Neither exsanguination to pulselessness nor cardiac arrest of 30 min duration can be reversed with complete neurologic recovery using conventional resuscitation methods. Techniques that might buy time for transport, surgical hemostasis, and initiation of cardiopulmonary bypass or other resuscitation methods would be valuable. We hypothesized that an aortic flush with high-volume cold normal saline solution at the start of exsanguination cardiac arrest could rapidly preserve cerebral viability during 30 min of complete global ischemia and achieve good outcome.


Sixteen dogs weighing 20-25 kg were exsanguinated to pulselessness over 5 min, and circulatory arrest was maintained for another 30 min. They were then resuscitated using closed-chest cardiopulmonary bypass and had assisted circulation for 2 h, mild hypothermia (34 degrees C) for 12 h, controlled ventilation for 20 h, and intensive care to outcome evaluation at 72 h. Two minutes after the onset of circulatory arrest, the dogs received a flush of normal saline solution at 4 degrees C into the aorta (cephalad) via a balloon catheter. Group I (n = 6) received a flush of 25 ml/kg saline with the balloon in the thoracic aorta; group II (n = 7) received a flush of 100 ml/kg saline with the balloon in the abdominal aorta.


The aortic flush decreased mean tympanic membrane temperature (Tty) in group I from 37.6 +/- 0.1 to 33.3 +/- 1.6 degrees C and in group II from 37.5 +/- 0.1 to 28.3 +/- 2.4 degrees C (P = 0.001). In group 1, four dogs achieved overall performance category (OPC) 4 (coma), and 2 dogs achieved OPC 5 (brain death). In group II, 4 dogs achieved OPC 1 (normal), and 3 dogs achieved OPC 2 (moderate disability). Median (interquartile range [IQR]) neurologic deficit scores (NDS 0-10% = normal; NDS 100% = brain death) were 69% (56-99%) in group I versus 4% (0-15%) in group II (P = 0.003). Median total brain histologic damage scores (HDS 0 = no damage; > 100 = extensive damage; 1,064 = maximal damage) were 144 (74-168) in group I versus 18 (3-36) in group II (P = 0.004); in three dogs from group II, the brain was histologically normal (HDS 0-5).


A single high-volume flush of cold saline (4 degrees C) into the abdominal aorta given 2 min after the onset of cardiac arrest rapidly induces moderate-to-deep cerebral hypothermia and can result in survival without functional or histologic brain damage, even after 30 min of no blood flow.

NORMOTHERMIC cardiac arrest (CA;i.e.,  temporary complete global brain ischemia) lasting 5 min or longer and reversed by standard resuscitation is almost invariably followed by brain damage. 1–5Hypothermia induced before arrest (protection) is more likely to mitigate postischemic brain damage than when induced after arrest (resuscitation). 2,4,6This study explored hypothermia induced during arrest (preservation). After normothermic CA of longer than 10 min in dogs, restoration of spontaneous circulation requires emergency cardiopulmonary bypass (CPB). 7We suspect that many military or civilian victims of traumatic exsanguination 8and presently unresuscitable patients with normovolemic sudden cardiac death 9could be saved with rapid preservation of brain and heart to buy time for transport, repair, and resuscitation with CPB. About one half of out-of-hospital cardiopulmonary–cerebral resuscitation attempts for normovolemic CA fail to restore heartbeat, 9and many long-term survivors suffer permanent brain damage. 2For sudden cerebral ischemia, the rapid loss of energy, 5disappointing pharmacologic cerebral resuscitation trials, 2,4and the benefits from even mild preservative and resuscitative cerebral hypothermia 2,3,6,10–12have been reported.

In 1984, Bellamy and Safar, 8considering combat casualties killed in action, recommended research into rapid induction of preservation of the organism for transport and surgical hemostasis without pulse, to be followed by delayed resuscitation to survival without brain damage. Using dog outcome models of exsanguination CA and CPB for the induction of profound hypothermia (5–10°C), we achieved cerebral preservation during complete circulatory arrest of 60 min, 13but not 120 min. 14We have recently documented effective rapid induction of cerebral hypothermia without CPB, using aortic arch cold flush to induce hypothermic preservation within 5 min of no blood flow for CA 15 min 15,16or CA 20 min. 17 

We hypothesized 18that flushing the aorta (and, hence, the brain) with normal saline at 4°C immediately after the onset of CA of 30 min can achieve survival without brain damage, and that the cold flush must include the spinal cord and abdominal viscera, which cannot tolerate normothermic ischemia of longer than 20 min. 17 

Materials and Methods

This study was approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh, Pittsburgh, Pennsylvania, and followed United States national guidelines for the treatment of animals. Sixteen male custom-bred hunting dogs, 8–12 months of age and with a body weight of 20–25 kg, were used. All experiments were performed by the same team within 1 y in mixed sequence without systematic randomization.


The dogs were fasted overnight with free access to water. After premedication with 10 mg/kg ketamine intramuscularly, anesthesia was induced with 50%:50% N2O:O2and 2–4% halothane via  mask. After tracheal intubation, the dogs were mechanically ventilated with tidal volumes of 15 ml/kg and positive end-expiratory pressure of 5 cmH2O, without paralysis. Anesthesia was continued with nitrous oxide and 0.5–1.5% halothane titrated to sustain normotension. The ventilatory rate was adjusted to achieve normocapnia (arterial partial pressure of carbon dioxide [Paco2], 35–40 mmHg), with an end-tidal carbon dioxide level of 4–5%. Electrocardiogram electrodes were attached to the extremities, and a pulse oximeter probe was placed on the tongue. Gastric and bladder catheters were inserted. Temperature probes were inserted for measuring tympanic membrane (Tty), esophageal (Tes), and rectal temperatures (Tr). Dextrose, 5%, in 5 ml · kg−1· h−1NaCl, 0.45%, was administered via  a peripheral intravenous line to maintain central venous pressure at a level higher than 3 mmHg.

A PE 90 catheter (Becton, Dickinson Co., Parsippany, NJ) was surgically inserted into the left femoral artery for monitoring of arterial pressure and blood sampling. A 7.5-French balloon catheter (Intellicath Continuous Cardiac Output Thermodilution Catheter; Baxter Co., Irvine, CA) was inserted via  the left femoral vein into the pulmonary artery for pressure monitoring, continuous cardiac output determination (Vigilance Monitor software 4.42, Baxter Co., Irvine, CA), temperature measurements (Tpa), and blood sampling. The right femoral artery was cannulated with a prototype 8-French catheter with one hole at the distal end (Cardeon Corp., Cupertino, CA) and an inflatable balloon 1 cm from the tip which, when inflated with 1.0–1.5 ml saline, occluded the aorta (as determined by disappearance of femoral artery pressure). The catheter had an internal diameter of 2.24 mm. To verify that the balloon was placed in the abdominal or descending thoracic aorta, the length was marked prior to insertion. The right external jugular vein was cannulated with a multiple-holed, spiral-reinforced, 18-French plastic cannula which was advanced to the level of the right atrium. This was used for venous bleeding and for venous return to the CPB system.

Arterial and central venous pressures and electrocardiography were continuously recorded on a polygraph (Grass Model 7D Polygraph; Astro-Med Inc., West Warwick, RI). Pulmonary artery pressure, pulmonary artery occlusion pressure, cardiac output, arterial and mixed venous blood gases, hemoglobin, hematocrit, sodium, potassium, glucose, and lactate were measured at regular intervals. Blood gases were measured at 37.5°C without correction for body temperature. Just before start of the insult, Tty was controlled at 37.5 ± 0.1°C using a heating blanket and lamp.

Cardiac Arrest

After two baseline measurements, heating devices, intravenous fluids, and halothane were discontinued while the dogs were weaned to spontaneous breathing of N2O:O275%:25%via  an endotracheal tube. When the canthal reflex returned, hemorrhage was initiated. Over a 5-min period the dogs were bled via  the arterial and venous cannulae into bags containing citrate. The hemorrhage was controlled to achieve mean arterial pressure (MAP) of 40 mmHg at 2 min, 30 mmHg at 3 min, and 20 mmHg at 4 min. At 5 min, ventricular fibrillation (VF) was induced with a transthoracic shock of 95–105 V alternating current at 60 Hz for 2 s, repeated as needed. Total no-flow time was 30 min.

Preservation by Aortic Flush

The flush strategies chosen were based on previous results 15–17and pilot experiments (see Discussion). Two min after the onset of VF, the balloon of the aortic catheter was inflated with 1.0–1.5 ml saline to occlude the aorta. The aorta was flushed with normal saline solution using a roller pump. In group I, with the balloon in the thoracic aorta, 25 ml/kg of 4°C saline was infused over 1 min. In group II, with the balloon in the lower abdominal aorta, 100 ml/kg of 4°C saline was infused over 4 min. (We had found in a pilot experiment that mean carotid artery pressure during this kind of aortic arch flush is about 100 mmHg). The venae cavae were allowed to drain into bags during the aortic flush. After the flush, during CA, the aortic catheter was replaced by a short, 7- to 8-gauge arterial CPB cannula to optimize flow for later resuscitation by CPB.


After CA 30 min, reperfusion was achieved with CPB. 3,7The CPB system included a centrifugal pump (Biomedicus, Eden Prairie, MN), a hollow-fiber membrane oxygenator, and a heat exchanger (Medtronic, Anaheim, CA). The circuit was primed with 400 ml of Dextran 40 10% in saline plus Ringer’s solution, 50%:50%. Sodium bicarbonate, 2 mEq/kg, and 1500 U heparin were added. Just before start of CPB, an additional 1500 U of heparin and 2 mEq/kg sodium bicarbonate were injected into the circuit. The dogs were paralyzed with 0.1 mg/kg intravenous pancuronium. CPB was started with a flow of 100 ml · kg−1· min−1, and enough shed blood was reinfused to achieve a central venous pressure of 10–15 mmHg. Repetitive doses of 0.01 mg/kg epinephrine were given into the femoral artery, if necessary, to increase the CBP-generated MAP to 100 mmHg. After CPB of 2–5 min with vigorous VF, defibrillation was attempted with an external direct-current counter-shock of 150 J. If necessary, shocks were repeated at 200 J, and a maximum of 300 J. Gas flow through the oxygenator was adjusted to keep Paco2at 30–35 mmHg and arterial partial pressure of oxygen (Pao2) greater than or equal to 100 mmHg. Controlled ventilation at a rate of 8–10 breaths/min was resumed to prevent atelectasis. CPB controlled Tty at 34°C from reperfusion to 12 h. The intravenous fluids were restarted at 100 ml/h. A base deficit of more than 6 mEq/l was corrected with sodium bicarbonate. When restoration of spontaneous circulation was established, CPB was continued for assisted circulation to 120 min, with 100 ml · kg−1· min−1for 60 min, 75 ml · kg−1· min−1for 30 min, and 50 ml · kg−1· min−1for 30 min. During CPB, the activated clotting time was maintained at greater than 300 s with additional heparin if needed. After restoration of circulation, norepinephrine by intravenous drip was titrated to achieve initially a brief hypertension with MAP greater than or equal to 150 mmHg, followed by MAP controlled at 90–150 mmHg. The remaining shed blood was gradually reinfused, avoiding central venous pressure higher than 15 mmHg.

Intensive Care

Controlled ventilation was continued to 20 h with N2O:O250%:50%. The dogs remained paralyzed with pancuronium until 20 h to assure a steady state of cardiovascular–pulmonary variables. For analgesia in case of suspected “stress” (mydriasis, tachycardia, hypertension), fentanyl boluses of 50 or 100 μg per dog were titrated intravenously to maintain the pupils at small size and to help reverse severe hypertension (MAP ≥ 150 mmHg) and tachycardia. More potent anesthesia was avoided because it can influence ischemic brain damage. In the dogs of this study, there were no painful stimuli during the 20 h paralysis after CA. There was also residual postischemic cerebral depression evident in all dogs at 20 h (evident after stopping all anesthesia and paralysis). Nevertheless, to be certain that this anesthetic regimen was sufficient, 19two dogs without ischemia and without paralysis were intubated under brief, light halothane and were then ventilated with N2O:O250:50%. Whenever movement, reaction to the endotracheal tube, or widening of pupils occurred, intravenous boluses of 100 μg fentanyl were given. As expected, the required doses were greater in normal dogs than in the post-CA dogs: namely, 100 μg/dog about every 15 min. The canthal reflex also remained active. Nevertheless, there was no purposeful escape behavior and no movement on paw pinch.

After resuscitation, hypotension (MAP < 90 mmHg) was treated with intravenous Ringer’s solution or titrated norepinephrine. After analgesia was assured by the presence of small pupils, severe hypertension (MAP > 150 mmHg) was also controlled with boluses of 0.25–0.5 mg/kg labetalol or 0.5–1.0 mg/kg hydralazine. For infection prophylaxis, the dogs received 250 mg cefazolin intravenously every 8 h. Respiratory care included rotation, suctioning, and “sighing” at regular intervals. At 20–24 h after resuscitation, paralysis was reversed with 0.05 mg/kg neostigmine plus 0.025 mg/kg atropine administered intravenously. Weaning to spontaneous breathing was accomplished via  endotracheal tube. The dogs were extubated when they were able to maintain normal Pao2and Paco2with spontaneous breathing, upper airway reflexes had returned, and circulation was stable. When dogs appeared awake, the catheters were removed under brief, light nitrous oxide–halothane anesthesia by mask, and the dogs were transferred to a stepdown area in the intensive care unit for continuous monitoring and life support by technicians and critical care physicians. Suspected discomfort (howling, restlessness), seizures, running movements, or opisthotonos were controlled with titrated intravenous 0.2-to 0.3-mg/kg boluses of diazepam. For diazepam requirements, see Results. Tty was controlled with external cooling and warming at 34°C to 12 h 12and at 37.5°C from 12–72 h.

Outcome Evaluation


The methods used for the evaluation of outcome in terms of function until 72 h and morphology at 72 h have been described elsewhere. 20–23Performance was evaluated according to overall performance categories (OPC) 1–5, where OPC 1 = normal; 2 = moderate disability; 3 = severe disability; 4 = coma; and 5 = brain death or death. Neurologic function was evaluated as neurologic deficit scores (NDS) 0–100%, where NDS 0–10%= normal and 100%= brain death. NDS included level of consciousness, breathing pattern, cranial nerve function, sensory and motor function, and behavior. Beginning 24 h after resuscitation, OPC and NDS were evaluated in dogs weaned from paralysis a nd fentanyl, and evaluation was continued every 8 h. The final evaluations at 72 h were independently recorded and agreed upon by two team members. Attempts were made to discontinue any sedation at least 4 h prior to final evaluation. If necessary, diazepam effect was reversed with 0.1 mg flumazenil intravenously, repeated as needed.


After functional outcome evaluation, the dogs were reanesthetized in the same manner as initially. The left hemithorax was opened, and the proximal descending aorta was ligated. A large-bore cannula was inserted proximal to the ligature. The dogs were then euthanized by infusing approximately 2 l of paraformaldehyde 3% (pH 7.4) with a roller pump with the right atrium opened, until clear fluid returned from the vena cava. A complete necropsy was performed. Macroscopic damage in gut and heart was estimated (dark hemorrhages, pale necroses), and samples of extracerebral organs were taken for histologic examination. After 1 h of brain fixation in situ,  the skull was sawed open and the brain was removed. 23Each brain was cut into 3-mm-thick coronal sections, which were immersed in paraformaldehyde for further fixation. Six selected slices were paraffin-embedded, cut into sections 4 μm thick, and stained with hematoxylin–eosin–phloxine. Using light microscopy, the same pathologist (Dr. Radovsky), who was unaware of treatment assignments, scored 19 distinct anatomic brain regions (see Results) for severity and extent of ischemic neuronal changes (shrunken, eosinophilic neuron with pyknotic nucleus), infarcts, and edema. 23For each region examined, the severity and extent of these lesions were assessed as HDS on a four-point scale, where minimal = 1+; moderate = 2+; severe = 3+; and maximal = 4+. The points were then multiplied by a weighting factor depending on the type of lesion (infarction 4×, neuronal necrosis 2×, and edema 1×). A maximal HDS of 56 per region was obtained if all three types of lesions occurred with maximal extent. Total HDS for the entire brain were obtained with a maximal possible score of 1,064 for all 19 regions combined. A total HDS of more than 100 represented extensive damage and had previously correlated with severe final NDS. 10–12,14,20–23In two pilot experiments (see Discussion), the spinal cord was also removed and macroscopically abnormal sections were processed, stained with hematoxylin–eosin, and evaluated by light microscopy.

Statistical Analysis

Dogs that did not follow protocol or that died from extracerebral causes were excluded from analysis. Brain death was included as a cerebral outcome. Data are given as mean ± SD or the median and IQR (the difference between the 25th and 75th percentiles) unless otherwise specified. We used the independent-samples t  test or the Mann–Whitney U test for the comparison of continuous variables (physiologic variables, final NDS, and final HDS). We used the chi-square test for trend to determine group differences of final OPC. Since our endpoint was final outcome (at 72 h), changes in OPC, NDS, and HDS over time were not statistically analyzed. To account for the change in temperature over time during arrest, we calculated the area under the temperature curve. All data were computed using SPSS for Windows, release 8.0 (SPSS Inc., Chicago, IL). A P  value less than 0.05 was considered statistically significant.


Of the 16 dogs exsanguinated to CA, three had to be excluded from outcome evaluation: in group I, one of the eight dogs died 9 h after restoration of spontaneous circulation as a result of unrecognized airway obstruction, and one died 36 h after restoration of spontaneous circulation with heart failure caused by heartworms. This left six dogs in the protocol. In group II, one of the eight dogs developed severe oropharyngeal edema of unknown cause post-CA, which made extubation and evaluation of OPC and NDS impossible. This left seven dogs in the protocol. In group I, three of the six dogs included in the protocol developed an increasing need for large doses of norepinephrine; to obtain their brains for histologic evaluation, their OPC and NDS were determined after brief weaning from controlled ventilation prior to anticipated severe hypotension, with MAP still within protocol parameters. They were then reanesthetized at 42 h, 52 h, and 60 h, respectively, for perfusion–fixation and morphologic evaluation. In group II, all seven dogs survived to 72 h.

There were no group differences in extracerebral variables important for cerebral recovery (table 1) at baseline and at 6 h after resuscitation, with the exception of central venous pressure, which was higher in group II at 6 h; lactate, which was higher in group I at 6 h; and O2extraction ratio, which was higher in group II at 6 h. Hematocrit immediately after resuscitation showed no significant intergroup difference (table 2).

Table 1. Physiologic Variables at Baseline and 6 h after Resuscitation

Aortic flush at the start of the arrest was in group I with 25 ml/kg saline at 4°C (n = 6) and in group II with 100 ml/kg saline at 4°C (n = 7). Data are given as median and interquartile range.

* This high 75th-percentile arterial oxygen tension (Pao2) value, above the expected value for a fractional inspired oxygen tension (Fio2) of 0.5, was caused by measuring the value for one dog by error under an Fio2of 1.0. †P < 0.05 comparing group I versus  group II.

ROSC = restoration of spontaneous circulation; Paco2= arterial carbon dioxide tension.

Table 1. Physiologic Variables at Baseline and 6 h after Resuscitation
Table 1. Physiologic Variables at Baseline and 6 h after Resuscitation

Table 2. Resuscitation and Restoration of Spontaneous Circulation

Aortic flush at start of arrest was in group I with 25 ml/kg saline at 4°C (n = 6) and in group II with 100 ml/kg saline at 4°C (n = 7). Data are given as median and interquartile range.

* Start of hypertensive bout = time after start CPB. † Duration of hypertensive bout = time with mean arterial pressure (MAP) greater than 150 mmHg.

ROSC = restoration of spontaneous circulation; CPB = cardiopulmonary bypass.

Table 2. Resuscitation and Restoration of Spontaneous Circulation
Table 2. Resuscitation and Restoration of Spontaneous Circulation

There was no difference between groups in the required number and energy of defibrillating counter-shocks, total epinephrine doses, and bicarbonate requirement (table 2). The time to achieve restoration of spontaneous circulation was significantly longer and total norepinephrine requirements significantly higher in group I. The beginning of the induced brief hypertension was significantly earlier and the peak MAP significantly higher in group II versus  group I (table 2). The duration of the hypertension (time with MAP >150 mmHg) varied greatly without intergroup difference.

Tty (fig. 1) just prior to the insult was 37.6 ± 0.1°C in group I versus  37.5 ± 0.1°C in group II (P = 0.1). Saline flush rapidly decreased Tty during CA to a minimum of 33.3 ± 1.6°C in group I versus  28.3 ± 2.4°C in group II (P = 0.001). Tty change over time (area under the curve) during CA was significantly different between the two groups (fig. 1). The lowest Tpa during CA was 28.2 ± 1.4°C in group I versus  20.9 ± 2.8°C in group II (P < 0.001). The lowest Tes was 32.9 ± 4.2°C in group I versus  21.1 ± 5.5°C in group II (P = 0.002). Tr remained normothermic and did not differ from baseline values during CA.

Fig. 1. Tympanic membrane temperatures (Tty) during exsanguination cardiac arrest of 30 min no-flow and resuscitation with cardiopulmonary bypass. Aortic flush with balloon catheter was with 25 ml/kg saline into the aortic arch in group I and with 100 ml/kg saline into the abdominal aorta in group II; in both cases, the saline was at a temperature of 4°C. Temperature areas under the curve during arrest were significantly different between groups (P = 0.003). NSS = normal saline solution.

Fig. 1. Tympanic membrane temperatures (Tty) during exsanguination cardiac arrest of 30 min no-flow and resuscitation with cardiopulmonary bypass. Aortic flush with balloon catheter was with 25 ml/kg saline into the aortic arch in group I and with 100 ml/kg saline into the abdominal aorta in group II; in both cases, the saline was at a temperature of 4°C. Temperature areas under the curve during arrest were significantly different between groups (P = 0.003). NSS = normal saline solution.


OPC and NDS between 24 and 72 h improved less in group I than in group II. Final OPCs were better in group II, with 100 ml/kg low aortic flush, than in group I, with 25 ml/kg high aortic flush (P = 0.001;table 3).

Table 3. Final 72-h Outcomes after Exsanguination and No Flow of 30 min for Each Dog

Exceptions were dogs 4, 5, and 6 of group I, which had to be terminated at 60, 52, and 42 h (see text). Dog 1 of group II was allowed to survive, was adopted, and is normal at 1 yr. Aortic flush at start of arrest was in group I with 25 ml/kg saline at 4°C (n = 6), with the tip of the balloon catheter in the thoracic aorta, and in group II with 100 ml/kg saline at 4°C (n = 7), with the tip of the balloon catheter in the abdominal aorta.

*P = 0.001. †P = 0.003. P = 0.004. § No diazepam required because anesthetized with fentanyl until final evaluation (overall performance category [OPC; 1 = normal, 5 = brain death] and neurologic deficit score [NDS; 0–100%] after fentanyl was reversed). ∥ Weakness of the hind legs (cerebral performance was normal).

Tty = lowest tympanic membrane temperature during cardiac arrest; diazepam = total dose required; HDS = total brain histologic damage score; NA = not available.

Table 3. Final 72-h Outcomes after Exsanguination and No Flow of 30 min for Each Dog
Table 3. Final 72-h Outcomes after Exsanguination and No Flow of 30 min for Each Dog

Final NDS (table 3) was 69% (IQR, 56–99%) in group I (including the three dogs that had to be killed before 72 h) versus  4% (IQR, 0–15%) in group II (P = 0.003). Four dogs in group II with OPC 1 had NDS 0–10% (i.e.,  were normal). The slightly higher NDS in the other three dogs in group II reflected hind leg weakness and difficulty in walking or standing; however, they showed normal cerebral function and behavior.

The total dose of diazepam administered in each dog is shown in table 3. Dogs 1, 2, and 3 of group I received 0.2 mg flumazenil intravenously before evaluation of final OPC and NDS. Dogs 4, 5, and 6 of group I remained intubated under fentanyl and hence received no diazepam. In these dogs, fentanyl was reversed with naloxone shortly before euthanasia for evaluation of final OPC and NDS.

At necropsy, all dogs in group II were macroscopically normal. In group I, all dogs had moderate to severe (widespread) hemorrhagic areas in the gut mucosa and over the epicardium and endocardium.

On brain histology at 72 h, total brain HDS was 144 (IQR, 74–168) in group I versus  18 (IQR, 3–36) in group II (P = 0.004). Two dogs in group II had essentially normal brains on light microscopic examination, with total HDS of 0 and 4, respectively (table 3). Regional brain HDS (fig. 2) in frontal, parietal, occipital, and temporal cortices, hippocampus, dentate gyrus, putamen, amygdala, and thalamus were higher in group I than in group II (P < 0.05;fig. 2). In group II, no ischemic changes were found in the most vulnerable neurons, i.e. , in the hippocampus (three dogs) and in the cerebellar Purkinje neurons (four dogs). In group I, the insular cortex was damaged in one dog, and the midbrain and substantia nigra were damaged in another dog. The medulla and dentate nucleus did not show any histologic damage in either group (these regions are not shown in fig. 2). Two dogs in group I and one dog in group II showed small infarcts. Spinal cord lesions in two pilot experiments are discussed later (see Discussion).

Fig. 2. Regional histologic damage scores (HDS). The box represents the interquartile range. The line across each box indicates the median. The whiskers indicate the highest and lowest values. The x indicates outliers (values between 1.5 and 3 box-lengths from the upper or lower edge of the box). The + indicates extremes (values >3 box-lengths from the upper or lower edge of the box). *P < 0.05. Gr = group; NSS = normal saline solution.

Fig. 2. Regional histologic damage scores (HDS). The box represents the interquartile range. The line across each box indicates the median. The whiskers indicate the highest and lowest values. The x indicates outliers (values between 1.5 and 3 box-lengths from the upper or lower edge of the box). The + indicates extremes (values >3 box-lengths from the upper or lower edge of the box). *P < 0.05. Gr = group; NSS = normal saline solution.


This study documents that a single flush of 100 ml/kg cold saline (4°C) into the abdominal aorta (not into the thoracic aorta) via  balloon catheter at the start of 30 min CA can allow survival without brain damage. To achieve similar results in a human, about 7 l of cold saline would be required. The mechanism is presumably related to the rapid induction of moderate to deep cerebral hypothermia (group II) at a rate much faster than could be achieved with surface cooling. This study also documents that in CA of 30 min, from which the healthy dog heart is unresuscitable under normothermia, 3,24,25mild hypothermia (group I) is sufficient to preserve the heart’s ability to beat, although severely damaged, but is insufficient for cerebral preservation. Aortic arch flush in group I left the nonflushed normothermic spinal cord and the viscera damaged.

Elective, slow, protective, pre-CA cooling has been practiced since the 1950s. 26,27This study provides the first documentation of rapid preservative cooling during arrest. The degree of cerebral preservation seemed better than that seen previously with cooling to mild 2,3or moderate hypothermia 26–28induced before normovolemic CA. Normothermic aortic arch flush provided no significant preservation in dogs using the same model. 15After normothermic VF-CA of 12.5 min, deep hypothermia by CPB does not result in better outcome than mild post-CA cooling. 10 

Pilot Experiments on Flush

This study was preceded by five pilot experiments in dogs to explore flush volume and catheter balloon position at start of CA 30 min to outcome at 72 h. Aortic arch flush with 50 ml/kg saline at 4°C resulted in one dog at 72 h with OPC 3, NDS 50%, and HDS 172 and with severe hemorrhagic areas in the duodenal mucosa, and in two dogs with normal cerebral function but with OPC 2 because of spastic paralysis of the hind legs (NDS 32% and 29%; HDS 20 and 40, respectively). Hearts and intestines were macroscopically normal. Two other dogs, flushed with 100 ml/kg saline at 4°C with the balloon at the level of the diaphragm, achieved normal cerebral function but with OPC 2 because of spastic paralysis of the hind legs (NDS 29% and 29%; HDS 16 and 12, respectively). Hearts and intestines were macroscopically normal. In these two dogs, the lumbar spinal cords showed histologic evidence of extensive degenerative changes, primarily in the ventral gray columns. This was characterized by neuronal degeneration with swelling and chromatolysis, neuronophagia, and prominent gliosis, including occasional glial nodule formation. Because of these pilot experiments, we decided for the final study group II to place the tip of the balloon catheter into the lower abdominal aorta.

Adjunctive Study of DNA Damage

All dog brains in the study were also stained using the TUNEL method. 29Evidence of DNA fragmentation was found predominantly in neurons which, on regular staining, appeared shrunken and had condensed nuclei. A new DNA damage scoring method was developed and revealed that these scores correlated with HDS.

Dietrich et al. , 30had shown that brief (4 h) postarrest mild hypothermia after normothermic incomplete forebrain ischemia in rats postpones but does not permanently salvage hippocampal neurons at 2 months. This is not relevant for our study, which produced moderate intraischemic (preservative) hypothermia, which in the Dietrich study gave lasting salvage of neurons at 2 months. Additionally, we used prolonged (12 h) post-CA mild hypothermia which, in a forebrain ischemia rat study by Colbourne et al. , 31gave permanent benefit. Our dog 1 from group II (table 3) is functionally normal 1 y after CA.

Attempts at cerebral resuscitation with drugs have so far been disappointing. 2,4,32In the mid-1980s we resumed research on resuscitative moderate hypothermia (30°C) after normothermic CA. 2,33Breakthrough effects in dogs on outcome, however, were documented only when mild hypothermia (34–36°C), which is simpler and safer than moderate hypothermia, was discovered to improve cerebral outcome when induced before VF-CA up to 15 min, 2,3,7and even when induced after normothermic VF-CA of 10–12.5 min. 2,10–12Mild resuscitative hypothermia essentially normalized cerebral outcome after VF-CA of 11 min when combined with cerebral blood flow–promoting measures. 12Thus, protective–preservative hypothermia, induced and reversed by CPB, has been shown to preserve the brain and whole organism for VF-CA up to 15 min at about 35°C, 3,7for CA up to 20 min at about 30°C, 2,7,26for CA up to 30 min at about 20°C (deep hypothermia), 28and for CA up to 60 min at about 10°C (profound hypothermia). 13For transport and repair in exsanguinated trauma victims, it is assumed that at least 30 min of preservation is needed until CPB can be initiated.

In a steady state, brain tissue temperature seems well reflected in Tty. 34However, when deep hypothermia was rapidly induced during CPB, large temperature gradients between brain temperature and Tty were found. 35In pilot experiments with the same dog model, we found that brain tissue may have been transiently 4–10°C below Tty. This might explain the unexpected complete cerebral preservation achieved in group II during CA of 30 min at Tty 28°C.

With CA of 30 min or longer, the heart needs to be protected to enable restoration of stable circulation. 24,25With aortic arch flush of 25 ml/kg saline (group I), all six dogs had macroscopically severe hemorrhagic damage of the heart. In contrast, with flush of 100 ml/kg saline (group II), all seven dogs had macroscopically normal hearts. In group I, with the low-volume flush and high balloon position, the intestines also showed severe hemorrhagic damage. There was also a greater need for norepinephrine after return of spontaneous circulation to maintain normotension. In group II, abdominal mild hypothermia seemed to have protected the abdominal viscera and spinal cord; this was for tissue supplied by the superior mesenteric artery. Rectal temperature did not deviate from baseline during CA in both groups. The protection of intestines by flush into the abdominal aorta in group II could also have been the result of lower norepinephrine requirement and better overall hemodynamics.

For group I, we debated whether to exclude from outcome evaluation the three of six dogs that had cardiovascular failure requiring increasing amounts of norepinephrine postarrest. To obtain brain tissue, these three dogs were weaned from paralysis and fentanyl early to determine OPC and NDS; they were then euthanized for brain histologic examination at 42, 52, and 60 h rather than at 72 h (table 3, group I, dogs 4, 5, and 6). With comparable damage, earlier HDS would show fewer ischemic neurons than at 72 h. Our exclusion criteria of extracerebral organ failure with cardiovascular–pulmonary variables out of protocol did not apply, because we could maintain normotension, normoxia, and other critical variables until euthanasia. A deleterious effect on the histologic damage by norepinephrine cannot be ruled out. The two dogs that developed brain death (with dilated, fixed pupils) did so despite normotension; they could not be weaned to spontaneous breathing. Group I results confirm that CA 30 min, even with mild cardiac hypothermia, is too severe an insult to expect cardiovascular resuscitability. 3,7,24,25 

The broader objectives of this preservation study include: (1) Helping to save victims of temporarily uncontrollable (internal) traumatic exsanguination, such as combat casualties and civilian trauma victims without severe brain trauma 8; (2) helping emergency medical services save some nontraumatic cases of normovolemic, normothermic, sudden cardiac death who are unresuscitable by standard cardiopulmonary resuscitation 2,9; and (3) enabling selected elective surgical procedures that are feasible only during a prolonged state of no blood flow. 36The last example would not require high-speed preservative cooling. With large fluid volumes and CPB, profound hypothermic asanguinous trickle flow could extend the tolerated preservation time to longer than 3 h. 36 

Various clinically feasible methods for the induction of cerebral hypothermia have been tested in animals and patients since the 1950s. 2,26None is as rapidly effective as aortic cold flush 15–18or CPB with heat exchanger. 10–14Initiation of CPB, however, takes more time than the brain can tolerate under normothermic CA. The percutaneous Seldinger technique, with or without minor cutdown, available for femoral vessels and other vessels, has been shown to be feasible and rapid 37but has not been explored for empty vessels in exsanguination. Almost as rapid as CPB cooling would be intracarotid cooling. 38,39 

For civilian emergency care, the methods described in this report should be feasible now, in the hands of physician-staffed teams of mobile intensive care unit ambulances and in hospital emergency departments. There, CPB should be available to continue with profound hypothermic total circulatory arrest 13or trickle flow 36to achieve preservation of at least 60 min. We are exploring novel methods for access to the aorta with 40and without thoracotomy. 2For field resuscitation by combat medics, rapid access to the aorta without thoracotomy, without CPB, and with small fluid volume at ambient temperature are needed. We therefore systematically explored the preservation achievable in the same dog model with six pharmacologic strategies, using aortic arch flush of 25 ml/kg saline at 24°C, which achieved a Tty of 36°C. 32Thus far, moderate-to-deep hypothermia 15–18proved much more preservative than any of the 14 drugs tested. 32 

We conclude that, in dogs, a single high-volume flush of cold saline (4°C) into the abdominal aorta at the start of exsanguination CA of 30 min rapidly induces moderate-to-deep cerebral hypothermia and can allow survival without functional or histologic brain damage. Future research should explore flush preservation for even longer arrest times, using profound hypothermia plus pharmacologic adjuncts.

Patrick Kochanek, M.D., (Director of the Safar Center for Resuscitation Research and Professor of Anesthesiology/Critical Care Medicine and Pediatrics, University of Pittsburgh) and Lyn Yaffe, M.D., (US Navy Medical Research and Development Command) made valuable suggestions. Robert Wagner, V.M.D., (Central Animal Facility, University of Pittsburgh School of Medicine) helped with assessing spinal cord damage. Rochelle Hans, Carol Korbanik, Jason Stezoski, and Sherman Culver, technicians (Safar Center for Resuscitation Research, Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh) helped with intensive care unit life support. The Cardeon Corp. (Cupertino, California) provided the flush catheter. Howell Sasser, Ph.D., (Carolinas Medical Center, Department of Research Planning and Evaluation, Charlotte, North Carolina) advised on statistical analyses. Patricia Boyle (Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh) helped with editing. Valerie Sabo (Safar Center for Resuscitation Research, University of Pittsburgh) helped with preparation of the manuscript.


Cole S, Corday E: Four minute limit for cardiac resuscitation. JAMA 1956; 161: 1454–8
Safar P: Resuscitation of the ischemic brain, Textbook of Neuroanesthesia with Neurosurgical and Neuroscience Perspectives. Edited by Albin MS. New York, McGraw-Hill, 1997, pp 557–93
Safar P: Resuscitation from clinical death: Pathophysiologic limits and therapeutic potentials. Crit Care Med 1988; 16: 923–41
Warner DS: Effects of anesthetic agents and temperature on the injured brain, Textbook of Neuroanesthesia with Neurosurgical and Neuroscience Perspectives. Edited by Albin MS. New York, McGraw-Hill, 1997, pp 595–611
Michenfelder JD, Theye RA: The effects of anesthesia and hypothermia on canine cerebral ATP and lactate during anoxia produced by decapitation. A nesthesiology 1970; 33: 430–9
Xiao F, Safar P, Radovsky A: Mild protective and resuscitative hypothermia for asphyxial cardiac arrest in rats. Am J Emerg Med 1998; 16: 17–25
Safar P, Abramson NS, Angelos M, Cantadore R, Leonov Y, Levine R, Pretto E, Reich H, Sterz F, Stezoski SW, Tisherman S: Emergency cardiopulmonary bypass for resuscitation from prolonged cardiac arrest. Am J Emerg Med 1990; 8: 55-67
Bellamy R, Safar P, Tisherman SA, Basford R, Bruttig SP, Capone A, Dubick MA, Ernster L, Hattler BG Jr, Hochachka P, Klain M, Kochanek PM, Kofke WA, Lancaster JR, McGowan FX Jr, Oeltgen PR, Severinghaus JW, Taylor MJ, Zar H: Suspended animation for delayed resuscitation. Crit Care Med 1996; 24: S24–47
Eisenberg MS, Horwood BT, Cummins RO, Reynolds-Haertle R, Hearne TR: Cardiac arrest and resuscitation: A tale of 29 cities. Ann Emerg Med 1990; 19: 179–86
Weinrauch V, Safar P, Tisherman S, Kuboyama K, Radovsky A: Beneficial effect of mild hypothermia and detrimental effect of deep hypothermia after cardiac arrest in dogs. Stroke 1992; 23: 1454–62
Kuboyama K, Safar P, Radovsky A, Tisherman SA, Stezoski SW, Alexander H: Delay in cooling negates the beneficial effect of mild resuscitative cerebral hypothermia after cardiac arrest in dogs: A prospective, randomized study. Crit Care Med 1993; 21: 1348–58
Safar P, Xiao F, Radovsky A, Tanigawa K, Ebmeyer U, Bircher N, Alexander H, Stezoski SW: Improved cerebral resuscitation from cardiac arrest in dogs with mild hypothermia plus blood flow promotion. Stroke 1996; 27: 105–13
Capone A, Safar P, Radovsky A, Wang YF, Peitzman A, Tisherman SA: Complete recovery after normothermic hemorrhagic shock and profound hypothermic circulatory arrest of 60 minutes in dogs. J Trauma 1996; 40: 388–95
Tisherman SA, Safar P, Radovsky A, Peitzman A, Marrone G, Kuboyama K, Weinrauch V: Profound hypothermia (less than 10 degrees C) compared with deep hypothermia (15 degrees C) improves neurologic outcome in dogs after two hours’ circulatory arrest induced to enable resuscitative surgery. J Trauma 1991; 31: 1051–61
Prueckner S, Woods RJ, Safar P, Takasu A, Stezoski SM, Radovsky A, Tisherman SA: Mild hypothermic aortic arch flush improves neurologic outcome from exsanguination cardiac arrest in dogs (abstract). Crit Care Med 1999; 27: A105
Woods RJ, Safar P, Takasu A, Prueckner S, Stezoski SW, Stezoski J, Tisherman SA: Hypothermic aortic arch flush for preservation of brain and heart during prolonged exsanguination cardiac arrest in dogs. J Trauma 1999; 47: 1028–36
Behringer W, Prueckner S, Safar P, Radovsky A, Stezoski W, Henchir J, Tisherman SA: Mild hypothermic aortic arch flush for cerebral preservation can achieve normal outcome after 20-minute exsanguination cardiac arrest in dogs. Acad Emerg Med 2000; (in press)
Behringer W, Prueckner S, Tisherman SA, Radovsky A, Safar P: Cold aortic flush for rapid induction of cerebral preservation during exsanguination cardiac arrest of 30 min in dogs (abstract). A nesthesiology 1999; 91: A250
Drummond JC, Todd MM, Saidman LJ: Use of neuromuscular blocking drugs in scientific investigations involving animal subjects. A nesthesiology 1996; 85: 697–9
Nemoto EM, Bleyaert AL, Stezoski SW, Moossy J, Rao GR, Safar P: Global brain ischemia: A reproducible monkey model. Stroke 1977; 8: 558–64
Safar P, Gisvold SE, Vaagenes P, Hendrickx HHL, Bar-Joseph G, Bircher N, Stezoski W, Alexander H: Long-term animal models for the study of global brain ischemia, Protection of Tissues Against Hypoxia. Edited by Wauquier A, et al., eds. Amsterdam, Elsevier, 1982, pp 147–70
Vaagenes P, Cantadore R, Safar P, Moossy J, Rao G, Diven W, Alexander H, Stezoski W: Amelioration of brain damage by lidoflazine after prolonged ventricular fibrillation cardiac arrest in dogs. Crit Care Med 1984; 12: 846–55
Radovsky A, Safar P, Sterz F, Leonov Y, Reich H, Kuboyama K: Regional prevalence and distribution of ischemic neurons in dog brains 96 hours after cardiac arrest of 0 to 20 minutes. Stroke 1995; 26: 2127–33
Angelos M, Safar P, Reich H: A comparison of cardiopulmonary resuscitation with cardiopulmonary bypass after prolonged cardiac arrest in dogs. Reperfusion pressures and neurologic recovery. Resuscitation 1991; 21: 121–35
Cerchiari EL, Safar P, Klein E, Cantadore R, Pinsky M: Cardiovascular function and neurologic outcome after cardiac arrest in dogs. The cardiovascular post-resuscitation syndrome. Resuscitation 1993; 25: 9–33
The Physiology of Induced Hypothermia. Edited by Dripps RD. Washington, DC, National Academy of Sciences, 1956
Bigelow WG, Linsay WK, Greenwood WF: Hypothermia: Its possible role in cardiac surgery. Ann Surg 1950; 132: 849–66
Livesay JJ, Cooley DA, Reul GJ, Walker WE, Frazier OH, Duncan JM, Ott DA: Resection of aortic arch aneurysms: A comparison of hypothermic techniques in 60 patients. Ann Thorac Surg 1983; 36: 19–28
Clark RS, Kochanek PM, Chen M, Watkins SC, Marion DW, Chen J, Hamilton RL, Loeffert JE, Graham SH: Increases in Bcl-2 and cleavage of caspase-1 and caspase-3 in human brain after head injury. FASEB J 1999; 13: 813–21
Dietrich WD, Busto R, Alonso O, Globus MY, Ginsberg MD: Intraischemic but not postischemic brain hypothermia protects chronically following global forebrain ischemia in rats. J Cereb Blood Flow Metab 1993; 13: 541–9
Colbourne F, Li H, Buchan AM: Indefatigable CA1 sector neuroprotection with mild hypothermia induced 6 hours after severe forebrain ischemia in rats. J Cereb Blood Flow Metab 1999; 19: 742–9
Behringer W, Prueckner S, Kentner R, Safar P, Radovsky A, Stezoski W, Wu X, Henchir J, Tisherman SA: Exploration of pharmacologic aortic arch flush strategies for rapid induction of suspended animation (SA) (cerebral preservation) during exsanguination cardiac arrest (ExCA) of 20 min in dogs (abstract). Crit Care Med 1999; 27: A65
Leonov Y, Sterz F, Safar P, Radovsky A: Moderate hypothermia after cardiac arrest of 17 minutes in dogs. Effect on cerebral and cardiac outcome. Stroke 1990; 21: 1600–6
Mariak Z, Lewko J, Luczaj J, Polocki B, White MD: The relationship between directly measured human cerebral and tympanic temperatures during changes in brain temperatures. Eur J Appl Physiol 1994; 69: 545–9
Stone JG, Young WL, Smith CR, Solomon RA, Wald A, Ostapkovich N, Shrebnick DD: Do standard monitoring sites reflect true brain temperature when profound hypothermia is rapidly induced and reversed? A nesthesiology 1995; 82: 344–51
Taylor MJ, Bailes JE, Elrifai AM, Shih SR, Teeple E, Leavitt ML, Baust ML, Maroon JC: A new solution for life without blood. Asanguineous low-flow perfusion of a whole-body perfusate during 3 hours of cardiac arrest and profound hypothermia. Circulation 1995; 91: 431–44
Phillips SJ, Ballentine B, Slonine D, Hall J, Vandehaar J, Kongtahworn, Zeff RH, Skinner JR, Reckmo K, Gray D: Percutaneous initiation of cardiopulmonary bypass. Ann Thorac Surg 1983; 36:223–5
Wolfson SKJ, Selker RG: Carotid perfusion hypothermia for brain surgery using cardiac arrest without bypass. J Surg Res 1973; 14: 449–58
White RJ: Cerebral hypothermia and circulatory arrest. Mayo Clin Proc 1978; 53: 450–8
Rhee P, Talon E, Eifert S, Anderson D, Stanton K, Koustova E, Ling G, Burris D, Kaufmann C, Mongan P, Rich NM, Taylor M, Sun L: Induced hypothermia during emergency department thoracotomy: An animal model. J Trauma 2000; 48: 439–47