Traumatic brain injury (TBI) is among the leading causes of death and permanent disability in urban and battlefield situations. Prognosis is especially grim if hypotension is superimposed on the injury (1–3). There are relatively few large animal models that mimic this condition (4,5), but we recently have observed diffuse axonal injury, as well as many of the other physiologic, neurologic, and histologic characteristics of human TBI 48 h after anesthetized swine were resuscitated from fluid percussion TBI + hemorrhagic hypotension (6,7). In those initial studies, fresh shed (autologous) blood was used for resuscitation. Of course, in real life a trauma patient's own blood is not available for initial fluid resuscitation, and isotonic crystalloid (lactated Ringer's solution or 0.9% NaCl) are the cornerstones of most emergency management guidelines (8,9). One purpose of this present study was to compare the effects of resuscitation with saline vs. shed blood in the experimental model. The primary outcome determinants were intracranial pressure (ICP), cerebral perfusion pressure (CPP), cerebral venous oxygen saturation (ScvO2), and cerebrovascular reactivity to CO2 challenges. If arterial O2 saturation (SaO2) is high and constant, changes in ScvO2 are a sensitive index of the ratio of cerebral blood flow to metabolism (3,9–11). CO2 challenges produce characteristic changes in ICP, CPP, and ScvO2 that reflect cerebral compliance and vascular reactivity (3,9–11). In the first series we tested the hypothesis that CPP, cerebrovascular CO2 reactivity, and ScvO2 would be comparable with blood and crystalloid resuscitation.
The results from that series of experiments led us to address a more timely question in the same basic model. Even if transfusions are the best treatment for hemorrhagic hypotension after TBI, blood transfusions are usually not administered within the “golden hour” after injury, which is the likely window of greatest therapeutic benefit. Several blood substitutes or hemoglobin-based oxygen carrying solutions (HBOCs) are in pre-clinical development or patient trials (12–14), but there is relatively little information on the actions of HBOCs in clinically relevant models that combine tissue injury and hemorrhagic shock (5). These novel compounds have many theoretical benefits over standard crystalloid resuscitation solutions, including their O2 carrying capacity and potential for volume expansion (15). HBOCs even have some theoretical benefits over blood because they can be carried in ambulances or can be administered in the field (16). Diaspirin-cross linked hemoglobin (DCLHb) is probably the most thoroughly characterized HBOC (12–14), but it has not been tested on CPP, ScvO2, and cerebrovascular CO2 reactivity after TBI + hemorrhagic hypotension. Thus, the second purpose of this study was to compare saline vs. DCLHb in the animal model. The magnitude of the ischemic insult was lessened so that neuropathological changes could be evaluated after a 72-h recovery. In this series we tested the hypothesis that the DCLHb resuscitation was superior to standard crystalloid.
MATERIALS AND METHODS
Swine were housed in the animal care facility that is approved by the American Association of Laboratory Animal Care with veterinarians available at all times. All procedures were performed according to NIH guidelines and were pre-approved by the University of Tennessee Animal Care and Use Committee.
Standard surgical procedures in all animals
Farm-raised, cross-bred pigs of either sex (35–45 kg) were fasted overnight with access to water only. Anesthesia was induced with an intramuscular injection of ketamine (30 mg/kg) and xylazine (3.5 mg/kg) and was maintained with a continuous intravenous infusion of fentanyl (30 μg/kg/h), supplemented with continuous infusion of ketamine (90 mg/h) plus xylazine (10 mg/h). After endotracheal intubation, the ventilator was set (10–15 mL/kg tidal volume; 12–15 breaths/min; FiO2 = 0.4) to produce an End Tidal CO2 (ETCO2) = 40 mmHg. One catheter was placed in the right common carotid artery for measurement of systolic arterial blood pressure (SAP), mean arterial blood pressure (MAP), and for blood withdrawal. Samples were drawn at regular intervals for arterial blood gases and electrolytes [Ultra C Analyzer, Nova Biomedical, Waltham, MA] and for complete and differential leukocyte counts [Abbott Cell-Dyn 1600, Abbott Park, IL]. Body temperature was maintained at 37°C to 39°C with a heating/cooling blanket. Heart rate (HR) was measured from the EKG.
Another catheter was placed in the right external jugular vein (Arrowflex Sheath with integral side port, Arrow International, Reading, PA) for fluid resuscitation. An 8F multi-lumen fiberoptic catheter (Edwards Swan Ganz Combo EDV Thermodilution Catheter, Baxter Laboratories, Irvine, CA) was placed in the pulmonary artery for continuous measurement of central venous pressure (CVP), mixed venous O2 saturation (SvO2), and cardiac output (Vigilance Computer, Baxter Laboratories). Cardiac index (CI) was calculated by dividing cardiac output by body weight.
Through a midline craniotomy, a 4F catheter was placed into the superior sagittal sinus for serial sampling of cerebral venous blood gases and electrolytes. A fiberoptic catheter (Post Craniotomy Subdural Pressure Monitoring Kit Model 110-4G, Camino Neurocare, San Diego, CA) was placed in the sub-dural space and connected to a pressure monitor (Camino 420 Series) for measurement of ICP. CPP was calculated by subtracting ICP from MAP. A separate craniotomy was made over the left frontal-parietal border, 1.5 cm lateral to the midline, and 2 cm anterior to the bregma. A multi-port hollow bolt was installed flush with the unbroken surface of the dura.
After the surgical instrumentation was complete, there was a 30- to 60-min stabilization period, then either the series 1 or series 2 protocol was performed.
Series 1 experimental protocol
Hyper/hypo-capnia was induced prior to injury and after injury (and fluid resuscitation). The hypercapnic challenge consisted of changing the inspired gas mixture to FiCO2 = 0.1 and FiO2 = 0.4 until steady-state conditions were reached (usually 10–15 min), after which measurements were taken. Then, a hypocapnic challenge was begun. The ventilator rate was adjusted so that ETCO2 was reduced to 25 mmHg with FiO2 = 0.4. After steady-state hypocapnic conditions, measurements were taken. The ventilator settings were then returned to normal. After an additional 10 to 15 min of ETCO2 = 40 mmHg, post-CO2 challenge data were collected in steady-state conditions. The total duration for the FiCO2 challenge averaged 30 to 45 min.
Fluid percussion method for TBI
A fluid percussion device (manufactured in the Department of Bioengineering, Medical College of Virginia) was attached to a multi-port, hollow bolt in the skull. A 4-kg hammer was suspended on a frictionless bearing at the end of a 90-cm metal arm. The hammer was dropped at 90° from horizontal and struck a plunger on a water-filled Plexiglas cylinder to transmit a force of 6 to 8 ATM directly to the dura. The ICP probe was removed immediately prior to TBI to prevent lacerating the tissue, and was then re-calibrated and re-inserted as soon as possible after TBI, usually within 2 min.
Hemorrhage and resuscitation
Within 2 min of the TBI, a stepwise arterial hemorrhage was begun by removing 30% of the estimated blood volume through the arterial catheter (based on 0.67 × kg body wt) in increments of 15%, wait 15 min, withdraw 10% more, wait 15 min, withdraw 5% more (average total time for hemorrhage = 52 ± 3 min). The stepwise arterial hemorrhage is a laboratory protocol developed at Letterman Army Trauma Institute (C.E. Wade, personal communication). Shed blood was stored in sterile plastic bags with integrally attached transfer containers (Miles, Inc., Cutter Biological, Elkhart, IN, anticoagulant, citrate phosphate dextrose adenine solution). One hour after hemorrhage (120 min post-TBI), resuscitation consisted of either shed blood (n = 7), a volume of previously frozen plasma equivalent to shed blood (n = 6), or bolus infusion of saline equivalent to 3× shed blood volume over 30 min (n = 13). Thereafter, supplemental 250 to 500 mL of saline boluses were administered as needed to all animals whenever SAP <100 mmHg, MAP <70 mmHg, or HR >110 beats/min. One additional control group received saline resuscitation after shock without TBI (n = 4). Fresh frozen plasma was prepared by centrifuging donor blood bags at 4°C, 5,000 g for 5 min (Sorvall RC-3, Newton, CT). The plasma was separated from the erythrocytes using a Fenwal plasma extractor (Morton Grove, IL), and was then frozen at −20°C within 6 h of collection.
The scalp wound was closed over the two craniotomies. The external jugular venous and sagittal sinus catheters were tunneled subcutaneously and left indwelling. Cefazolin (1 g) was administered for prophylaxis. Those that were conscious and capable of spontaneous breathing were returned to their cages. During a 3-day recovery period, intravenous fluids (5% dextrose or lactated Ringers), diazepam (5–20 mg, intravenously), or fentanyl (100–200 μg, intravenously) was administered for dehydration, distress, or pain, as recommended by the veterinarians during their regular cage checks.
During morning and evening cage checks, motor movements (1–8 points), eyelid reflexes (1–3 points), and respiratory activity (1–3 points) were evaluated and graded using a coma scale that reflects normal (14 points) to moribund (1 point) behavior (17).
Series 2 experimental protocol
During hypercapnia, FiCO2 was reduced to 0.05 (instead of 0.1), and the duration of the hyper/hypo-capnia was reduced to no more than 10 min (instead of as long as necessary to reach steady state).
Fluid percussion method for TBI
The magnitude of the cortical impact was reduced to 4 to 5 ATM (instead of 6–8 ATM).
Hemorrhage and resuscitation
The stepwise hemorrhage was increased to 35% of estimated blood volume (instead of 30%), the duration of the blood withdrawal period was shortened to 30 min (instead of 50 min), and the total period of post TBI hypotension was shortened to 60 min (instead of 120 min). In addition, the resuscitation protocol was slightly changed. An initial fluid bolus of either 500 mL of 10% DCLHb or 500 mL of 0.9% saline was administered as rapidly as possible to simulate “pre-hospital resuscitation.” After 15 min, both groups received 3× shed blood as saline followed by supplements as needed whenever SAP <100 mmHg, MAP <70 mmHg, or HR >110 beats/min. The duration of the post-shock observation period was increased to 180 min (instead of 120 min).
After the last data point was collected and the FiCO2 challenge was completed, one-half of the shed blood was transfused (no blood was administered in Series 1). Naloxone (1 mL, 0.4 mg/mL; Endo Laboratories, Wilmington, DE) was administered intravenously to reverse the narcotic anesthesia (naloxone was not administered in Series 1). During morning and evening cage checks, all animals received 1-L infusions of either lactated Ringers or 5% dextrose (fluids were only administered as needed in Series 1).
Day 4 protocol
After a 72-h recovery, the animals were induced, anesthetized, and ventilated as in day 1, and catheters were placed in the left femoral artery and internal jugular vein for systemic pressure measurement and blood sampling. Lactated Ringers was administered to balance any weight change (e.g., loss of 1.5 kg body wt = 1.5 L of fluid), then day 4 baseline and CO2 challenge measurements were taken. An overdose of anesthetic was administered for euthanasia.
Collection of specimens for histological analysis
Both common carotid arteries were retrograde perfused with 1 L of saline followed by 1 L of 10% buffered formalin. Both jugular veins were severed to avoid brain distention from formalin loading. The brain was allowed to fix in situ for an additional 1 h. The brain was removed, examined grossly, then stored in 10% buffered formalin. The brains were examined in a systematic fashion macroscopically using 5-mm coronal sections for the hemispheres and transverse sections for the brain stem and cerebellum. Six to seven standard tissue blocks at various levels of the neuraxis were embedded in paraffin, and representative 6- to 8-μm sections were stained with hematoxylin and eosin. These sections were evaluated by a neuropathologist (J.B.S.) who was unaware of the resuscitation protocol that was used. Hemorrhagic contusions were noted as present or absent, as was necrosis of the cerebellum in the midline. A semi-quantitative score was assigned as follows: subarachnoid hemorrhage was assessed as absent (0), trace (1), present (2), or severe (3) over the convexities, dorsum, or ventrum of brainstem. After microscopic examination, diffuse axonal injury was judged as absent (0), mild (1), moderate (2), or severe (3) based on a subjective interpretation of numbers of axonal balls and the number of slides in from a given animal in which they could be demonstrated. The presence of hypoxic-ischemic changes in neurons was also noted as present or absent.
Animals were studied in pairs at side-by-side operating stations. Data are expressed as means ± SE. Within-group and between-group comparisons were made with commercially available statistical software (Statview 4.5, Abacus Concepts, Berkeley, CA) using repeated measures analysis of variance a priori and Fisher's protected least significant difference test post hoc. Significance was assessed at the 95% confidence interval.
Series 1 qualitative observations
A total of 34 animals was instrumented. In four animals, the skull was damaged during the craniotomies, so the TBI could not be delivered. These animals were subjected to sham TBI, hemorrhage, shock, and saline resuscitation, and were euthanized at the end of day 1. In the remaining 30 animals, there were four deaths after TBI + hemorrhagic hypotension, prior to resuscitation. Those animals were excluded from further analysis. Of the 26 who received TBI + shock and who survived to the end of resuscitation, six could not be weaned from the ventilator and were euthanized on day 1 (n = 5 saline, n = 1 plasma, and n = 0 blood). The chi-squared P value was 0.0581 for blood vs. saline. Of the 20 animals returned to the cage for recovery, there were six additional deaths prior to day 4 (n = 2 saline; n = 2 plasma; and n = 2 blood). By day 4, there were 6/13, 3/5, and 5/7 survivors in the saline, plasma, and blood groups, respectively.
By day 4, all 14 survivors had normal eyelid (coma score = 3/3) and respiratory reflexes (coma score = 3/3). However, motor impairment was fairly significant, ranging from drowsiness and spontaneous purposeful movements (coma motor score = 7/8) to total lethargy with sternal recumbency (coma motor score = 6/8) to deep coma with spontaneous pedaling movement and withdrawal to pinch only (coma motor score = 4/8). The motor scores averaged 5 ± 1, 6 ± 1, and 5 ± 1, and the total coma scores averaged 11 ± 1, 11 ± 1, and 12 ± 1 in the saline, plasma, and blood groups, respectively. On the basis of these observations, we decided that the relative lack of neurologic critical care during the recovery might have confounded interpretation of the data. For this reason, only the acute (non-survivor) data from day 1 are presented in Figures 1 through 4, and several changes were made for the Series 2 experiments.
Series 1 quantitative observations
Table 1 shows that the groups were similar in terms of weight and hemorrhage volume. The total leukocyte and platelet counts were also similar before and after injury (data not shown). The saline group required significantly more total fluid than did the plasma (P = 0.0003) or blood (P < 0.0001) groups to stabilize SAP and HR during resuscitation.
Figure 1 shows time-related changes in SAP, HR, arterial lactate, and hematocrit in the saline and blood groups. In all groups, SAP rapidly corrected to the target value (>100 mmHg) with fluid resuscitation within the first 30 min after the shock period (150 min after TBI). In the saline group, MAP stabilized at 84 ± 6 mmHg (data not shown). In the saline group over the next 45 min both SAP and MAP remained unchanged, but then decreased 5 to 10 mmHg with the hypercapnic (240 min after TBI) and hypocapnic challenges (250 min after TBI). In the blood and plasma groups, SAP and MAP were both significantly higher (all P < 0.05) than the values in the saline group for the entire 120-min post-resuscitation period (150–270 min post-TBI). Similarly, in all groups, HR rapidly corrected to the target value (<110 beats/min) with fluid resuscitation within the first 30 min after the shock period. While there were no differences at individual time points, the average value during the post resuscitation period was significantly higher for saline vs. blood (P = 0.0002).
The bottom right panel of Figure 1 shows that hematocrit was significantly higher at each individual time point after resuscitation with blood vs. saline (all P < 0.05). The changes were similar within the other two non-blood groups. Despite the hemodilution, plasma osmolarity was not significantly different between groups at any point (data not shown). Other plasma electrolytes (Na+, K+, and Ca+2) remained essentially constant over the whole protocol except for transient changes at the time of fluid administration (data not shown). The bottom left panel of Figure 1 shows that lactate rose about 2.5× during the shock period, but then progressively cleared in all groups with resuscitation. However, lactate clearance was delayed in the plasma group.
Figure 2 (top left panel) shows that pulmonary artery pressure (PAP) transiently increased during the CO2 challenges and during the post-TBI resuscitation, but there were no between group differences. The top right panel shows with blood or plasma resuscitation, SvO2 fully corrected to the respective baseline. However, within the saline group, SvO2 remained significantly less than its baseline at each post resuscitation time point (all P < 0.05). Despite these within group differences vs. respective baselines, there were no significant between group differences at individual time points until the end of the observation period. The average value during the post-resuscitation period was significantly lower for saline vs. blood (P = 0.0029).
With all three fluids, CI was restored to supranormal levels (>120% baseline) and was similar between groups. However, because of the significantly higher HR, stroke volume was significantly lower at each post-resuscitation time point for blood vs. saline (all P < 0.05; data not shown). CVP, as an index of filling pressure, was at or above baseline after resuscitation in all groups.
Figure 3 shows that after resuscitation, ScvO2 was 5% to 10% higher with blood vs. saline at each individual time point. However, these apparent differences were only significant at the CO2 challenge (P = 0.0022 vs. blood). While there were no differences at individual time points, the average value during the post-resuscitation period was significantly higher for blood vs. saline (P = 0.02). Prior to injury, PaCO2 changes at the arrows produced ScvO2 increases to >80%. After injury, with saline, CO2-evoked an increase ScvO2 to only 60%. With blood, the CO2-evoked increase in ScvO2 was similar before and after injury.
Figure 4 shows the time-related changes in ICP and CPP. The top panel shows that upon resuscitation, ICP rapidly increased in all groups (all P < 0.05) and remained near 20 mmHg in the saline group. For comparison, ICP remained at 15 mmHg in the blood and plasma groups and was lowest in the group that received sham TBI and saline resuscitation. The stable post-resuscitation value was significantly different with blood vs. saline (P = 0.0023), but the apparent differences at each individual post-resuscitation time point did not reach the 5% confidence level until the time of the post injury CO2 challenge. Prior to injury, PaCO2 changes at the arrows produced small changes in ICP and essentially no change in CPP. With saline, there was no CO2-evoked ICP increase. With blood, the CO2-evoked ICP increase was potentiated relative to baseline.
The lower panel of Figure 4 shows that CPP was significantly higher with blood vs. saline immediately upon resuscitation and at each time point thereafter (all P < 0.05). The changes in the plasma group were intermediate. By 240 min after TBI in the saline vs. blood or plasma group, CPP was lower at PaCO2 = 25 (hypocapnia), 37 (normocapnia), or 71 (hypercapnia) mmHg (P = 0.0021, 0.0005, and 0.0004 vs. blood, respectively).
Series 2 qualitative observations
In the population of 12 animals subjected to a less severe TBI and shorter hypotensive period, one saline animal was excluded due to technical (i.e., ventilator) failure. All others were successfully weaned from the ventilator and were returned to the cage for recovery.
At approximately 12-h intervals, neurologic exams were conducted and total coma scores averaged 10, 12, 12, 13, 13, and 14. There were no clinical differences in the two pigs recovering in side by side cages. By 72 h, all had normal eyelid (coma score = 3/3), and respiratory reflexes (coma score = 3/3), with some motor impairment ranging from mild drowsiness with spontaneous purposeful movements (coma motor score = 7/8) to lethargy with sternal recumbency (coma motor score = 6/8). All animals survived until day 4 and there was no difference between treatments.
Series 2 quantitative observations
Table 2 shows that blood loss, maintenance fluids, and resuscitation fluids were similar between groups. The total leukocyte and platelet counts were also similar before and after injury (data not shown). Compared to DCLHb, the amount of supplemental fluid required to maintain SAP and HR was significantly higher with saline (P = 0.0166). As a result, the total amount of fluid was significantly higher with saline (P = 0.0314).
Figure 5 shows that SAP was significantly higher immediately upon infusion of DCLHb vs. saline at each time point thereafter (all P < 0.05). The changes in MAP were comparable and also significant at each time point (data not shown). While there were no differences at individual time points, the average HR during the post-resuscitation period was significantly higher for saline vs. DCLHb (P = 0.01). The lower panel of Figure 5 shows that lactate increased to about 2 mmol/L by the end of the shock period. During resuscitation with saline, lactate rapidly corrected to pre-injury baseline. However with DCLHb, lactate continued to rise and the washout was delayed (all P < 0.05). All these values had recovered to their respective pre-injury baseline by day 4. Hematocrit changes after shock and resuscitation with DCLHb and saline were virtually identical to those shown in Figure 1 for the saline group, so the data are not shown. By day 4, hematocrit had recovered to 22 ± 1 in both groups, compared to a pre-injury baseline of 27 ± 2.
Figure 6 shows that PAP rapidly rose with DCLHb and remained significantly higher than the corresponding value with saline throughout the observation period (all P < 0.05). CI was transiently depressed with DCLHb vs. saline immediately following resuscitation (P = 0.0122), probably in response to increased afterload. Thereafter, CI stabilized at an average post-resuscitation value that was about 20 mL/min/kg lower with DCLHb and this difference was significant (P = 0.034). Despite the lower HR in the DCLHb group, stroke index (data not shown) was similar between groups even though CVP was significantly higher with DCLHb vs. saline at all time points (P = 0.0146 to < 0.0001) after resuscitation.
SaO2 was constant and >99% at all time points (data not shown), so whole body O2 extraction is reflected by SvO2, which fell with shock (saline P = 0.037, DCLHb P = 0.0142), and was restored to baseline with resuscitation in both groups.
Figure 7 shows that ScvO2 was 5% to 10% lower after resuscitation with DCLHb vs. saline, but this apparent difference did not reach the 5% level of statistical significance at any individual time point. However, the average ScvO2 during the post-resuscitation period was significantly lower with DCLHb vs. saline (P = 0.01). With the CO2 challenge post-TBI, ScvO2 increased to 60% with DCLHb and to 80% with saline (p = NS). According to the manufacturer of DCLHb, this compound can interfere with some instruments that measure oxy-hemoglobin saturation (J.C. Hartman and K.E. Burhop, personal communication). For this reason, cerebral venous PO2 is also shown in Figure 7. The PcvO2 data are qualitatively similar to the ScvO2 data.
Figure 8 shows that ICP was near the pre-injury baseline, during the shock period, but CPP was reduced by about one-half. Following resuscitation, ICP rose in both groups (saline P = 0.0138, DCLHb P < 0.0001) compared to baseline. With DCLHb vs. saline, ICP was significantly higher at several time points (all P < 0.05). After the 72-h recovery period, ICP was similar between groups. Despite the higher ICP, the DCLHb group maintained a normal (pre-injury) CPP at all times post-injury at normocapnia (PaCO2 = 38–44 mmHg). These values were significantly higher than those with saline at several points post-resuscitation (all P < 0.05).
At 240 min post-TBI, cerebrovascular reactivity was tested by increasing FiCO2 to 0.05 (which increased PaCO2 to 58 ± 2 mmHg). In this condition, ICP increased and CPP fell and the values were similar in both groups. With hyperventilation (PaCO2 = 27 ± 1 mmHg), ICP and CPP were restored to the normocapnia values.
Figure 9 shows a typical section from a brain 72 h after resuscitation with saline. The gross appearance showed inferior cerebellar tonsilar necrosis and hemorrhage, and a trace of ventral brain stem subarachnoid hemorrhage. On the cut surface, a small (<1 mm) hemorrhage was found buried in the cortex of the left frontal pole. There was evidence of a catheter tract, and examination of transverse sections of the cerebellum showed midline cerebellar hemorrhagic necrosis. Microscopic examination confirmed the presence of an acute contusion in the left frontal pole, and a zone of subtotal damage on the inferior aspect of the frontal pole that was interpreted as representing a contrecoup lesion. Scattered axonal bulbs were demonstrable in the white matter of the left hemisphere and the thalamus, and rare axonal bulbs were visualized in the raphe of the mid brain. Collections of numerous axonal bulbs were seen in the dorsolateral aspect of the pons near the angle of the fourth ventricle and extending into middle cerebellar peduncle. The presence of midline cerebellar necrosis and related dorsal column necrosis was confirmed. Some large neurons in the midline of the medulla appeared chromatolytic. Some axonal bulbs were identified in one posterior horn and in the corticospinal tract.
The brain illustrated in Figure 10 was obtained from an animal resuscitated with DCLHb. The only gross abnormalities detected on the cut surface were the presence of a catheter tract and questionable necrosis in the midline of the cerebellum, the presence of which was confirmed by microscopic examination. Subarachnoid hemorrhage was present over the right convexity, with a trace on the ventral brainstem and a severe amount over the dorsal brainstem. The right hemisphere showed moderate numbers of axonal bulbs and the presence of fairly widespread numbers of eosinophilic neurons in the deeper layers of the cortex of the gyral crests. Fairly numerous axonal bulbs were identified in the dorsolateral pons near the angle of the fourth ventricle and extending into middle cerebellar peduncle.
Semi-quantitative histopathologic grading of injury according to type, location, and severity is shown in Table 3. All animals displayed structural evidence of brain injury ranging from subarachnoid hemorrhage to ischemic changes to diffuse axonal injury. There were no significant differences between the groups in type or severity. In particular, there was no trend toward reduced ischemic injury with DCLHb vs. saline.
Figures 1 and 2 show resuscitation with blood or saline to the same SAP and HR endpoints in the Series 1 protocol corrected most systemic variables and markers of volume status, including CI, lactate, and cardiac filling pressures. However, there were 0/7 vs. 5/13 deaths within 5 h (P = 0.058) and CPP, ScvO2, and CO2 evoked responses were all higher with blood vs. saline (Figs. 3 and 4).
According to the Monro-Kelley Doctrine (2,9), rapid and reversible ICP increases and decreases reflect vasodilation and vasoconstriction in the closed cranial vault, and the magnitude of the ICP change indicates cerebral compliance. At constant SaO2, rapid and reversible increases or decreases in ScvO2 that accompany changes in PaCO2, reflect increases or decreases in the ratio of global cerebral blood flow to metabolism (3,9–11). For example, at a relatively constant cerebral metabolism, a large CO2-evoked ScvO2 increase with a small ICP increase suggests a brisk vasodilation with high cerebral compliance. That pattern was observed prior to injury; a biphasic change in PaCO2 from a baseline value of 40 to 80 to 25 mmHg produced corresponding changes in ScvO2 from a baseline value of 45% to 85% to 35% (Fig. 3), as well as corresponding changes in ICP from a baseline value of 4 to 8 to 4 mmHg (Fig. 4). After injury and resuscitation with saline, the ScvO2 excursions in response to the same PaCO2 challenge were blunted compared to those in the blood group. With saline, ICP averaged about 20 mmHg post-injury (compared to a baseline value <5 mmHg) and CPP was 50 to 60 mmHg. With blood, ICP and CPP were 15 and 80 to 85 mmHg post-injury. In both groups, the CO2-evoked changes in CPP were greater post-injury. Altogether, these data are consistent with reduced cerebral compliance in both groups and reduced cerebrovascular CO2 reactivity with saline vs. blood. That finding may have some practical implications because many TBI patients are resuscitated to SAP and HR endpoints and would not have CPP monitors, jugular bulb catheters, or Swan-Ganz catheters in place in emergency or field situations. But the main point is that blood was more effective than saline for resuscitation after severe TBI (6–8 ATM) + 120 min of hypotension.
Figures 5 through 8 show that after a less severe TBI (4–5 ATM) + 60 min hypotension, CPP was higher with DCLHb vs. saline. There were no other obvious benefits with this blood substitute, except that less fluid was required to stabilize and maintain SAP and HR (Table 2). The higher CPP and reduced fluid requirement might be helpful in some field situations or during emergency medical transport where unlimited fluids, sophisticated monitoring, or pressors are not readily available. On the other hand, less volume combined with the pressor action of DCLHb (12–15) probably contributed to worsened indices of systemic resuscitation. Figures 5 and 6 show that with DCLHb vs. saline, CI was lower and lactate was higher. The reduced CI could be attributed, in part, to an increased right and left ventricular afterload, but cannot be attributed to reduced preload because CVP was markedly elevated throughout the post-resuscitation period. There was a trend for cerebral vasoconstriction because ScvO2 averaged 5% lower with DCLHb vs. saline (Fig. 7). There was no neuropathologic evidence that DCLHb improved ischemia-related structural damage by 72 h relative to saline (Figs. 9 and 10;Table 3). Thus, there was almost no evidence that resuscitation with DCLHb was better than saline, which is surprising because blood was superior to saline after a more severe injury.
Critique of experimental design
The data interpretation must be considered within the framework of at least six major limitations in the experimental design. First, every animal model of TBI shows important differences when compared human TBI. Second, these data were collected under conditions of mechanical ventilation, FiO2 = 0.4, and fentanyl anesthesia. Third, even if the coupling between cerebral blood flow and metabolism is a vital outcome determinant after TBI, neither cerebral blood flow nor metabolism was directly measured. Fourth, catheters in the carotid artery and jugular veins on the same side could have biased outcome. Fifth, the lack of critical care during the post-trauma recovery could have masked any positive or negative effect of the fixed volume or unlimited volume resuscitation strategies. Sixth, hemodilution is unavoidable when using crystalloid resuscitation and the consequences of hemodilution are debatable. Each of these limitations will now be briefly discussed.
The first point is that all animal models have some weaknesses. The pig brain is much smaller than the human brain and the dural investment is thicker (relatively) and tightly applied, so the pathological effects of inertial loading and rotational forces are not the same. The shape and the thickness of the skull relative to brain are also quite different, so the transmission of force through the skull, onto the brain, and reflections of force inside the skull are likely to be different than in humans. Finally, in any animal model that is in current use, the distribution of axonal swellings and axonal bulbs, the hallmark of diffuse TBI, tends to be different from that in a human TBI patient (4,18). A unique approach to address the problems of force transmission through the pig skull uses a freeze-thaw (cryogenic) method for producing TBI. This method has its own set of limitations, but it produces histologic changes that resemble certain aspects of human brain injury (19,20). In this present study we observed histological changes similar to those in human TBI (Figs. 9 and 10;Table 3), but the neuropathologic evidence was only semi-quantitative.
With regard to the second point, all these animals were mechanically ventilated with FiO2 = 0.4 to eliminate the respiratory depression associated with TBI (21) and were anesthetized for ethical reasons. There is no question that anesthesia alone can influence hemodynamic changes, as well as outcome after TBI (22), but all animals received the same drugs, so it is unlikely that anesthesia altered outcome.
With regard to the third limitation, neither cerebral blood flow nor cerebral metabolism was directly measured. We assumed that the global relationship between blood flow and metabolism was proportional to alterations in the ScvO2 at constant SaO2 (3). This assumption has been validated in anesthetized pigs by comparing local and global blood flow (using the microsphere method) with global estimates derived from ScvO2 (21). Unlike a human, the pig lacks a jugular bulb, but the sagittal sinus is an analogous anatomical structure that collects cerebral venous outflow. In humans, changes in jugular bulb O2 saturation have been used for bedside assessment of CO2-reactivity and autoregulation in patients during the acute phase of severe head injury (3,10,11,24,25).
The fourth limitation of the experimental design was that catheters were placed in the right carotid artery and jugular vein, which could have interfered with cerebral perfusion. However, unlike humans, pigs have an extensive collateral flow through the vertebral arteries and direct measurement of regional cerebral blood flow has demonstrated no obvious effect with the total unilateral occlusion of one carotid artery and one jugular vein (26).
Considering the fifth limitation, there is no question that 24-h critical care would have influenced outcome. Nevertheless, in the Series 1 experiments, despite a substantial TBI (6–8 ATM) followed by 120 min of hemorrhagic hypotension which caused four deaths out of 30, 20/26 pigs were extubated after only 5 h. Despite the lack of critical care during recovery, there were only six additional deaths in the cage prior to day 4. In the Series 2 experiments, there were no deaths within 72 h and all neurologic indices were improving. Thus, even with no critical care, the morbidity and mortality associated with this severe experimental TBI is hardly different than what might be expected in the same time period for a human that suffered a similar insult.
Finally, with regard to the sixth limitation, hemodilution was unavoidable. The effect of hemodilution has been examined in both occlusive and traumatic models of brain injury and the effects vary. After cerebral artery occlusion in rats, hemodilution reduces damage and edema if normotension is maintained (27,28). If hemodilution is achieved with DCLHb, instead of albumin, focal ischemic injury is further reduced, especially at doses that induce hypertension (27,28). However, after traumatic injury, the benefits of hemodilution are equivocal. Prior to cryogenic injury in rabbits, hemodilution increased blood flow in all brain regions, but after injury, the response was attenuated in the ipsilateral hemisphere (29). In cats, neither hemorrhage alone nor fluid percussion TBI alone produced significant changes in cerebral blood flow, cerebral oxygen delivery, or EEG. When the two insults were combined, there were significant decreases in all three, but the changes were similar in those resuscitated with shed blood or in those hemodiluted with hetastarch (30). Therefore, the potential benefits of hemodilution may depend on several factors such as MAP, the nature of the resuscitation fluid, as well as the type of brain injury. However, certain basic physiologic principles must apply: First, with hemodilution, CI must be increased to maintain an equivalent systemic O2 delivery. If CI is not increased, then O2 extraction must be increased to maintain an equivalent systemic O2 consumption. Second, with hemodilution, systemic vascular resistance is usually reduced because hematocrit influences blood viscosity and viscosity is a component of resistance. If MAP remains constant, and if resistance is decreased, then CI must be increased. These present results clearly showed that CI was the same, but SAP (and MAP) was 20 mmHg lower with saline vs. blood or plasma (Fig. 1). At constant SaO2, SvO2 was lower with saline vs. plasma or blood (Fig. 2). Similarly, ScvO2 was lower with saline vs. plasma or blood at normocapnia, hypercapnia, or hypocapnia (Fig. 3). Thus, it is theoretically possible that the reduced CPP and lower ScvO2 were an inevitable consequence of hemodilution, and not necessarily a bad outcome with saline vs. blood in Series 1, but that could not explain 5/13 deaths for saline vs. 0/7 deaths for blood (P = 0.0581). In addition, that explanation could not account for the reduced ScvO2 in the DCLHb group in the Series 2 experiments (Fig. 7) because both groups had similar hematocrits.
Comparison to previous work
This is the first study in which DCLHb has been tested after a fluid percussion form of TBI + hemorrhage, but its properties have been studied in several other injury models.
Plasma volume expansion and hemodynamic responses in conscious hemorrhaged and normovolemic splenectomized sheep were compared after DCLHb, human albumin, or lactated Ringers (15). All three solutions expanded blood volume and increased MAP and CI after hemorrhage, but DCLHb was associated with greater plasma volume expansions, higher MAP, and lower CI for at least 6 h compared to the other two. The authors speculated that volume expansion with DCLHb was due to greater mobilization of endogenous interstitial protein or to reduced transcapillary loss (15). Furthermore, DCLHb-mediated vasoconstriction could lower blood-to-tissue transport of fluid and protein, which could explain why DCLHb expanded vascular volume out of proportion to its colloid osmotic pressure (15). However, after chest trauma, resuscitation with DCLHb caused pulmonary vasoconstriction, greater lesion size, reduced compliance, and increased lung wet weight (31), which led those authors to conclude that blood-to-tissue transport of fluid and protein was aggravated by DCLHb. We recently evaluated DCLHb and three other HBOC formulations after chest trauma and observed vasoconstriction, but no evidence that blood-to-tissue transport of fluid and protein was aggravated (32). Thus, further study is necessary to reconcile the effects of various HBOCs on plasma volume expansion.
Most authors have interpreted the effects of DCLHb after various types of head injury as beneficial (19,20,27,28,33,34). One series of studies were performed after cryogenic TBI combined with constant pressure hemorrhage (19,20). Resuscitation consisted of 4 mL/kg of either lactated Ringer's or DCLHb, followed by supplemental Ringer's to maintain MAP at baseline (19). Even though fluid requirements were 20% lower, DCLHb increased MAP and CPP for at least 24 h. Neither ICP nor cerebral O2 delivery was significantly altered (19). The same investigators later showed that hemodilution with DCLHb increased MAP, increased CPP, and reduced ICP with significantly less total fluid (20), also consistent with this present study. Neither regional cerebral blood flow, cerebral O2 delivery, nor lesion volume differed between treatments. These results basically agree with our Figures 5 through 8.
In rodents with diffuse brain injury produced by a weight drop (33) combined with hemorrhagic hypotension, resuscitation with DCLHb improved CPP and cerebral blood flow and lowered ICP, relative to saline. The CPP findings also concur with those in Figure 8.
Finally in a systematic series of studies in rats (34), the actions of DCLHb have been observed after subarachnoid hemorrhage. Those data show that hemodilution with DCLHb improves CPP and decreases neuronal death, relative to blood or saline. They found no evidence that DCLHb exacerbated cerebral ischemia, which is consistent with our Table 3 and Figures 9 and 10.
After severe TBI, increased CPP tends to improve cerebral oxygenation, as reflected by improved ScvO2, after severe TBI (35,36). In preliminary observations, we have reported that ScvO2 and CO2 reactivity are improved when CPP is increased with a pressor (phenylephrine) infused along with standard saline resuscitation (37). We have no immediate explanation why ScvO2 and CO2 reactivity were not improved, relative to saline, when CPP was increased with a pressor dose of DCLHb in this present study. Perhaps the effect is related to mechanism of the pressor action. Whereas phenylephrine acts on adrenergic receptors, it is generally assumed that nitric oxide scavenging probably explains the pressor action of DCLHb, as well as almost every almost other HBOC in current human trials (12,13). There is some evidence that endothelin is also released and that altered viscosity of HBOC solutions reduces sheer stress on the endothelium, which reduces basal nitric oxide production, which evokes vasoconstriction, and decreased functional capillary density (12,13). This combination of events may be lethal because decreased nitric oxide availability can also increase the intrinsic tissue oxygen consumption (13). Whatever the explanation for the adverse treatment effect of DCLHb in the human trauma trial (16), new generation HBOCs with fewer side effects are already in the development and testing stages in various animal models (14).
Summary and conclusions
The first series of experiments tested the hypothesis that CPP, ScvO2, and cerebrovascular CO2 reactivity would be comparable after resuscitation with saline vs. blood to the same SAP and HR. The data clearly showed that CPP was lower, cerebrovascular CO2 reactivity was less, and mortality trended higher. The second series of experiments tested the hypothesis that saline was inferior to DCLHb, using the same basic endpoints. Those results showed that CPP was lower, but quite unexpectedly, there was no other obvious benefit with DCLHb vs. saline resuscitation. For example, afterload was increased, CI was depressed even though filling pressure was increased, and lactate clearance was delayed. Therefore, in these experimental conditions, whole blood was more effective than saline resuscitation, whereas DCLHb was not more, and possibly even less, effective than saline. We speculate that further investigation in similar experimental models might provide some plausible explanations why DCLHb unexpectedly increased patient mortality in a recent multicenter trial for the treatment of severe traumatic shock.
We are grateful to J. Craig Hartman, Ph.D., of Baxter Hemoglobin Therapeutics (Boulder, CO) for providing the DCLHb; to Kenneth Weber of NeuroCare Group (Pleasant Prairie, WI) for providing the Camino Monitor and catheters; to Aline Pendergrast-Fuess of Baxter Healthcare Corporation (Irvine, CA) for providing cardiac output computers and Swan-Ganz catheters; to Jill Lomauro of Arrow International, Inc. (Reading, PA) for providing catheter introducers; and to Terry Shirey, Ph.D. of Nova Biomedical (Waltham, MA) for providing the Stat Ultra Blood Analyzer.
1. Chesnut RM, Marshall LF, Klauber MR, Blunt BA, Baldwin N, Eisenberg HM, Jane JA, Marmarou A, Foulkes MA: The role of secondary brain injury in determining outcome from severe head injury. J Trauma 34( 2):216–222, 1993.
2. Lang EW, Chesnut RM: Intracranial pressure: monitoring and management. Neurosurg Clin N Am 5( 4):573–605, 1994.
3. Robertson CS, Cormio M: Cerebral metabolic management. New Horiz 3( 3):410–422, 1995.
4. Maxwell WL, Povlishock JT, Graham DL: A mechanistic analysis of non-disruptive axonal injury: a review. J Neurotrauma 14( 7):419–440, 1997
5. Pope A, French G, Longnecker DE:Fluid Resuscitation: State of the Science for Treating Combat Casualties and Civilian Injuries.
Report from the Committee on Fluid Resuscitation for Combat Casualties, Division of Health Science Policy, Institute of Medicine, Washington, DC: National Academy Press, 1999.
6. Glass TF, Fabian MJ, Schweitzer JB, Weinberg JA, Proctor KG: Secondary neurologic injury resulting from non-hypotensive hemorrhage combined with mild traumatic brain injury. J Neurotrauma 16( 9):771–782, 1999.
7. Glass TF, Fabian MJ, Schweitzer JB, Weinberg JA, Proctor KG: The impact of hypercarbia on the evolution of brain injury in the setting of traumatic brain injury and systemic hemorrhage. J Neurotrauma 18( 1):57–71, 2001.
8. American College of Surgeons:Advanced Trauma Life Support Course for Physicians,
ed., Chicago: American College of Surgeons, 1997.
9. Bullock R, Chesnut RM, Clifton G, Ghajar J, Marion DW, Narayan RK, Newell DW, Pitts LH, Rosner MJ, Wilberger JE:Guidelines for the Management of Severe Head Injury.
A joint initiative of the Brain Trauma Foundation, The American Association of Neurological Surgeons, and the Joint Section of Neurotrauma and Critical Care, 1995
10. Sahuquillo J, Poca MA, Ausina A, Baguena M, Gracia RM, Rubio E: Arterio-jugular differences of oxygen (AVDO2
) for bedside assessment of CO2
-reactivity and autoregulation in the acute phase of severe head injury. Acta Neurochir (Wien) 138( 4):435–444, 1996.
11. Sahuquillo J, Munar F, Baguena M, Poca MA, Pedraza S, Rodriguez-Baeza A: Evaluation of cerebrovascular CO2
-reactivity and autoregulation in patients with post-traumatic diffuse brain swelling (diffuse injury III). Acta Neurochir Suppl (Wien) 71:233–236, 1998.
12. Hess JR: Blood substitutes. Semin Hematol 33( 4):369–378, 1996.
13. Intaglietta M: Microcirculatory basis for the design of artificial blood. Microcirculation 6( 4):247–258, 1999.
14. Burhop KE, Doyle M, Lemon D: Vasoactive properties of hemoglobin solutions: facts, fiction, and myths. Shock 12(Suppl 1):10, 1999.
15. Fischer SR, Burnet M, Traber DL, Prough DS, Kramer GC: Plasma volume expansion with solutions of hemoglobin, albumin, and Ringer lactate in sheep. Am J Physiol 276(Pt 2):H2194–H2203, 1999.
16. Sloan EP, Koenigsberg M, Gens D, Cipolle M, Runge J, Mallory MN, Rodman Jr: G Diaspirin cross-linked hemoglobin
(DCLHb) in the treatment of severe traumatic hemorrhagic shock: a randomized controlled efficacy trial. J Am Med Assoc 282( 19):1857–1864, 1999.
17. Hilton Jr, DL Einhaus SL, Meric 3d, AL White RP, Schweitzer JB, Park MR, Robertson JT: Early assessment of neurologic deficits in the fluid percussion model of brain injury. J Neurotrauma 10( 2):121–133, 1993.
18. Ross DT, Meaney DF, Sabol MK, Smith DH, Gennarelli TA: Distribution of forebrain diffuse axonal injury following inertial closed head injury in miniature swine
. Exp Neurol 126( 2):291–299, 1994.
19. Chappell JE, McBride WJ, Shackford SR: Diaspirin cross-linked hemoglobin
resuscitation improves cerebral perfusion after head injury and shock. J Trauma 41( 5):781–788, 1996.
20. Chappell JE, Shackford SR, McBride WJ: Effect of hemodilution with diaspirin cross-linked hemoglobin
on intracranial pressure, cerebral perfusion pressure
, and fluid requirements after head injury and shock. J Neurosurgery 86( 1):131–138, 1997.
21. Zink BJ, Sheinberg MA, Wang X, Mertz M, Stern SA, Betz AL: Acute ethanol intoxication in a model of traumatic brain injury with hemorrhagic shock: effects on early physiological response. J Neurosurg 89( 6):983–990, 1998.
22. McGrath BJ, Matjasko MJ: Anesthesia and head trauma. New Horizons 3( 3):523–533, 1995.
Prielipp RC, Coursin DB: Sedative and neuromuscular blocking drug use in critically ill patients with head injuries. New Horizons 3( 3):456–468, 1995.
24. Rieger A, Menzel M, Rainov NG, Sanchin L, Rot S, Furka I, Gorombey Z, Burkert W: Continuous monitoring of the partial pressure of oxygen in cerebral venous blood. Neurosurgery 41( 2):462–468, 1997.
25. Robertson CS, Gopinath SP, Goodman JC, Contant CF, Valadka AB, Narayan RK: SjvO2 monitoring in head-injured patients. J Neurotrauma 12( 5):891–896, 1995.
26. Leffler CW, Busija DW, Beasley DG, Fletcher AM, Green RS: Effects of indomethacin on cardiac output distribution in normal and asphyxiated piglets. Prostaglandins 31( 2):183–190, 1986.
27. Cole DJ, Schell RM, Drummond JC, Reynolds L: Focal cerebral ischemia in rats: effect of hypervolemic hemodilution with diaspirin cross-linked hemoglobin
versus albumin on brain injury and edema. Anesthesiology 78( 2):335–342, 1993.
28. Cole DJ, Schell RM, Drummond JC, Pryzbelski RJ, Marcantonio S: Focal cerebral ischemia in rats: effect of hemodilution with α-α cross-linked hemoglobin on brain injury and edema. Can J Neurol Sci 20( 1):30–36, 1993.
29. Todd MM, Wu B, Warner DS: The hemispheric cerebrovascular response to hemodilution is attenuated by a focal cryogenic brain injury. J Neurotrauma 11( 2):149–160, 1994.
30. DeWitt DS, Prough DS, Taylor CL, Whitley JM: Reduced cerebral blood flow, oxygen delivery, and electroencephalographic activity after traumatic brain injury and mild hemorrhage in cats. J Neurosurg 76( 5):812–821, 1992.
31. Cohn SM, Zieg PM, Rosenfield AT, Fisher BT: Resuscitation of pulmonary contusion: effects of a red cell substitute. Crit Care Med 25( 3):484–491, 1997.
32. Maxwell RA, Gibson JB, Fabian TC, Proctor KG: Resuscitation of severe chest trauma with four different hemoglobin-based oxygen carrying (HBOC) solutions. J Trauma 49( 2):200–211, 2000.
33. Piper IR, Garrioch MA, Souter MJ, Andrews PJ, Thomson D: Effects of diaspirin cross-linked haemoglobin on post-traumatic cerebral perfusion pressure
and blood flow in a rodent model of diffuse brain injury. Br J Anaesth 80( 5):639–643, 1998.
34. Cole DJ, Reynolds LW, Nary JC, Drummond JC, Patel PM, Jacobsen WK: Subarachnoid hemorrhage in rats: effect of singular or sustained hemodilution with αα diaspirin cross-linked hemoglobin
on cerebral hypoperfusion. Crit Care Med 27( 5):972–977, 1999.
35. Cruz J: The first decade of continuous monitoring of jugular bulb oxyhemoglobin saturation: management strategies and clinical outcome. Crit Care Med 26( 2):344–351, 1998.
36. Robertson CS, Valadka AB, Hannay HJ, Contant CF, Gopinath SP, Cormio M, Uzura M, Grossman RG: Prevention of secondary ischemic insults after severe head injury. Crit Care Med 27( 10):2086–2095, 1999.
37. Malhotra AK, Schweitzer JB, Fabian TC, Proctor KG: Cerebral perfusion pressure
(CPP)-directed therapy after traumatic brain injury (TBI). Shock 13:60, 2000.