City, MO), and the ambient temperature of the room was maintained at 28 degrees Celsius. Other than these measures, no attempts were made to control body temperature near any target value. At 24 h postischemia, dogs were moved to standard metabolic cages. They were inspected every 1-2 h for the first 6-12 h post-ischemia, then a minimum of every 8 h thereafter for 72 h. Dogs that were unable to care for themselves or exhibited clinical evidence of dehydration (e.g., dry oral mucosa, poor skin turgor, weak pulse by palpation, or resting tachycardia) were given 5% glucose in lactated Ringer's solution, 500 ml intravenously or subcutaneously, in divided doses, every 12 h.

Neurologic assessment was performed by a blinded observer (R.E.H.) at 24, 48, and 72 h postischemia. The detailed 100-point neurologic deficit scoring scale of D'Alecy et al. [15]was used. In brief, a total neurologic deficit score was derived for each dog based on the sum scores of five categories. These categories included (1) level of consciousness (deficit score range: 0-18 points); (2) motor function (0-28 points); (3) respiratory effort (0-18 points); (4) function of cranial nerves (0-16 points); and (5) function of spinal nerves (0-20 points). The total neurologic deficit score of D'Alecy et al. was converted to a percent neurologic function score according to the formula Equation 1. According to this convention, a normal dog received a percent neurologic function score of 100 and a brain dead animal received a score of 0.

After the final observation at 72 h postischemia, the dogs were anesthetized with intramuscular ketamine and intravenous sodium thiopental. After tracheal intubation, the lungs were mechanically ventilated with room air. A left thoracotomy was performed, and, in rapid succession, the heart was electrically fibrillated, the descending thoracic aorta was cross-clamped, the proximal descending thoracic aorta was cannulated with an 8-G needle, and the right atrium was opened. Immediately thereafter, 0.9% saline at a pressure of 100 cmH²O was infused into the aortic cannula until the atrial effluent was clear. This was followed by infusing a 4% paraformaldehyde solution at a pressure of 100 cmH²O to produce in situ brain preservation. The brains were then removed and stored in buffered paraformaldehyde. All brains were fixed for at least 4 weeks (2 weeks in paraformaldehyde; the remainder in 10% formalin) before microscopic examinations.

Histologic evaluation was performed by a neuropathologist (B.W.S.) blinded to treatment groups. Coronal whole-mount, paraffin-embedded microsections were cut to 6-micro meter thickness and stained with hematoxylin and eosin. Histopathologic findings were graded according to a previously described scale. [16]The severity of injury was assigned points, bilaterally, according to the following scale: normal, 0; minimal, 1; moderate, 2; severe, 3; and maximal, 4. The points were then multiplied by a weighting factor (infarction, 4X; ischemic nerve cell change, 2X; edema, 1X) to obtain a score for each of 19 brain regions. The bilateral scores were averaged to produce a single score for each brain region.

The analysis for one of these regions, the hippocampus, was subdivided further to provide an assessment of the CA1, CA2, and CA3,4 segments. The hippocampus was chosen for segmental analysis because of previous reports, in rats exposed to global cerebral ischemia, of aggravated regional (CA1) histologic damage during mild hyperthermia (39 degrees Celsius) [4,11]and CA1 sparing during mild hypothermia. [5,6].

The scores of individual brain regions, hippocampal segments, and the total score for all regions (inclusive of the hippocampus but exclusive of the hippocampal subdivisions) were analyzed.

Animals that did not meet all preestablished protocol criteria were excluded from data analysis. Exclusion was based on strict criteria: (1) preischemic blood glucose < 110 or > 130 mg/dl; (2) failure to achieve complete cerebral ischemia for 12.5 min; (3) failure to attain a cerebral perfusion pressure greater or equal to 60 mmHg within 1 min after the restoration of cerebral perfusion; (4) failure to maintain Pa^O2> 60 mmHg and Pa^CO2< 45 mmHg while spontaneously breathing room air; (5) the appearance of postischemic MAP < 60 mmHg or > 150 mmHg for more than 60 min or a MAP < 50 mmHg at any time; or (6) death from a cause other than ischemic neurologic disease (e.g., aspiration of gastric contents, pulmonary edema, or cerebral hemorrhage).

Previous studies have shown that postischemic histologic changes in the brain require 24 h or more to mature. [17,18]Thus, any dog that died within 24 h of the ischemic insult was not included in the histologic data analysis, but was included in the analysis of neurologic function, provided the study criteria had been fulfilled. If the dog died after 24 h but before the 72 h observation and data acquisition period, the brain was immediately excised and preserved, and the histologic data from the dog were included in the final statistical analysis. In all cases, the cause of death was determined by a veterinary pathologist (A.G.A.) after reviewing the clinical history (excluding information of the treatment group and ischemic period) and performing an extensive necropsy. Also, these findings were discussed with the blinded observer (R.E.H.) to determine whether protocol criteria had been violated. All decisions to exclude a dog from the final data analysis were made by the blinded observer.

With all data, the 38 degrees Celsius and 39 degrees Celsius groups were statistically compared with the reference group; however, the 38 degrees Celsius and 39 degrees Celsius groups were not compared with each other. Physiologic variables were compared by using unpaired t tests. Functional outcome scores and histopathology scores were compared by using Mann-Whitney rank sum tests and the Spearman rank correlation coefficient (r^s). P < 0.05 was considered significant. Parametric data are presented as mean plus/minus SD, unless otherwise indicated.

The three treatment groups were well matched for preischemic and postischemic physiologic variables, except for temperature (Table 1). Infusion of phenylephrine (approximately 20 micro gram) was used to increase blood pressure at the completion of ischemia in only one dog (39 degrees Celsius group).

Two dogs from the 37 degrees Celsius group and one dog from the 39 degrees Celsius group were excluded from statistical analysis because of failure to meet preestablished protocol criteria. Of these dogs, the 39 degrees Celsius dog and one 37 degrees Celsius dog died within 15 and 20 h, respectively, after cerebral ischemia because of massive aspiration of gastric contents. The latter dog's course was influenced, in part, by a 2-ml hematoma at the medullopontine junction that produced a palsy (observed upon emergence from anesthesia) of the cranial nerves responsible for airway protection. The other 37 degrees Celsius dog died within 49 h of the ischemic event as a result of neurogenic cardiomyopathy and pulmonary edema. [19,20]Of note, the two 37 degrees Celsius dogs had a good level of neurologic function (alertness, normal respiratory pattern, standing or attempts to stand, ability to track movement with their eyes, and ability to feed themselves) before their death. Conversely, the 39 degrees Celsius dog had severely impaired neurologic function (comatose, grossly abnormal respiratory pattern, and lack of purposeful movement) before death.

Also, two dogs from the 39 degrees Celsius group died within 24 h of the ischemic insult as a result of profound central neurologic damage. Based upon the preestablished protocol criteria (outlined above), these dogs were included in the final neurologic function analysis, but not the final histologic analysis.

Twenty-one dogs were included in the functional outcome analysis (n = 7 per group). Functional neurologic deficits in the reference group ranged from none (a percent function score of 100) to moderate ataxia, slightly depressed consciousness, and minimally impaired cranial nerve function (a percent function score of 90, Figure 2). Dogs in the 38 degrees Celsius group were more variable in outcome at 72 h postischemia. Their functional neurologic deficits ranged from moderate ataxia (a percent function score of 95) to neurogenic death (a percent function score of 0). Dogs in the 39 degrees Celsius group all sustained severe neurologic deficits ranging from extended-spastic hind limbs and blindness (a percent function score of 65) to neurogenic death (a percent function score of 0). When compared with the reference group, both the 38 degrees Celsius and 39 degrees Celsius dogs had significantly worse neurologic function scores (P < 0.01 and 0.001, respectively).

Nineteen dogs (n = 7 in both the reference and 38 degrees Celsius groups and n = 5 in the 39 degrees Celsius group) fulfilled protocol criteria for inclusion in the histologic analysis.

In all groups, histologic evidence of ischemic injury was most noticeable in gray matter structures, and these same structures were most influenced by temperature changes (Table 2). When compared with the reference group, both 38 degrees Celsius and 39 degrees Celsius dogs had significantly worse total histopathology scores (P < 0.01 for both; Figure 3). In general, significant differences were less common in 38 degrees Celsius dogs (3 of 19 brain areas, total score, and 1 of 3 hippocampal subdivisions) than in 39 degrees Celsius dogs (6 of 19 brain regions, total score, and 3 of 3 hippocampal subdivisions).

When data from all three groups were combined, there was a significant correlation between overall neurologic function and histopathology rank scores (n = 19 pairs of data; r^s= 0.96; P < 0.001) (Figure 4).

It is well appreciated that alterations in temperature can affect the brain's ability to survive an ischemic insult. This phenomenon has been demonstrated repeatedly in humans subjected to profound hypothermia and circulatory arrest to facilitate the surgical repair of complex cardiac or cerebrovascular anomalies. [21-23]When temperature is reduced to 15-20 degrees Celsius, the human brain can tolerate approximately 1 h of circulatory arrest without sustaining permanent injury. [21-23]The proposed mechanism of protection during profound hypothermia is a parallel reduction in the cerebral metabolic rate of oxygen consumption. [24].

Early attempts to use moderate hypothermia for cerebral protection, outside the setting of assisted circulation, resulted in no benefit. Instead, it enhanced morbidity and mortality resulting from systemic physiologic aberrations (e.g., depression of cardiac output, regional blood flow disturbances, and severe metabolic acidosis). [25,26]Perhaps because of this dismal result, there was little interest in examining the effect of mild alterations in temperature (i.e., alterations that would have minimal effects on basal metabolic rate) on outcome. This trend has dramatically changed in recent years.

Studies in models of global cerebral ischemia have reported that temperature alterations of 2-6 degrees Celsius can significantly influence postischemic neurologic function [7-10]and histopathology. [4-6,11,12].

Our study expanded on this collective research by examining the effects of smaller temperature changes, a greater number of histologic sites, and neurologic function using an extensive scoring scale. We found that brain regions rich in white matter were resistant to ischemic damage and its modulation by temperature (Table 2). In contrast, structures rich in gray matter were more susceptible to ischemic injury and more likely to be influenced by temperature increases (Table 2).

Our study also discovered that temperature changes of either 1 degree Celsius or 2 degrees Celsius significantly altered functional outcome (Figure 2). For example, dogs maintained at 37 degrees Celsius were able to stand, walk, feed themselves, and interact sociably with their handlers. Conversely, dogs maintained at 39 degrees Celsius were either comatose or died from cerebral ischemia. Dogs maintained at 38 degrees Celsius were intermediate between 37 degrees Celsius and 39 degrees Celsius dogs, but significantly worse than 37 degrees Celsius dogs. There also was a highly significant correlation between the functional and histologic assessment of postischemic injury (Figure 4).

Our observation, that temperature changes of 2 degrees Celsius or less can significantly affect postischemic neurologic outcome, is consistent with the recent report by Warner et al. [27]Using a model of focal cerebral ischemia (transient middle cerebral artery occlusion) in rats, they discovered that a 1.2 degrees Celsius increase in pericranial temperature adversely affected postischemic neurologic function and total infarct volume.

The physiologic basis of small temperature changes producing significant alterations in postischemic neurologic outcome is not clear. Increasing or decreasing temperature 1 degree Celsius or 2 degrees Celsius should increase or decrease cerebral metabolic rate of oxygen consumption by a mere 7-14%, respectively. [28,29]Such changes in temperature are probably insufficient to produce alterations in the intraischemic depletion of energy substrates or accumulation of toxic metabolites by an amount that could meaningfully alter functional and histopathologic outcome. [30]For example, Busto et al. [5]and Natale and D'Alecy [7]reported an improved neurologic outcome in subjects exposed to mild hypothermia during complete cerebral ischemia. However, under the same experimental conditions, they were unable to detect any significant differences in intraischemic depletion of energy substrates [5](glycogen, glucose, adenosine triphosphate, or phosphocreatine) or lactate accumulation [5,7]between hypo- and normothermic groups. Other proposed mechanisms by which small temperature changes can affect outcome include alterations in membrane stability (including the blood-brain barrier), membrane depolarization, temperature-induced ion homeostasis (including calcium fluxes), neurotransmitter release or reuptake (e.g., glutamate or aspartate), enzyme function (e.g., phospholipase or xanthine oxidase activity), and free radical production or endogenous scavenging. [31,32].

In interpreting our data, three methodologic factors should be considered: the extent of injury in the reference group, the temperature of the reference group, and postischemic temperature management. Because we planned to evaluate the correlation between functional and histopathologic outcome after ischemia, it was necessary that a large number of animals survive more than 24 h, and, thus, be candidates for histologic analysis. [17,18]To facilitate this, we designed the study so that the reference group would sustain minimal neurologic injury. The reference group was compared with groups in which we anticipated a detrimental effect of temperature.

We maintained the reference animals at 37 degrees Celsius, even though true normothermia is 38.9 degrees Celsius in the awake dog. [33]The use of a 37 degrees Celsius reference group allowed us to increase brain temperature by 1 degree Celsius or 2 degrees Celsius without introducing temperature-dependent, ischemia-independent, neuronal injury into the model. Specifically, it is well appreciated that cerebral hyperthermia can cause irreversible brain injury, regardless of whether there is a coexisting ischemic insult or not. [34]It should be noted that the use of a temperature other than true normothermia in the reference group is not unusual. The practice is consistent with the previous experience of our laboratory [13,18,35,36]and other laboratories [37-39]using canine models of complete cerebral ischemia.

In all study groups, temperature was maintained at 37 degrees Celsius for the final portion of the dogs' stay in the acute study environment (Figure 1). It is unlikely that a post-ischemic deviation from true normothermia had any bearing on the final results. Specifically, Kuboyama et al. [8]reported that, in a model of global cerebral ischemia, delayed (> 15 min after recirculation) temperature alterations did not affect neurologic outcome.

In the context of this study design, we discovered that dogs maintained at 39 degrees Celsius during ischemia had functional and histopathologic outcomes that were significantly worse than dogs maintained at 37 degrees Celsius, the historical reference temperature in this model (Figure 2and Figure 3and Table 2). Temperature changes of half this amount (1 degree Celsius in dogs maintained at 38 degrees Celsius) also produced a significantly worse outcome. It became apparent that study design was critical to the testing of our hypotheses because temperature changes of a mere 2 degrees Celsius had virtually exhausted the ability of the scoring scale to evaluate neurologic function. That is, at 72 h postischemia, reference group dogs were normal or near-normal and 39 degrees Celsius dogs were severely injured or dead.

Traditionally, clinicians have tolerated mild temperature increases in critically ill patients, or they have treated the increases with antipyretic drugs or physical means, using reactionary techniques. Such a therapeutic approach often results in ongoing mild hyperthermia during periods of potential risk for an ischemic neurologic insult. For example, Young and colleagues** have observed that temperature increases to greater or equal to 38 degrees Celsius are a common occurrence during the initial 24 h after craniotomy. Should the patient experience an ongoing or new-onset ischemic insult, there may be, as stated by Minamisawa et al., [4]"devastating effects of fever." Thus, it is our opinion that, in select, high-risk patients, traditional temperature management should be replaced with more aggressive, proactive techniques.

These temperature management concepts may have implications for cardiac surgery patients as well. For example, peri- and intraoperative cerebral ischemia may occur in patients having major cardiac surgery, with resulting postoperative neurologic and cognitive deficits. [14,40-43]A current trend in cardiac surgery is "warm" cardiopulmonary bypass: bypass conducted during systemic normothermia or near-normothermia. [44-47]With this technique, brain temperature is typically not measured. It is possible that the brain, with its inherently high rate of metabolism and heat production, may experience mild hyperthermia--independent of systemic temperature--during periods of increased risk for ischemic injury. The possibility of such a detrimental brain-to-core temperature gradient occurring during cardiopulmonary bypass has been confirmed in children undergoing low-flow cardiopulmonary bypass and profound hypothermia. During the rewarming phase, Bissonnette*** discovered that jugular bulb temperatures were often greater or equal to 38 degrees Celsius during periods in which core temperatures were approximately 36 degrees Celsius. This observation suggests that monitoring and control of brain temperature may be warranted in select cardiac surgery patients, to complement the traditional assessment of systemic temperature.

By extrapolating from the results of previous studies, using models of global cerebral ischemia in dogs [7-10]and rats, [4-6,11]it is possible to conclude that the temperature-related phenomena we observed are part of a continuum that occurs over a temperature range of 5.9 degrees Celsius less than [7]to 1.5 degrees Celsius greater than [4,11]true normothermia. This temperature range was derived from the existing outcome literature, and assumed that 38.9 degrees Celsius and 37.5 degrees Celsius represented true normothermia in dogs [33]and rats, [48]respectively. If this interpretation of the literature is correct, and the findings from these studies and our current canine study are transferable to humans, the results suggest that, in patients at risk for imminent ischemic neurologic injury, temperature should be closely monitored and controlled at relatively hypothermic temperatures, provided this intervention does not introduce secondary systemic morbidity (e.g., perioperative myocardial ischemia [49]).

In summary, our study demonstrated, in a canine model of complete cerebral ischemia, that temperature changes of a mere 1 degree Celsius or 2 degrees Celsius resulted in a significant alteration of both neurologic function and histopathology. To the extent that our results are transferable to humans, these data suggest that, in patients at imminent risk for ischemic neurologic injury, body temperature should be closely monitored. Further, the clinician should aggressively treat all episodes of hyperthermia until the patient is no longer at risk for ischemic neurologic injury.

The authors thank Richard Koenig, Audrey Kroening, James Milde, and Rebecca Wilson for their technical support and Kenneth Offord for statistical consultation.

* Tommasino C, Shapiro HM: Effects of nitrous oxide analgesia on local cerebral glucose utilization during immobilization stress (abstract). ANESTHESIOLOGY 59:A334, 1983.

** Young WL: Personal communication. 1994.

*** Bissonnette B: Personal communication. 1993.