Hemorrhagic Shock after Experimental Traumatic Brain Injury in Mice: Effect on Neuronal Death

Abstract

Traumatic brain injury (TBI) from blast injury is often complicated by hemorrhagic shock (HS) in victims of terrorist attacks. Most studies of HS after experimental TBI have focused on intracranial pressure; few have explored the effect of HS on neuronal death after TBI and none have been done in mice. We hypothesized that neuronal death in CA1 hippocampus would be exacerbated by HS after experimental TBI. C57BL6J male mice were anesthetized with isoflurane, mean arterial blood pressure (MAP) was monitored, and controlled cortical impact (CCI) delivered to the left parietal cortex followed by 1) continued anesthesia (CCI only), or either 60 or 90 min of volume controlled HS. Parallel 60 or 90 min HS only groups were also studied. After HS (± CCI), 6% hetastarch was used targeting MAP ≥50 mmHg during a 30 min Pre-Hospital resuscitation phase. Then, shed blood was re-infused and hetastarch given targeting MAP ≥60 mmHg during a 30 min Definitive Care phase. Neurological injury was evaluated at 24h (fluorojade C) or 7 days (CA1 and CA3 hippocampal neuron counts). HS reduced MAP to 30−40 mmHg in all groups, p<0.05 vs CCI only. Ipsilateral CA1 neuron counts in the 90 min CCI+HS group were reduced at 16.5±14.1 vs 30.8±6.8, 32.3±7.6, 30.6±2.2, 28.1±2.2 neurons/100 Im in CCI only, 60 min HS only, 90 min HS only, and 60 min CCI+HS, respectively, all p<0.05. CA3 neuron counts did not differ between groups. Fluorojade C staining confirmed neurodegeneration in CA1 in the 90 min CCI+HS group. Our data suggest a critical time window for exacerbation of neuronal death by HS after CCI and may have implications for blast injury victims in austere environments where definitive management is delayed.

INTRODUCTION

The important role of secondary insults in increasing morbidity and mortality after traumatic brain injury (TBI) is widely recognized, both experimentally and clinically. The combination of TBI and hemorrhagic shock (HS) has taken on special importance in both military and civilian settings as the result of terrorist attacks with improvised explosive devices–which inflict TBI and other extra-cerebral injuries (Gawande, 2004; Gutierrez de Ceballos et al., 2005). The report of Chesnut et al., (1993) reviewing the NIH Traumatic Coma Databank correlated hypotension and hypoxemia with doubled morbidity and mortality after TBI in humans – identifying hypotension as the single most critical parameter. Subsequent work has confirmed the critical detrimental role of secondary insults after TBI in the ICU (Gopinath et al, 1994). The marked deleterious effects of secondary insults have been confirmed in the setting of blast polytrauma (Nelson et al., 2006). Early reports of exacerbation of brain injury by secondary insults in experimental TBI included work in the cats by Nelson et al, (1979), Jenkins et al, (1986) and Barron et al., (1988), and after fluid percussion injury (FPI) in rats by Ishige et al., (1987a; 1987b), where brief periods of hypoxemia were used. Exacerbation of hippocampal neuronal death in the CA3 region by a secondary hypoxic insult was later observed by Clark et al., (1997) using a hypoxic admixture to achieve a PaO₂ of ∼40 mmHg after controlled cortical impact (CCI), and exacerbated hippocampal neuronal death in CA1 was reported by Jenkins et al., (1989) with transient carotid occlusion and hemorrhagic hypotension (producing forebrain ischemia) after FPI in rats. In rodent models of experimental TBI alone, neurons in the CA3 and hilar sectors of the hippocampus generally exhibit the greatest vulnerability (Lowenstein et al., 1992), while pyramidal neurons in the CA1 sector of the hippocampus are classically selectively vulnerable to ischemia or hypoxemia (Kirino, 1982). Both CA1 and CA3 hippocampal neuronal death are often identified on post-mortem examination after fatal TBI in humans (Kotapka et al., 1994). And these patients typically had secondary insults.

The post-TBI milieu is characterized by primary injury, cascades of secondary injury and repair, and less well recognized perturbations of normal homeostatic mechanisms. Considerable evidence supports the existence of marked vulnerability of the traumatically injured brain early after insult. A number of mechanisms are proposed to mediate this enhanced vulnerability including hypoperfusion and reduced oxygen delivery, disturbed autoregulation of cerebral blood flow (CBF), excitotoxicity, and mitochondrial failure, among others (DeWitt et al., 1995). These mechanisms may create an environment in the acutely injured brain that renders it vulnerable to a level of hypotension and anemia from HS that would otherwise be tolerated. In addition, secondary injury mechanisms during resuscitation and reperfusion may further exacerbate the evolution of damage. Given that HS produces maximal vasoconstriction in the splanchnic circulation – while the brain is relatively protected – visceral ischemia with resultant release of pro-inflammatory mediators and/or translocation of intestinal flora may also play a role in amplifying secondary brain damage (Vatner, 1974; Myers et al., 1994). This complex secondary injury cascade in the setting of combined TBI and HS is poorly understood.

Previous experimental models of combined TBI and HS have focused on hemodynamics and the effect of fluid resuscitation on intracranial pressure (ICP) and related intracranial dynamics using large animals (DeWitt et al., 1992a; 1992b; Glass et al., 1999). Few studies have explored the effect of HS on neuronal death mechanisms after TBI in rodents (Matsushita et al., 2001) but none, to our knowledge, has specifically examined the impact of HS on hippocampal neuronal death in selectively vulnerable brain regions. There has also been a paucity of investigation of combined TBI and secondary insults in mice, a species ideal for mechanistic and therapeutic investigation due to ready availability of targeted mutant strains.

We report the characterization of a clinically relevant mouse model of combined TBI and HS, and resuscitation including physiologic monitoring and neuropathologic evaluation. We hypothesized that a level of HS that alone produces no neuronal damage (Carrillo et al., 1998) would increase neuronal death after experimental TBI in the CA1 ischemia-vulnerable region of the hippocampus.

METHODS

Study Groups and Experimental Protocol

The Institutional Animal Care and Use Committee of the University of Pittsburgh School of Medicine approved all experiments. Male C57BL6J mice (Jackson Laboratories, Bar Harbor, ME), 12 to 15 weeks of age and weighing 27 ± 1.8 grams, were housed under controlled environmental conditions and allowed ad libitum food and water until study.

Anesthesia was induced with 4% isoflurane in oxygen and maintained with 1% isoflurane in 2:1 N₂O/O₂ via nose cone. Inguinal cut-down and insertion of central femoral venous and arterial catheters was accomplished under sterile conditions using modified PE-50 tubing. After placement of the mouse in a stereotaxic frame, a 5-mm craniotomy was performed over the left parietotemporal cortex with a dental drill, and the bone flap was removed. A brain temperature micro-probe (Physitemp, Clifton, NJ) was inserted through a left frontal burr hole, and a rectal probe placed to monitor body temperature. Immediately after craniotomy, the inhalational anesthesia was changed to 1% isoflurane and room air for a 10 min equilibration period prior to beginning the injury protocols.

While brain temperature was maintained at 37 ± 0.5 °C, mild-moderate CCI was performed with a pneumatic impactor (Bimba, Monee, IL) as previously reported with modifications (Sinz et al., 1999; Whalen et al., 1999). A 3-mm flat-tip impounder was deployed at a velocity of 5 m/sec and a depth of 1 mm. This injury level for CCI was specifically chosen to produce a contusion but no appreciable loss of hippocampal neurons in any subfield—in the absence of HS, based on prior work with this model in mice by our group (Kochanek et al., 2006; Foley et al., submitted), along with additional pilot studies.

A diagram of the experimental paradigm for TBI, shock and resuscitation is provided in Figure 1. To model a level of HS that was clinically relevant, we performed a series of pilot experiments to determine the amount of hemorrhage volume necessary to reduce the mean arterial blood pressure (MAP) to achieve a stable MAP of ∼ 35−40 mmHg in all groups. Based on prior studies, we did not anticipate that this level of HS alone would produce brain injury (Carillo et al., 1998). Similarly, based on prior studies of the effect of HS on MAP in rodents with or without TBI (Yuan and Wade, 1992), and by a series of pilot experiments, it was determined that in mice subjected to HS alone, a volume of 2.7ml/100g was needed to achieve this target MAP range. In contrast, but as anticipated, after CCI, a smaller volume of 2.0/100g was required, consistent with the well-described enhanced sensitivity to the hypotensive effects of hemorrhage after TBI (Yuan and Wade, 1992; Law et al., 1996). In all mice, HS was induced over 15 min in a decelerating fashion, with 50% of the total volume removed over the first 5 min, 25% over the next 5 min and the final 25% over the last 5 min. Mice remained in unresuscitated HS for an additional 45 or 75 min for a total Shock phase of either 60 or 90 min—to study the effect of HS duration on neuropathological outcome after CCI. After completion of the blood withdrawal, mice transiently auto-resuscitated to a MAP of ∼ 45−55 mmHg, but then rapidly re-equilibrated and maintained MAP in the target range for the remainder of the desired 60−90 min Shock phase. After completion of the HS interval, a 30 min Pre-Hospital phase was initiated and 6% hetastarch (Hextend, Hospira, INC., Lake Forest, IL) was rapidly infused in 0.1 mL aliquots to achieve a MAP P50 mmHg. To simulate arrival at more Definitive Care, mice were then switched from 1% isoflurane in room air to 1% isoflurane in oxygen. For this 30 min interval, shed blood was first rapidly re-infused, and a goal MAP of 60 mm Hg was maintained with additional 6% hetastarch, again administered in 0.1 mL aliquots. At completion of the Definitive Care phase, catheters were removed, anesthesia discontinued and mice recovered in supplemental oxygen for 30 min before being returned to their cages.

Figure 1.

Diagram depicting the overall scheme and time-course of experiment protocol used in this study.

Mice were randomized to one of 5 study groups (n= 10 per group), and underwent procedures or equivalent anesthesia and monitoring as designated: 1) CCI only, 2) 60 min of HS only [60HS only], 3) 90 min of HS only [90HS only], 4) CCI followed immediately by 60 min of HS [60CCI+HS], or 5) CCI followed immediately by 90 min of HS [90CCI+HS]. Mice in the CCI only group underwent CCI without HS but were maintained under identical anesthesia and monitoring to the combined injury groups for a 60 min interval. Mice in the HS only group underwent either 60 or 90 min of HS without craniotomy or CCI, but again, were maintained under identical anesthesia and monitoring as the combined injury groups. Mice in the CCI+HS groups underwent CCI followed by HS of either 60 or 90 min duration as described above.

Monitoring Protocol

MAP was continuously monitored via the femoral artery and recorded at baseline, after CCI, and every 5 min during HS and resuscitation; heart rate was continuously monitored and recorded at baseline and once during each phase. Laboratory evaluation with arterial blood gas determinations, and blood lactate, glucose, hematocrit, sodium, potassium, ionized calcium and ionized magnesium was obtained at baseline, 30 min into the Shock phase, and at the end of the Definitive Care phase.

Histology Protocol

At 24 hours (n= 4 per group) or 7 days (n= 6 per group) after experiments, mice were reanesthetized with 4% isoflurane and killed by ice-cold saline transcardial perfusion followed by 10% buffered formalin phosphate perfusion and fixation of brains with subsequent embedding in paraffin at 2 weeks. Multiple 5 μm sections, 200 μm apart, from bregma −1.86 to −2.26, were prepared from each brain, and stained with hematoxylin and eosin (H&E, Thermo Scientific, Pittsburgh, PA). Additional 5 μm sections were obtained from the interval tissue and stained with Fluoro-Jade C (FJC, Chemicon, Temecula, CA) to evaluate for neuronal degeneration at 24 hr (Schmued et al, 2005). Sections stained with FJC were assessed qualitatively. Hippocampal neuronal damage was quantified with 7 day cell counts in H&E sections by blinded evaluator using Image J (http://rsb.info.nih.gov/ij/). Cell counts were quantified in CA1 and CA3, and are reported as the average number of normal appearing neurons per 100μm hippocampal pyramidal cell layer length. The 5μm H&E sections taken from bregma −1.86 to −2.26 were also qualitatively evaluated by a neuropathologist (RHG) blinded to treatment group.

Statistical methods

Physiologic parameters and cell counts were compared between groups using one-way analysis of variance (ANOVA), and post-hoc tests with appropriate correction for multiple comparisons. All data are provided as mean ± SEM. The primary outcome parameter of the study was neuron counts in CA1 hippocamus ipsilateral to CCI (or in the left hippocampus in HS only). Significance was determined by a p value ≤ 0.05.

RESULTS

Physiology

Table 1 provides a summary of important physiologic variables. MAP (the mean of all values for each group during the HS interval) was significantly lower during Shock in all groups with HS compared to CCI only: (60HS only, 90HS only, 60CCI+HS, and 90CCI+HS respectively, all p < 0.05 vs CCI only). In addition, the MAP during HS for all groups was within the target range of between 30 and 40 mmHg. During Pre-Hospital resuscitation, MAP increased into the target range of ≥ 50 mmHg in all groups and was between 50 and 60 mmHg in all of the HS groups (with or without TBI, data not shown). In contrast, MAP in the CCI only group was higher, as anticipated, at 77.3 ± 3.0 mmHg. During Definitive Care resuscitation, MAP recovered to ≥70 mm Hg in the HS only groups while 60 CCI+HS and 90 CCI+HS, were nearly 70 mmHg at 69.9 ± 5.0 and 69.9 ± 6.9 mm Hg respectively.

Table 1.

		Baseline	Shock	Definitive Care
MAP	CCI only	85.4 ± 6.6	80.2 ± 4.0	77.0 ± 2.5
	60 HS only	90.2 ± 3.6	35.5 ± 3.6	74.5 ± 5.4
	90 HS only	89.8 ± 5.3	39.3 ± 4.1	72.8 ± 4.3
	60 CCI+HS	88.5 ± 8.4	34.3 ± 5.0	69.9 ± 5.0
	90 CCI+HS	85.4 ± 5.5	36.3 ± 6.4	69.9 ± 6.9
HCT	CCI only	38.7 ± 4.6	36.3 ± 4.5	32.4 ± 4.3
	60 HS only	36.6 ± 1.4	27.6 ± 1.8	30.1 ± 1.9
	90 HS only	35.7 ± 1.1	26.6 ± 1.7	29.3 ± 1.8
	60 CCI+HS	37.1 ± 3.7	27.7 ± 3.7	28.5 ± 2.4
	90 CCI+HS	36.8 ± 2.6	28.3 ± 2.0	28.3 ± 2.0
Lactate	CCI only	2.5 ± 0.1	2.4 ± 0.3	1.8 ± 0.2
	60 HS only	2.2 ± 0.1	4.0 ± 0.4	1.6 ± 0.1
	90 HS only	2.3 ± 0.1	3.2 ± 0.2	1.6 ± 0.1
	60 CCI+HS	2.6 ± 0.2	3.5 ± 0.3	1.9 ± 0.2
	90 CCI+HS	2.5 ± 0.2	4.2 ± 0.4	2.2 ± 0.1
Base Deficit	CCI only	−4.7 ± 0.4	−5.2 ± 0.4	−5.7 ± 1.5
	60 HS only	−4.7 ± 0.7	−6.9 ± 0.4	−4.8 ± 0.2
	90 HS only	−4.7 ± 0.3	−6.6 ± 0.2	−5.4 ± 0.4
	60 CCI+HS	−4.7 ± 0.3	−8.1 ± 0.6	−4.9 ± 0.5
	90 CCI+HS	−4.7 ± 0.4	−7.2 ± 0.3	−5.1 ± 0.5

Hematocrit decreased by ∼30% in mice subjected to HS and CCI+HS during Shock and Pre-Hospital phases; hematocrit partially recovered in all groups in Definitive Care (Table 1). Ostensibly, lack of complete recovery resulted from hemodilution from volume resuscitation with hetastarch required to maintain MAP. Fluid requirements during Pre-Hospital and Definitive Care were 0.29 ± 0.1, 0.24 ± 0.1, 0.34 ± 0.1, and 0.34 ± 0.1 mL of 6% hetastarch in 60HS only and 90HS only versus 60CCI+HS and 90CCI+HS groups, respectively, and did not significantly differ between groups.

Not surprisingly, compared to CCI only, groups subjected to HS or CCI+HS had higher lactate levels and greater base deficits during Shock; all measurements were taken at the same protocol time-point. These values were significant at p ≤ 0.05 for CCI vs 60CCI+HS and 90CCI+HS groups for base deficit during Shock. For blood lactate levels, there was a predictable rise and fall during Shock and both resuscitation phases in 60HS, 90HS and 60CCI+HS groups compared to CCI only (p ≤ 0.05) ; however, lactate levels in the 90CCI+HS group continued to be significantly higher during Definitive Care as well, likely reflecting continued lactate “wash-out” with resuscitation after prolonged HS.

All groups had similar trends in PaO₂ (Table 2); there was appropriate equilibration after room air administration, and expected increase with initiation of 100% oxygen during Definitive Care. PaCO2 decreased in HS only and CCI+HS groups during Shock, while pH did not, a difference likely related to compensatory hyperventilation. There were no significant differences between groups with regard to glucose, osmolality, sodium, potassium, ionized calcium or ionized magnesium at any of the sampling times, (Table 3, all data not shown).

Table 2.

		Baseline	Shock	Definitive Care
pH	CCI only	7.34 ± 0.01	7.37 ± 0.01	7.33 ± 0.02
	60 HS only	7.38 ± 0.01	7.40 ± 0.01	7.35 ± 0.01
	90 HS only	7.36 ± 0.01	7.41 ± 0.01	7.33 ± 0.01
	60 CCI+HS	7.36 ± 0.01	7.36 ± 0.02	7.34 ± 0.01
	90 CCI+HS	7.37 ± 0.01	7.38 ± 0.01	7.31 ± 0.02
paCO2	CCI only	37.0 ± 3.4	31.7 ± 3.9	35.2 ± 4.7
	60 HS only	31.6 ± 2.4	25.3 ± 1.6	35.8 ± 2.2
	90 HS only	33.6 ± 3.1	25.3 ± 1.5	36.7 ± 3.4
	60 CCI+HS	33.5 ± 3.9	26.9 ± 1.8	36.4 ± 4.2
	90 CCI+HS	33.3 ± 4.3	26.9 ± 1.9	40.0 ± 6.6
paO2	CCI only	162.3 ± 7.9	79.4 ± 9.0	446.4 ± 25.0
	60 HS only	163.0 ± 17.0	99.9 ± 8.5	441.1 ± 10.5
	90 HS only	157.6 ± 13.7	93.2 ± 6.8	432.3 ± 18.4
	60 CCI+HS	154.2 ± 14.3	87.6 ± 9.9	444.6 ± 27.2
	90 CCI+HS	168.7 ± 12.3	92.2 ± 10.7	465.9 ± 19.9

Table 3.

		Baseline	Shock	Definitive Care
Sodium	CCI only	142.1 ± 0.3	142.6 ± 0.6	142.6 ± 0.7
	60 HS only	142.6 ± 0.3	140.8 ± 0.2	143.5 ± 0.3
	90 HS only	142.4 ± 0.4	140.2 ± 0.5	143.3 ± 0.2
	60 CCI+HS	142.7 ± 0.4	140.4 ± 0.4	143.1 ± 0.5
	90 CCI+HS	142.1 ± 0.5	140.5 ± 0.5	144.0 ± 0.5
Glucose	CCI only	88.8 ± 10.9	134.0 ± 19.9	87.1 ± 6.7
	60 HS only	92.9 ± 7.2	169.0 ± 11.0	74.8 ± 8.1
	90 HS only	101.9 ± 9.3	151.1 ± 12.2	71.6 ± 5.8
	60 CCI+HS	94.7 ± 9.7	148.5 ± 20.6	73.8 ± 4.2
	90 CCI+HS	87.6 ± 11.9	179.9 ± 10.2	85.4 ± 6.7
Osmolality	CCI only	290.0 ± 1.1	294.8 ± 1.7	292.3 ± 1.3
	60 HS only	289.9 ± 1.2	293.2 ± 1.2	293.5 ± 0.9
	90 HS only	293.2 ± 1.3	293.4 ± 1.8	296.3 ± 1.8
	60 CCI+HS	291.2 ± 1.3	291.9 ± 1.7	293.0 ± 1.3
	90 CCI+HS	291.9 ± 2.7	293.4 ± 1.8	296.3 ± 1.8

Neuropathology

Hippocampal Neuron Counts

Surviving CA1 neuron counts in dorsal hippocampus ipsilateral to injury in the 90CCI+HS group were significantly reduced compared to all other study groups (Figure 2). Average ipsilateral CA1 neuron counts 90CCI+HS were 16.5 ±14.1 versus 30.8 ±6.8, 32.3 ±7.6, 30.6 ±2.2, 28.1 ±2.2 neurons per 100 μm pyramidal cell layer length (CCI only, 60HS only, 90HS only, and 60CCI+HS, respectively, all p <0.05). Examples of mice from the four insult groups are shown in Figure 4). There were no significant differences between groups for hippocampal neuron counts in CA3 ipsilateral to injury (Figure 3) or in either CA1 or CA3 contralateral to injury suggesting that CCI alone at this relatively mild level, HS alone or combined injury did not produce significant neuronal loss in these hippocampal subfields.

Figure 2.

Average number of surviving CA1 hippocampal neurons per 100 μm length for each experimental group, both ipsilateral (light bars) and contralateral (dark bars). Data are mean and SEM, n= 6 for each group, * p < 0.05 compared to all other groups.

Figure 4.

Representative 20x H&E microphotographs depicting the CA1 hippocampal subfield in A. 90HS only, B. CCI only, C. 60CCI+HS and D. 90CCI+HS. 60HS not shown. Pyramidal neuron loss is evident within the medial region of CA1 in the 90CCI+HS group.

Figure 3.

Average number of surviving CA3 hippocampal neurons per 100 μm length for each experimental group, both ipsilateral (light bars) and contralateral (dark bars). Data are mean and SEM, n= 6 for each group. There were no differences between groups.

H&E Neuropathologic Survey

Review of 7 d H&E sections from 90CCI+HS mice consistently demonstrated hemorrhage and focal, full-thickness necrosis of the parietal cortex overlying the dorsal hippocampus ipsilateral to injury. Moderate acute eosinophilic degeneration was observed in the hippocampal neurons of the underlying dorsal subiculum, CA1, CA4, and dentate gyrus, particularly the dorsal blade. Occasional eosinophilic neurons were noted in CA3 of the hippocampus as well as in the dorsal thalamus. Scattered microglial and neutrophil infiltrates were present, as was neuropil vacuolation. The contralateral sides lacked abnormal histologic alterations in the 90CCI+HS group, as well as in all other study groups. Sections from the CCI only and 60CCI+HS groups demonstrated identical full-thickness necrosis of the injured parietal cortex. However, unlike 90 CCI+HS, eosinophilic neuron degeneration in CA1 and dentate gyrus was more mild in quality, and eosinophilic change in thalamic neurons was rare. HS only sections revealed no evidence of neuronal damage. Additional selected images from the neuropathological survey are presented in Figure 5.

Figure 5.

A. Representative medium-high power (20x objective) micrograph of a 90CCI+HS mouse hippocampus demonstrating normal pyramidal neurons (e.g. at arrowhead) stained blue with DAPI and degenerating neurons stained yellow-green with Fluoro-Jade C (e.g. at long arrow) within the CA1 subfield at 24 hours. B. High power (40x objective) micrograph of a H&E-stained section demonstrating normal pyramidal neurons (e.g. arrowheads) interspersed with degenerating eosinophilic neurons (e.g. long arrows) at 7 days in the CA1 hippocampal subfield of a 90CCI+HS mouse. C. Low power (4x objective) microphotograph of the most severe spectrum of hippocampal damage sustained by several mice in the 90CCI+HS group. Robust microglial and PMN infiltration is noted along the superior aspect (arrows), and there is substantial loss of neurons in the CA1 subfield (asterisks). D. Higher magnification (20x objective), of the boxed region in Figure 4C demonstrates the absence of viable pyramidal neurons and loss of architecture within much of the CA1 subfield in this example.

FluoroJade-C staining at 24 h post-insult

FJC positivity at 24 h was seen predominantly in CA1 and dentate gyrus and largely restricted to mice in the 90CCI+HS group (Figure 5). This corresponded with regions of CA1 neuron loss at 7 d as assessed by H&E staining, corroborating neuronal degeneration in the observed areas of subsequent neuron loss. Rare FJC positive neurons were seen in 60CCI+HS mice.

DISCUSSION

We specifically chose a level of TBI that produced a cortical contusion that was just below the threshold for overt neuronal loss in the underlying dorsal hippocampus. We also selected a clinically relevant level of HS, based on the work of Carillo et al. (1998), with a MAP that we anticipated would not produce neurological injury in mice subjected to HS alone at the durations studied in our protocol. We were, however, surprised that 90 min rather that 60 min of HS was required to exacerbate neuronal death after the chosen level of CCI. When Shock duration was extended to 90 min we observed a neuronal loss pattern previously well defined in experiments of ischemia and hypoxemia, namely, ∼ 60% loss of selectively vulnerable CA1 pyramidal neurons in the dorsal hippocampus by 7 days after the insult. Whether or not additional neuronal loss would be seen at longer outcomes remains to be determined. The duration of HS required to produce hippocampal neuronal death after TBI was longer than anticipated since in studies of CCI in rats, addition of 30 min hypoxemia (PaO₂ ∼ 40 mmHg) was sufficient (Clark et al., 1997). However, systemic hypoxemia in those studies resulted in the development of hypotension after ∼ 15−20 min, and combined hypoxemia and hypotension is likely to be particularly deleterious (Siesjo 1978). The fact that hypotension often develops in TBI models where secondary hypoxemia is superimposed is, in our opinion, underappreciated. In addition, those studies with hypoxemia in rats used a relatively greater injury severity level than used in our study, which could also importantly increase the level of vulnerability of the injured hippocampus to a secondary insult. Given that the normal MAP in mice anesthetized with isoflurane in our model was ∼ 85−90 mmHg, our studies indicate that HS to a MAP that is 50−60% below baseline can be tolerated for 60 min after TBI–suggesting that there may be a greater therapeutic time window for successful resuscitation to mitigate deleterious consequences of a secondary insult in the traumatically injured brain. This finding is similar to the work of Stern et al., (2000) who reported that acute cerebral hemodynamic parameters were preserved in pigs after FPI, despite HS to a MAP of 30 mmHg for a period of 60 min. Longer periods of shock were not studied in that model.

In pilot studies, to produce an identical level of hypotension in mice subjected to either HS or CCI+HS, it was necessary to use a greater degree of hemorrhage in HS only mice (2.7 ml/100g blood withdrawal vs 2.0 ml/100g in HS only and CCI+HS groups, respectively). This relationship between TBI and reduced tolerance to HS has been reported (Law et al., 1996; Yuan and Wade, 1992) and suggests a systemic consequence of CNS trauma on blood pressure regulation. Chesnut et al., (1998) observed hypotension in humans with isolated TBI and without significant extracerebral injury. Mahoney et al., (2003) confirmed and expounded on this observation, citing possible brainstem involvement, altered autonomic tone, or massive catecholamine surge with ensuing transmitter depletion, receptor saturation and consequent myocardial depression and cardiovascular collapse. Using a rat model of FPI and HS, Law et al., (1996) showed that rodents subjected to either isolated brain injury or HS were able to adjust vascular tone to maintain MAP, however, when these insults were combined, compensatory vasoconstriction during shock failed to occur. Yuan et al., (1992) showed both suppression of spontaneous MAP recovery in rats subjected to combined injury and attenuation of the MAP response to fluid resuscitation. We initially attempted to use higher MAP resuscitation goals in pilot studies, ≥ 60 mm Hg MABP in the Pre-Hospital phase and ≤ 80 mm Hg (normotension) MAP in the Definitive Care phase. However, targeting these goals resulted in uniform mortality in unintubated, spontaneously breathing mice. We observed frank pulmonary edema, possibly neurogenic in origin or due to myocardial depression. The possibility of cerebral edema in the face of aggressive fluid resuscitation also cannot be excluded, as we have not yet measured ICP in this mouse model. In this initial study, we were concerned that addition of invasive ICP monitoring would potentially compromise long-term survival in these small rodents. Future studies of this important parameter are needed. Nevertheless, our data clearly confirm that the cardiovascular system is sensitized to the hemodynamic consequences of HS by a preceding TBI.

We chose to use hetastarch as our resuscitation fluid in this model since it is the current standard of care in United States Army for combat casualty resuscitation (Holcomb, 2003). Other resuscitation fluids will need to be tested in this model since they may exhibit differing efficacies in neuroprotection even if the same MAP targets are used.

While we did not evaluate CBF in these mice, the literature is replete with evidence of cerebrovascular dysregulation and regional blood flow reductions after TBI. Posttraumatic hypoperfusion after CCI in rats has been reported across laboratories and assessment techniques (Bryan et al., 1995, Hendrich et al., 1999) and loss of blood pressure autoregulation of CBF after TBI has been reported (Lewelt et al., 1980)–potentially allowing exacerbation of cerebral hypoperfusion with even modest reductions in MAP. In injured regions with increased metabolic requirements, small reductions in CBF, or perhaps simply the failure to increase CBF and substrate delivery to match demands could be sufficient to damage vulnerable neurons when thresholds of energy failure are reached. Currently, we are using perfusion magnetic resonance imaging to study CBF in this model (Dennis et al., 2006).

In the CCI model in rodents, marked cerebral hyperglycolysis has been reported in the hippocampus underlying the contusion (Hovda et al., 1995; Statler et al., 2003). In HS, anemia accompanies the reduction in MAP and we observed a significant reduction in HCT; therefore the anticipated reduction in oxygen and substrate delivery produced by hypoperfusion would be amplified. The contribution of anemia to the exacerbation of neuronal injury in combined TBI and HS, however, remains to be defined.

We used isoflurane anesthesia in our model. Isoflurane by inhalation provides a consistent level of anesthesia that can be readily titrated and promptly discontinued. However, isoflurane reduces cerebral metabolic demands, provides some degree of CBF promotion – particularly in subcortical structures, and is neuroprotective after TBI (Statler et al., 2000; 2006). Severe TBI in humans is generally not treated with sedatives or anesthetics in the field. Nevertheless, it is necessary to provide anesthesia in animal models of TBI and thus our model may underestimate the amount of damage that a similar level of TBI and HS would produce in the clinical setting. Studies of combined TBI plus HS in this model using less protective and more clinically relevant anesthetics and/or analgesics such as fentanyl are warranted.

Most studies of combined experimental TBI and HS have been carried out in large animals and have been focused on the influence of resuscitation fluids on intracranial dynamics, including ICP and CBF (Gibson et al., 2002; Shackford, 1997; Bedell et al., 1998). Long-term outcomes in these models are extremely expensive and have generally not been investigated, thus, the impact of HS on delayed neuronal death has been subjected to limited investigation. There have been a few studies in rodent models, however, that are noteworthy.

In rats, using combined mild-moderate FPI and HS to a MAP of 50≥60 mmHg for 30 min, Schütz et al., (2006) did not observe either an increase in cerebral edema or an exacerbation of cortical tissue loss when FPI+HS was compared to FPI alone. They did, however, observe a delay in cognitive recovery in FPI+HS rats, suggesting subtle cerebral injury not necessarily related to edema or cortical tissue damage alone. They used a milder level of hypotension compared to our model, as well as a shorter shock duration, and without re-infusion of shed blood in the resuscitation phase, clinical extrapolation may be limited. Failure to treat the marked anemia that accompanies HS after TBI would be deemed to be outside of the current standard of care in TBI management. Matsushita et al., (2001) performed a similar protocol in rats also using moderate FPI and HS to a MAP of 60 mmHg for 30 min, however, they observed exacerbated contusion size in the posterior parietal cortex in FPI+HS rats when compared to FPI alone; they did not report assessment of the hippocampus. Similarly, with a different experimental TBI mechanism, impact acceleration, Ito et al., (1996) added 30 min of hypoxemia and hypotension (a PaO₂ of 40 mmHg and MAP of 30 mmHg, respectively) and used diffusion-weighted imaging to discern apparent diffusion coefficients (ADCs) and extrapolate cytotoxic versus extracellular edema. Combined injury rats demonstrated marked and sustained increases in ICP and reduced CBF as well as reduced ADC, consistent with cytotoxic edema, despite resuscitation, when compared to controls. When Barzó et al., (1997) used this identical protocol but added an additional combined injury group of impact acceleration, hypoxemia to a PaO₂ of 40 mmHg and hypotension with a MAP of 40−50 mmHg, they observed recovery of ADC and clinical condition in the MAP 40−50 mmHg combined insult group--while the MAP 30−40 mmHg group did not recover, and progressed to death. This suggested a critical MAP threshold of between 30 and 40 mmHg, which was the MAP seen in our study during HS in all groups exposed to shock. Similar to the work of Clark et al., using CCI and secondary hypoxic insult in rats, the earlier work of Ishige et al., (1987c) in FPI used 2,3,5-triphenyltetrazolium chloride to reveal an ischemic area surrounding the contusion not seen with either isolated FPI or hypoxemia alone. They also used MRI to confirm extension of the contusion and surrounding edema in combined insult rats vs TBI alone and observed a decrease in CBF in the entire ipsilateral cortex in these rats. Further work by Ishige et al., (1988) used phosphocreatine (PCr)/ inorganic phosphate ratios obtained by in vivo phosphorus-31 magnetic resonance spectroscopy as evidence for changes in high energy metabolite concentrations, and observed that depletion of high energy metabolites was markedly accelerated by combined FPI and hypoxemia in a dose dependent fashion, vs FPI alone. They added a hypotensive insult (MAP 30−40 mmHg) to the combination of FPI and hypoxemia and observed even further depletion of high energy phosphates in the brain—again, consistent with the MAP used in our study. These studies support the possibility that HS exacerbated energy failure in the pericontusional brain regions, including hippocampus, thus triggering neuronal death in our model.

We re-infused autologous shed blood –which would not be available for clinical use. It is recognized that massive transfusion of packed red blood cells can produce a number of unwanted side effects (i.e., immunosuppression, hyperkalemia, coagulopathy) and that complications from transfusion are generally related to the duration of blood storage prior to use. Thus, one might anticipate that secondary injury and complications are underestimated in our mouse model relative to the human condition. Recently, in the setting of massive blood loss in combat casualty care, greater emphasis has been placed on the use of fresh whole blood.

Finally, we focused our histopathological examination to hippocampal neuronal counts given the anticipated vulnerability of that brain region to secondary ischemic insults. However, further studies are needed to assess cortical damage and other brain regions, given the fact that our qualitative survey suggested the possibility of enhanced damage in other structures.

While technically challenging in mice, an experimental model of combined TBI plus HS is feasible with reasonable clinical fidelity. This initial study characterizes a new model which, given the ready availability of genetic variant mice, is unique in its potential for application to mechanistic and therapeutic study of this injury combination.