Secondary Logo

Journal Logo

Basic Science Aspects

Cerebral Alterations Following Experimental Multiple Trauma and Hemorrhagic Shock

Vogt, Nina; Herden, Christiane; Roeb, Elke; Roderfeld, Martin; Eschbach, Daphne§; Steinfeldt, Thorsten||; Wulf, Hinnerk||; Ruchholtz, Steffen§; Uhl, Eberhard; Schöller, Karsten

Author Information
doi: 10.1097/SHK.0000000000000943
  • Free



Multiple trauma (MT) is one of the leading causes of death in the age group up to 45 years, and therefore of considerable clinical and economical relevance (1). The overall mortality is about 20%, and the in-hospital case fatality rate averages up to 10.3%; 60% of the survivors still experience relevant disabilities 2 years after the trauma which include limitations of mobility and self-care, as well as restrictions in activities of daily living due to pain, neurological deficits, and posttraumatic stress disorder. About 12% of the MT patients are affected by a hemorrhagic shock (HS) that can potentially cause organ hypoperfusion and dysfunction, being responsible for 50% of all trauma-related deaths within the first 48 h (2). The brain reacts very sensitive to a reduction of perfusion and oxygenation and, in the worst case, this might result in global cerebral ischemia (3). However, the cerebral sequelae after MT/HS without traumatic brain injury (TBI) have not been well described yet.

The traditional guideline-supported therapeutic approach after MT/HS consists of optimized ventilation and immediate fluid resuscitation (4). Mild therapeutic hypothermia (33°C) has just recently been applied for organ protection after MT (5, 6). Fröhlich et al. (5) demonstrated that induced hypothermia attenuates the hepatic inflammatory response, and might therefore represent a potential therapeutic strategy to avoid posttraumatic organ dysfunction. Furthermore, the cerebroprotective effects of hypothermia in cardiac surgery, as well as after stroke or traumatic brain injury, are well known (7), and hypothermia is a therapeutic mainstay after cardiopulmonary resuscitation. However, hypothermia in conjunction with MT is not uncritical, since it has been shown that hypothermia, when combined with acidosis and coagulopathy—which is known as lethal triad—can also worsen the prognosis (6). Therefore, the primary aim of our study was to identify cerebral alterations after MT and HS without traumatic brain injury. Hence, we utilized an elaborate and realistic porcine MT model that has previously been described by our workgroup (8–11). The model includes tibia fracture, lung contusion, liver laceration, and HS with subsequent fluid resuscitation started within 90 or 120 min. The secondary aim of our study was to investigate potential cerebroprotective effects of mild therapeutic hypothermia (33°C).


Experimental animals were cared for prior to and at all stages of the experiment in compliance with all institutional guidelines and regulations of the Animal Care Committee of the District Government of Giessen, Hesse (Protocol Number MR 20/17, Ref. 22/2013). We acted in accordance with the accepted principal of the “3Rs” (restriction, refinement, reduction) and collected the data presented in this study as part of a larger trial on multiple injured animals of which different results were previously published (8, 9). For our current investigation, we used data of 60 male pigs (sus scrofa, Deutsche Landrasse) with a mean body weight of 35 ± 5 kg and an age of 3 to 6 months.

Experimental groups

We analyzed data from six different groups: MT with 45% blood loss and 90 min. HS phase, normothermia (T90; n = 15); MT with 45% blood loss and 90 min. HS phase, hypothermia (TH90; n = 15); MT with 50% blood loss and 120 min. HS phase, normothermia (T120; n = 10), MT with 50% blood loss and 120 min. HS phase, hypothermia (TH120; n = 10), no trauma (sham; n = 5) and no trauma, hypothermia (shamH; n = 5). Furthermore, brains of five non-handled animals served as controls for histological and immunohistochemical investigations.

This model is a final experiment; animals that died during experimental period were not replaced and not included in the evaluations.

Anesthesia and monitoring

Anesthesia and monitoring were conducted as previously described by Eschbach et al. and Horst et al. (8, 9). Briefly, after a fasting period of 12 h anesthesia was induced, animals were intubated (7.5 ch tube) and continuously ventilated using pressure control (Draeger, Evita, Danvers, Mass) with a tidal volume of 6 mL/kg to 8 mL/kg, FiO2 of 30%. Respiratory rate was adjusted to achieve an expiratory pCO2 of 45 mm Hg to 55 mm Hg. General anesthesia was maintained during the entire study period by the application of propofol and sufentanil, as well as additional midazolam 2% in case of shivering. Vital signs were monitored by electrocardiogram (ECG)-synchronized pulse oximetry, ECG recording and by a naso-oesophageally placed temperature probe. Single shot antibiotic prophylaxis (cefuroxime 80 mg/kg/BW) was administered prior to further interventions. Under aseptic conditions arterial and venous catheters were inserted and tracheostomy was performed (8, 9). Multimodal cerebral monitoring was conducted with the NEUROVENT PTO 2L probe (Raumedic AG, Helmsbrecht, Germany; Fig. 1A) in all Trauma and Sham groups. After drilling a burr hole with a gimlet at the right frontal bone 1 cm lateral from the midline, and 1.5 cm below an imaginary line between the two orbitae, the probe was advanced into the right frontal lobe up to the 5 cm mark of the probe (Fig. 1B), and was subsequently fixed by the bolt kit. A flexible element attached the brain probe to the skin, and allowed a strain relief when animals were moved.

Fig. 1
Fig. 1:
A, Multimodal cerebral probe with bolt kit.

Trauma induction

Induction of multiple trauma was carried out as previously described by our study group (8). In brief, the right hind leg was placed in a drop-weight gadgetry, and a tibia fracture was induced by a 20 kg plumb-cuboid, which was dropped from a height of 100 cm. Afterward, blunt thoracic trauma was induced by a captive bolt (9 × 17, Dynamite Nobel, AG, Troisdorf, Germany). Subsequently, laparotomy was performed with two lacerations of the liver (caudal lobe). HS started parallel to laparotomy in a pressure-controlled mode by withdrawing blood from the femoral vein, until a mean arterial blood pressure (MABP) of 30 ± 5 mm Hg was reached in T90/TH90 groups (45% of estimated total blood volume), and until a MABP of 25 ± 5 mm Hg was reached in T120/TH120 groups (50% of estimated total blood volume), respectively.

The shock period was maintained for either 90 (T90/TH90) or 120 min (T120/TH120). Afterward, reperfusion was started and maintained for 1 h with warmed Ringer-Acetate solution and the 4-fold of the removed blood volume. Sham animals received no injury, hemorrhage, or fluid resuscitation.

Controlled induced hypothermia

After fluid resuscitation controlled hypothermia was induced by the ARCTIC SUN 5,000 temperature management system (Bard Medical, Medivance, Louisville, Ky) (9). Here, two hydrogel pads were placed from the end of the thorax to the flanks of the animals. Pigs were then cooled for 12 h (a core temperature of 33°C was reached within 3 h), followed by a rewarming period (0.5 °C/h) of 10 h. After the rewarming period animals were observed until euthanasia at 48.5 h after MT.

Multimodal cerebral monitoring

The following parameters were continuously monitored at a rate of 1 Hz during the whole monitoring period by the NEUROVENT PTO 2L probe: partial brain tissue oxygen pressure (PtiO2), temperature (T), and intracranial pressure (ICP). Data were recorded in the MRP2 logO DATALOGGER (Raumedic AG, Helmsbrecht, Germany), which was connected to a notebook for final data evaluation. Cerebral perfusion pressure (CPP) was calculated from MABP and ICP values. Time point “0” was set 15 min prior to trauma induction. Data were averaged in hourly values using Microsoft Excel 2010 (Microsoft Office Professional Plus 2010, Version 14.0.7166.5000, Microsoft Deutschland GmbH, Unterschleissheim, Germany).

Collection of blood samples

Blood gas analyses were carried out half-hourly from the beginning of the experiment until the end of the reperfusion phase, and subsequently, in two-hourly intervals. Full blood samples were collected from a central venous line using serum monovettes (SARSTED AG & Co, Nümbrecht, Germany) at the following six time points: prior to induction of trauma (0 h), right after trauma (1.5/2 h), right after fluid resuscitation (2.5/3 h), right after hypothermia (14.5 h), during the observation period (24.5 h), and before termination of the experiment (48.5 h). Samples were allowed to clot for 30 min prior to centrifugation. Subsequently, serum was removed and stored at −80°C.

Removal of the brain

For the removal of the brain the head was dislocated from the spine at the C0/C1 joint immediately after euthanasia. Subsequently, the skull was opened with tree saw-cuts. The brain was elevated from the skull and cut in half. The right hemisphere was fixed in 4% paraformaldehyde and embedded in paraffin (Tissue-TekVIP5Jr, Sakura Finetek, Staufen, Germany); the left hemisphere was placed in liquid nitrogen for 1 h, then removed and stored at −80°.

Quantification of serum neuron-specific enolase (NSE) and protein S-100B levels

Serum levels of NSE and Protein S-100B were determined using the following ELISA kits: Porcine Neuron Specific Enolase, E07N0025 (BlueGene, Shanghai, China) and Pig soluble protein-100B, CSB-E14063p (CUSABIO, Wuhan, Hubei Province, China), according to the manufacturer's protocol. All serum samples were centrifuged again before using.

Quantification of morphological brain damage and cerebral inflammation

To determine regions of brain lesions the entire right cerebral hemispheres of six animals (three normothermic and three hypothermic) were cut serially (slice thickness 0.5 cm, Rotatory microtome, RM 2255, Leica Biosystems, Nussloch, Germany) and stained with hematoxylin and eosin (H&E). Brain slides from the left hemisphere were investigated in two exemplary animals (one normothermic and one hypothermic animal) to exclude differences between the two cerebral hemispheres. The most significant changes were detected in the frontal lobe. Therefore, in further analyses of this study we focused on this region. Brain slices were examined for histological lesions using ×100 and ×200 magnifications. The presence of cerebral inflammation in meninges and brain tissue of the frontal lobe (magnification up to ×100) was assessed semiquantitatively using an arbitrary three-point inflammation score; either 1, 2, or 3 points with possible intermediates (0.5) were assigned according to the three following categories: low grade inflammation (1 point)—one to two layers of inflammatory cells in perivascular and meningeal areas; moderate grade inflammation (2 points)—three to four layers of inflammatory cells with moderate perivascular or vascular infiltration and meningeal localization; severe grade inflammation (3 points)—four or more layers of inflammatory cells in perivascular, meningeal, and parenchyma localization with vascular infiltration. An example of a severe inflammation (3 points according to our score) is provided in Figure 2.

Fig. 2
Fig. 2:
Exemplary case of a high grade (3 points) meningeal inflammatory reaction with neutrophil granulocytes, macrophages, and lymphocytes (arrows) in a T120 animal (frontal section HE-stained, ×100 magnifications).

Immunohistochemical staining

Immunohistochemical staining of the frontal lobe was performed in eight animals of each Trauma group as well as in all Sham and Control animals using an Iba1-antibody (to assess microglia reactivity) and protein-S100 antibody (to assess astrocytic activation). An iNOS (inducible nitric oxide synthase) antibody was used in five animals of each group for detection of M1 microglia and astrocytes.

All paraffin sections were deparaffinated (3× xylol, 2× isopropanol, 96% ethanol, 80% ethanol; 3 min each) and incubated in 0.3% H2O2 in methanol for 30 min to inhibit endogenous peroxidase activity. For the detection of iNOS, slices were pretreated with target retrieval buffer pH 9 for 23 min in 95°C following 15 min of cooling. Slides were blocked with 1.5% (Iba1, 2 h) and 5% (iNOS, 1 h) normal goat serum. The sections were then incubated with a rabbit anti-Iba1 polyclonal antibody and rabbit anti NOS2 polyclonal antibody (1:500; anti Iba1, Wako Code No. 019-179741, Neuss, Germany and 1:70 NOS2 sc-651, Santa Cruz Biotechnology, Heidelberg, Germany) overnight at 4° C. On the next day slices were washed three times in tris-buffered saline (TBS), and incubated with a biotinylated anti-rabbit made in goat antibody (1:500 (Iba1), 1:100 (iNOS); Vector Laboratories No. BA-1000, Burlingame, Calif). Sections were then washed three times in TBS and incubated with ABC-Complex Peroxidase Standard for Iba1 (9 μL A+B to 1 mL TBS/BSA 1%; Vector Laboratories No. PK 400; Burlingame, Calif) for 1 h at room temperature (RT) and with steptavidin (1:700, No. 43-4323 Invitrogen Corporation, Camarillo, Calif). After washing three times for 5 min in TBS and once in aqua dest., sections were incubated in fresh 0.05% di-aminobenzidine tetrahydrochloride (DAB) in 0.1 mol imidazol/HCL buffer and 0.01% H2O2 for 2 min (iNOS) and 10 min (Iba1). For control staining normal rabbit serum and TBS/ BSA 1% were used instead of the primary antibody. Finally, sections were washed ×3 with TBS and once in aqua dest., 5 min in Kardasewitch to reduce background, and then again two times in aqua dest. To stain the sections they were placed in Papanicolaous/hematoxylin S 1:10 for 30 s, and then for 5 min in warm tap water. Finally, slices were processed using an ascending alcohol row. For protein S-100 staining paraffin sections were deparaffinated, as described above, and incubated in 0.3% H2O2 in Methanol for 30 min to inhibit endogenous peroxidase activity, washed in TBS, and blocked with 5% normal goat serum and TBS/BSA 1% for 10 min. The sections were then incubated with a polyclonal rabbit anit-S100 (1:6,400; Polyclonal Rabbit Anti-S100; Code 0311, Dako; Glostrup, Denmark) overnight at 4°C. On the next day slices were washed three times with TBS, and then incubated with a biotinylated anti-rabbit antibody (1:500; Vector Laboratories No. BA-1000, Burlingame, Calif) for 30 min at RT. Sections were washed three times in TBS, and were subsequently incubated with ABC-Complex Peroxidase Standard (9 μL A + B to 1 mL TBS/ BSA 1%; Vector Laboratories No. PK 400; Burlingame, Calif) for 30 min at RT. Further steps are in analogy to the Iba1/iNOS staining protocol despite the DAB incubation of 2 min.

Quantification of microglia and astrocyte reactivity

Microglia and astrocyte activation was investigated in the frontal lobe using ×400 or ×200 magnification, respectively. In five visual fields of each region activated microglia and astrocytes were counted using light microscopy (Leica DMRB with counting grid 1 × 1 cm), and the mean values were calculated for the respective region. The morphological changes were noticed and described. Furthermore, iNOS-positive cells (microglia and astroglia) were examined in five animals of each group and 10 fields of view in the region of the frontal lobe using light microscopy (×400 magnifications; Leica DMRB with counting grid 1 ×1 cm).

Only cells that we could clearly identify as microglia or astrocytes based on the morphology were included in the analysis. Astrocytes exhibit a round light brown cell body with stellate branches which often attached neighbor cells. iNOS-positive microglia showed a smaller inhomogeneous cell body (triangular, rod-shaped, or lunate) with strongly branched processes and never contact other microglia in the neighborhood.

Statistical analysis

Statistical analysis was performed using SigmaPlot 12.3 Exact Graphs and Data Analysis (Systat Software GmbH, Erkrath, Germany). Data were analyzed with Kruskal–Wallis ANOVA on ranks test followed by Dunn Method for comparison of more than two groups. For comparison of two groups, the Rank Sum Test was applied. Statistical significance was assumed at P < 0.05. Data are presented as mean ± SEM if not indicated otherwise. The normal distribution was always rejected because it was not present at all time points. We compared the groups with one another at the same time point and not the individual times, because our interest was the influence of trauma intensity on the monitoring parameters.


Two animals of the T90 and two animals of the TH120 group, as well as one animal of the T120 group, died within the first 12 h after MT/HS. These animals were therefore excluded from the analysis.

Physiological parameters

Physiological parameters including blood gases were previously described by our workgroup (8, 10, 11).

Monitoring parameters

ICP values showed slight variations during the monitoring period that remained within the normal range in all groups. There were no inter-group differences (Fig. 3A). Hypothermia did not have a significant effect on ICP values when compared with normothermic animals. Subsequent to the trauma and shock phase, CPP values (Fig. 3B) reached its nadir in the trauma groups being significantly lower in the T90 (38.69 ± 9.22 mm Hg; P = 0.006 vs. sham) and in the T120 (35.11 ± 14.61 mm Hg; P = 0.006 vs. sham) groups compared with the sham group (63.02 ± 10.73 mm Hg). Values recovered in both trauma groups during the monitoring period, but were still significantly lower right after resuscitation in the T90 (44.01 ± 14.27 mm Hg, P = 0.016 vs. sham) and in the T120 group (39.24 ± 14.25 mm Hg; P = 0.016 vs. sham) compared with sham group (75.18 ± 16.70 mm Hg). During the further course of the experiment CPP of sham animals showed a gradual decrement to values around 40 mm Hg. On comparison of normothermic and hypothermic groups we noticed a significantly lower CPP at +14.5 h in the TH90 group (TH90: 34.34 ± 9.21 mm Hg vs. T90: 40.69 ± 7.52 mm Hg; P = 0.034).

Fig. 3
Fig. 3:
A, Time course of intracranial pressure (ICP) up to 48.5 h after termination of the shock phase in sham and normothermic trauma animals.

Initially, PtiO2 values of the T90 group were lower than in Sham animals or in the T120 group without reaching a statistically significant difference. Subsequent to MT/HS, PtiO2 values were significantly lower in the T90 (22.96 ± 17.54 mm Hg; P = 0.007 vs. sham) and in the T120 group (20.09 ± 11.75 mm Hg; P = 0.007 vs. sham) compared with the sham group (65.20 ± 36.52 mm Hg). During further course of the experiment PtiO2 curves of the trauma and sham animals converged at a plateau of around 40 mm Hg. The time course of PtiO2 is displayed in Figure 3C. At the end of the rewarming phase 24.5 h (20.44 ± 13.35 mm Hg; P = 0.029) and at the end of the monitoring phase 48.5 h (17.56 ± 10.00 mm Hg; P = 0.043) hypothermic animals of the TH90 group exhibited a significantly lower PtiO2 when compared with normothermic animals of the T90 group (+24.5 h: 33.49 ± 11.14 mm Hg and +48.5 h: 27.52 ± 10.00 mm Hg). These differences were not detectable in the groups with more pronounced trauma (T120/TH120).

Brain temperature differed around 0.2°C from the body temperature and varied between 37.9°C and 39.3°C during normothermia; there were no significant differences between groups. The temperature minimum during hypothermia phases was 33.2 ± 0.16°C in TH90 group (14.5 h) and 33.3 ± 0.34°C in TH120 group (14.5 h), and there were no significant differences between the hypothermia groups.

Biomarkers in blood serum

Protein S-100B values (Fig. 4A) were temporarily increased subsequent to MT/HS in the T120 group compared with sham animals (P = 0.02), whereas NSE values (Fig. 4B) were increased after MT/HS in the T90 group (P = 0.06) and in the T120 group (P = 0.006) when compared with sham pigs. Hypothermia in the TH90 and TH120 groups did not influence serum protein S-100B and NSE levels when compared with respective normothermic groups.

Fig. 4
Fig. 4:
A, Serum protein S-100B levels in normothermic trauma and sham animals at baseline (0 h), after the shock phase (1.5/2 h), after the resuscitation phase (2.5/3 h) and at the end of the experiment (48.5 h post-trauma).

Morphological brain damage

H&E stained brain sections did not reveal severe morphological damage. Brain tissue surrounding the implantation site of the cerebral probe showed, as expected, mild tissue damage including mild hemorrhage but no significant cerebral edema or infarct.

Cerebral inflammation

A relevant accumulation of inflammatory cells was detectable in brains of sham as well as trauma animals of both groups, whereas in control animals inflammation was almost not present. Interestingly, inflammatory cells were mainly found in the meninges consisting of neutrophil granulocytes, macrophages, monocytes, and lymphocytes. There was a significant difference in the inflammation score between T90 (P = 0.05 vs. control) and T120 (P = 0.05 vs. control) animals and control animals (Fig. 5A). Hypothermia significantly reduced cerebral inflammation (Fig. 5B) in the TH90 group (P = 0.014 vs. T90) but without significant reduction in the T120 versus TH120 group.

Fig. 5
Fig. 5:
A, Cerebral inflammation quantified by the 3-point score in control, sham and normothermic trauma animals.

Microglia and astrocyte reactivity

We found relevant microglia activation in the frontal lobe of both Trauma groups using Iba1 immunohistochemistry. The microglia cell count (Fig. 6A) was significantly elevated in the T90 (P < 0.001 vs. sham; P < 0.001 vs. control) and the T120 animals (P < 0.001 vs. sham; P < 0.001 vs. control). There was a tendency (P = 0.072) toward an increase of the microglia cell count in hypothermic animals (TH120) compared with normothermic animals of the T120 group in the frontal lobe. Despite the missing significant difference in cell count, hypothermic animals showed reactive microglia with enlarged cell somata and thickened processes that were more pronounced compared with normothermic animals and particularly evident in the hypothermic trauma groups. Furthermore, rod-shaped microglia with narrow cell somata and few processes were exclusively found in trauma animals (Fig. 6B).

Fig. 6
Fig. 6:
A, Microglia activation in the frontal lobe using Iba1-immunohistochemistry in control, sham and normothermic trauma animals.

The staining of protein S-100 revealed astrocytes with enlarged somata and elongated branches in all normothermic trauma groups when compared with control and sham groups. These morphological changes appeared to be more distinct in the hypothermic than in the normothermic trauma groups. However, astrocyte cell count did not differ between groups (data not shown).

There was a significant iNOS activation in microglia cells in the T120 group (1.04 cells/10 fields of view; P < 0.05 vs. control, Fig. 7), which was even more pronounced (1.96 cells/10 fields of view; P = 0.008 vs. T120) in the TH120 group. The cell count of the iNOS activated astrocytes was also significantly higher (5.36 cells/10 fields of view) in the T120 group compared with the control group (P < 0.05) and sham group (P < 0.05). There was no difference in the cell count of iNOS positive astrocytes between normothermic and hypothermic animals.

Fig. 7
Fig. 7:
Cell count of iNOS-stained microglia in the frontal lobe in control, sham and normothermic trauma animals.


Here, we present data on cerebral alterations following multiple trauma and hemorrhagic shock using an elaborate and realistic porcine long-term trauma model without direct traumatic brain injury. Due to our knowledge, this is the first study that assesses the cerebral sequelae of MT/HS without TBI in a large animal model. Major findings were temporary trauma/shock-related drops of CPP and PtiO2 values on multimodal cerebral monitoring as well as temporary serum protein S-100B and NSE elevations, which were overcome subsequent to fluid resuscitation. Furthermore, histological and immunohistochemical investigations revealed a cerebral inflammation in both trauma and sham and groups that was most probably influenced by the long-term intensive care setting. In contrast, cerebral microglia activation was clearly trauma-related, and iNOS staining pointed toward a cerebral injury pattern.

Hypothermia reduced cerebral inflammation when initiated early (TH90), but not when it was induced in a delayed fashion (TH120). Furthermore, morphological signs of microglia activation were pronounced in the hypothermia groups, which could potentially indicate a neuroprotective mechanism, but has to be clarified in further studies.

Multiple trauma and hemorrhagic shock

Multiple trauma often affects young and healthy patients leading to serious medical and socio-economic consequences (12). The long-term survival of patients after multiple trauma is significantly reduced compared with the general population, depending on the extent of the initial injury and the degree of permanent disability (13). Acute trauma-related mortality usually occurs due to traumatic brain injury (TBI) and/or severe hemorrhagic shock. Thus, about 50% of the patients who do not survive the trauma, die because of uncontrolled hemorrhage (2) leading to severe organ hypoperfusion and, ultimately, to organ failure. Despite fluctuations in systemic blood pressure the brain normally keeps the cerebral blood flow constant via autoregulation mechanisms. However, autoregulation fails when systemic blood pressure drops below 50 mm Hg, and ischemic brain damage can be the consequence (14). Thus, delayed morbidity and mortality after MT/HS can be caused by a secondary brain damage, but also by a systemic inflammatory response syndrome (SIRS) (1). Blunt trauma, especially chest trauma, results in a release of pro-inflammatory cytokines and activates the coagulation and complement cascades, which by themselves trigger an excessive response of the immune system. The resulting systemic inflammation may interact with initially unaffected organ systems mounting in a multiple organ dysfunction syndrome (1, 15). Our workgroup investigated interleukins and HMGB1 in serum following MT/HS and found severe signs of shock (increase of lactate) as well as a systemic increase of the pro-inflammatory mediators in the trauma groups (9–11).

When MT includes TBI, the extent and pattern of cerebral damage depends on the severity of TBI itself but also on the secondary damages, e.g., hypoxia, hypotension, inflammation with SIRS, and finally multi-organ dysfunction syndrome (MODS). Here, hemorrhagic shock is known to double the mortality and morbidity of TBI patients (16). Studies that investigated the combination of TBI and HS primarily focused on the effects of fluid substitution/resuscitation on intracranial dynamics including CPP, CBF and MABP (17–19). Further studies found that a HS following TBI can exacerbate functional deficits or delay long-term cognitive recovery, but the effects on structural damage vary between studies and may be explained, e.g., by different severities of TBI or shock/hypotension, or the anesthetic agents used (20–22). In this context the role of a TBI and HS triggered systemic inflammation that can potentially contribute to secondary brain damage, has also to be taken into account. Most interestingly, recently published studies (16, 23) showed that the addition of valproic acid to saline resuscitation was neuroprotective in a porcine model of combined TBI+HS, and that this effect was potentially mediated by down-expression of genes that ultimately regulate nuclear factor-kB (NF-kB)-mediated production of cytokines. Shein et al. (24), however, investigated the differences between TBI, HS, and TBI+HS in an animal model and found higher levels of anti-inflammatory IL10 in TBI+HS. They concluded that HS caused a shift toward an anti-inflammatory process. Overall, interactions between shock, multiple trauma ± TBI and the immune system are complex and warrant further investigations.

Animal model

There are many large animal models used in multiple trauma research. These are either models with isolated hemorrhage (volume/pressure controlled or uncontrolled) or trauma models that combine hemorrhage with brain, chest or abdominal trauma, or fracture (25). Interestingly, only few studies examined the long-term effects (>24 h) after combined trauma (25–27). Our model is reproducible (8) and realistically reflects the situation after multiple trauma including bone fracture, soft tissue damage, and uncontrolled bleeding from the liver stitch injury. The additional hemorrhagic shock aspect, however, was volume and pressure-controlled as a matter of standardization. Furthermore, our model simulates a long-term intensive care setting (study period up to 48.5 h) including multimodal organ monitoring (8). The cerebral sequelae of multiple trauma and/or hemorrhagic shock without traumatic brain injury have only been addressed by a few studies yet (3, 28–31). Pfortmueller et al. (29) retrospectively investigated serum protein S-100B concentrations in adult patients with major trauma independent of the presence of head injury. They found no statistically significant difference in protein S-100B concentrations between patients with and without head injury, indicating a systemic trauma-related release. Cavus et al. (30) examined the course of CPP, cerebral tissue oxygenation index (TOI) and cortical electrical activity (BIS), as well as arterial Protein S-100B concentrations (3) using a porcine hemorrhagic shock model. In conformity with our findings CPP and TOI decreased temporarily and recovered depending on the degree of fluid resuscitation until the end of the experiment. Protein S-100B was shown to increase temporarily after shock followed by a recovery, close to baseline values after therapy, which was confirmed by our results. The researchers discussed several reasons for the increased protein S-100B elevation including increased production, BBB leakage and cell damage. Interestingly, the increase in protein S-100B levels was significantly associated with a critical threshold CPP that has already been shown to correlate with inadequate cerebral perfusion. However, due to our knowledge long-term multimodal cerebral monitoring, long-term determination of serum markers for brain damage, as well as in-depth histological and immunohistochemical investigations to assess long-term cerebral alterations after multiple trauma without brain injury have not yet been conducted in a clinical or experimental setting.

Multimodal cerebral monitoring

ICP was rather constant over the entire monitoring period with values between 9.16 ± 2.36 mm Hg and 18.51 ± 1.96 mm Hg. Effects of hypothermia on ICP were not detectable, presumably due to the generally low values that were always within the physiological range. CPP values were expectedly temporarily impaired as a result of MABP reduction following trauma and hemorrhagic shock with minimum values of 34.04 ± 13.26 mm Hg. Despite this drop under the autoregulation threshold of 50 mm Hg and subsequent fluid resuscitation, histological examinations revealed no ischemic damage or significant cerebral edema. However, the trauma and shock-related microglia activation that we found might have been triggered by the temporary ischemia.

Ultimately, fluid resuscitation was sufficient to result in a recovery of MABP and CPP values until the end of the monitoring period.

The PtiO2 monitoring allows for a direct measurement of local tissue oxygen tension; hence, the relationship between cellular oxygen consumption and regional oxygen supply. Measurement is influenced by factors like the microvascular composition, cerebral perfusion, diffusion distance between capillaries and cells and the dominance of arterioles and venules in the area of probe placement (32). Stabilization of PtiO2 probe measurements can take 1 to 2 h according to the literature (33). In the current experiments baseline values of the brain probe were established 15 min prior to trauma induction after a stabilization time that ranged between 15 min and 1 h, dependent on the duration of the entire monitoring setup. This might explain the intergroup deviations of the initial PtiO2 values supported by the fact that later recordings showed almost identical courses in the two trauma groups indicating reliable and reproducible measurements. Furthermore, comparison of CPP and PtiO2 curves suggest, as expected, an interrelation of values. Overall, PtiO2 values only temporarily dropped under the hypoxia threshold of 20 mm Hg (minimum value 10.61 ± 11.17) in the trauma groups followed by a recovery during the course of the experiment to values of ∼ 30 mm Hg. There are few experimental studies in which the Raumedic probe had been used. The probe seems to deliver reliable and stable values with the PtiO2 measurement being the most sensitive modality, which can be confirmed by our experiments (14, 33, 34). Our PtiO2 values were higher than previously shown (32, 34), particularly at the beginning of the experiment, which could be related to different stabilization times as already described, but also to a different position of the tip of the probe compared with other studies.

Serum biomarkers of brain damage

There is a wide range of serum biomarkers for prognostic assessment after trauma (35). The most prominent markers are Protein S-100B and NSE, which have been extensively investigated as predictors of neurological outcome, mainly following traumatic brain injury, stroke, shock, and hypoxic-ischemic encephalopathy (36, 37). However, these markers are also upregulated following trauma without traumatic brain injury including femur fracture, but also after cardiac surgery and after hemorrhagic shock (3, 38–41). Protein S-100B is a calcium-binding protein, located primarily in schwann cells and the cytoplasm of astroglia. The physiological function depends on its concentration ranging from stimulation of neurite growth to induction of apoptosis (39). NSE is a dimeric intracellular glycolytic enzyme which is localized in the cytoplasm of neurons and other neuroendocrine cells, but it is also found in platelets and erythrocytes, being passively released by cell destruction. NSE participates in axonal transport, and its expression levels can fluctuate depending on energy demand within a cell. In the case of axonal injury it is upregulated to maintain homeostasis (38). In serum of healthy humans protein S-100B is present in very low concentrations around 0.05 μg/L, concentrations above 2.0 μg/L are associated with a poor outcome after neurological damage (29). However, protein S-100B is also frequently elevated in critically ill patients without obvious brain tissue damage (41); increased lactate levels and low arterial pressure, pH, and Hb levels seem to have an influence on the synthesis of protein S-100B (41). Meybohm et al. (3) investigated the release of serum protein S-100B in a porcine hemorrhagic shock model and found an increase up to 0.75 μg/L or more when the CPP was lower than 30 mm Hg. In our experiment we temporarily measured protein S-100B concentrations greater than 3.0 μg/L. However; histologically we could not detect gross cerebral cell damage or significant astrocytosis in our trauma animals. Furthermore, normal protein S-100B levels of pigs have not yet been clearly determined, and the values highly depend on the assay kits used for analysis. Overall, the rise of protein S-100B and NSE concentrations that we measured is presumably triggered by the trauma and/or the shock, but not by cerebral damage.

Histological and immunohistochemical findings

Our histological examination 2 days after MT/HS demonstrated a significant cerebral inflammatory reaction, particularly in the rostral brain regions, that was present in all groups except of the control group. Furthermore, immunohistochemistry revealed a microglia activation, reactive astrocytes, and iNOS-positive microglia in all trauma groups with different expression. The reason for the detected cerebral inflammation is not entirely clear, but the fact that inflammatory changes were also found in sham animals speaks against a purely trauma-related phenomenon. One explanation would be a systemic inflammation triggered by the intensive care setting including long-term ventilation and anesthesia, as well as invasive monitoring devices. The cerebral probe might have also triggered a cerebral inflammation; however, inflammatory changes were not particularly pronounced around the trajectory of the probe or in the hemisphere ipsilateral to the probe.

Microglia cells are activated early in a pathological setting, and their activation can be triggered by a peripheral inflammation, a systemic inflammatory response, neuro-immune interactions, either humorally by proinflammatory cytokines or by direct interactions with the brain at circumventricular organs and impaired BBB. After cerebral damage the cells exhibit dual functions that are dependent on their cytokine activation profile: they can either polarize in the pro-inflammatory M1 phenotype (classically activated), which maintains inflammatory responses and triggers neurodegenerative processes, or in the M2 phenotype (alternatively activated), which can remove cell detritus by phagocytic function and allows repair processes (42). Four major cytokines that are employed by M2 microglia to antagonize a pro-inflammatory response are IL-4, IL-13, IL-10, and TGF-β; Ym1 and arginase-1 are potent markers for M2 polarization. Pro-inflammatory cytokines that play a role in M1 activation are IL-8, IL-6, TNF-α, and iNOS (42). The obvious trauma/shock-related microgliosis and activation that we found including the iNOS-positive differentiation pattern might indicate a persistent mild form of cerebral injury. Since iNOS activation was more pronounced and only significant in the T120 group, this process might be dependent on the trauma/shock intensity.

Cerebral effects of therapeutic hypothermia

Hypothermia did not influence any neuromonitoring parameters or serum biomarkers. However, cerebral inflammation, most probably triggered by the intensive care setting, was reduced by hypothermia in the TH90 but not in the TH120 group. Microglia and also astrocytes demonstrated more pronounced morphological signs of activation in the hypothermic groups, and also iNOS reactivity was also increased in hypothermic animals. Previous studies demonstrated a differential pattern of microglial behavior including activation and migration depending on the timing and degree of hypothermia (43). The M1 polarization indicated by the positive iNOS expression of microglia, which was accentuated by hypothermia, points toward the same direction as the higher IL-6 and IL-8 values in the hypothermia groups that our workgroup demonstrated previously (9). Overall, in our experiments hypothermia did not lead to a downregulation of the microglia cell count or activation, as one could have expected (44); furthermore, the reduced cerebral inflammation in the TH90 group, which could not be reproduced in the TH120 group with higher trauma/shock severity and delayed hypothermia, has to be interpreted with caution.

Taken together, the polarization of microglia ultimately determines whether these cells are neuroprotective or neurodestructive. The TH120 group showed a significant higher iNOS activation in microglia than the TH90 group (P = 0.008 vs. TH90), so it seems as if induced hypothermia, at an early stage (TH90), might reduce iNOS-positive cells and inflammation, but not after severe trauma and delayed hypothermia (TH120). Nevertheless, further investigation with M1 and M2 markers was needed to realize the microglia polarization.

Limitations of the study

The parameters that we chose to assess cerebral consequences after trauma and shock were either derived from invasive cerebral monitoring or from histological/immunohistochemical investigations. The importance for clinical outcome, particularly of the microglia activation that we found, is difficult to guess. Thus, functional outcome parameters evaluated by pig behavior assessment tools would have been very helpful and sensitive to allow a clear judgement on cerebral alterations cognitive outcome. There are several tests like the neurological deficit score for pigs (45) and the neurologic alertness score (46) that can improve a study's informative value. However, with the severe injury pattern that we used this would not have been acceptable in terms of animal welfare.

The multimodal cerebral probe that we used has a tip diameter of 1.65 mm and a PtiO2 sensitivity area of ∼22 mm2 (data provided by the manufacturer). The probe thus allows for a punctual measurement of pathophysiological changes that potentially lead to a global cerebral injury pattern. Therefore, it cannot be excluded that we missed more pronounced ischemic or hypoxic changes due to the limited spatial coverage. However, histological und immunohistochemical investigations did not show signs of severe neuronal damage. Baseline values of the PtiO2 measurements were determined 15 min prior to trauma induction but after a varying duration after probe implantation. The interval between probe implantation and baseline reading, thus the stabilization time, was determined by the duration of the entire monitoring setup including placement of other probes that changed with the experience of the team. Thus, baseline PtiO2 values have to be interpreted with caution. However, later measurements were conducted well after the known stabilization time and were conclusive on inter-group comparison but also on comparison with CPP values. With our immunohistochemical investigations we only determined the M1 polarization of the microglia cells using an iNOS antibody. Future labeling of M2-positive microglia might allow for a better insight into the cerebral pathomechanisms following multiple trauma and hemorrhagic shock.


In this realistic porcine long-term model we did not find evidence of gross cerebral damage when fluid resuscitation was initiated within 120 min after multiple trauma and hemorrhagic shock without direct traumatic brain injury. However, trauma-related microgliosis and M1 microglia activation might be a consequence of a temporary hypoxia/ischemia and future research is necessary to clarify its relevance. Mild hypothermia seems to be capable of reducing cerebral inflammation in our model when initiated early, but its connection to microglia activation warrants further investigation.


1. Keel M, Trentz O. Pathophysiology of polytrauma. Injury 2005; 36:691–709.
2. Wutzler S, Maegele M, Wafaisade A, Wyen H, Marzi I, Lefering R. Risk stratification in trauma and haemorrhagic shock: scoring systems derived from the TraumaRegister DGU. Injury 2014; 45:29–34.
3. Meybohm P, Cavus E, Doerges V, Weber B, Stadlbauer KH, Wenzel V, Scholz J, Steffen M, Bein B. Release of protein S100B in haemorrhagic shock: effects of small volume resuscitation combined with arginine vasopressin. Resuscitation 2008; 76:449–456.
4. Hussmann B, Lendemans S. Pre-hospital and early in-hospital management of severe injuries: changes and trends. Injury 2014; 45:39–42.
5. Fröhlich M, Hildebrand F, Weuster M, Mommsen P, Mohr J, Witte I, Raeven P, Ruchholtz S, Flohé S, van Griensven M, et al. Induced hypothermia reduces the hepatic inflammatory response in a swine multiple trauma model. J Trauma Acute Care Surg 2014; 76:1425–1432.
6. Weuster M, Mommsen P, Pfeifer R, Mohr J, Ruchholtz S, Flohé S, Fröhlich M, Keibl C, Seekamp A, Van Griensven M, et al. Induced hypothermia does not harm hemodynamics after polytrauma: a porcine model. Mediators Inflammation 2015; 2015:829195.
7. Dietrich WD, Bramlett HM. The evidence for hypothermia as a neuroprotectant in traumatic brain injury. Neurotherapeutics 2010; 7:43–50.
8. Eschbach D, Steinfeldt T, Hildebrand F, Frink M, Schöller K, Sassen M, Wiesmann T, Debus F, Vogt N, Uhl E, et al. A porcine polytrauma model with two different degrees of hemorrhagic shock: outcome related to trauma within the first 48 h. Eur J Med Res 2015; 20:73.
9. Horst K, Eschbach D, Pfeifer R, Relja B, Sassen M, Steinfeldt T, Wulf H, Vogt N, Frink M, Ruchholtz S, et al. Long-term effects of induced hypothermia on local and systemic inflammation—results from a porcine long-term trauma model. PLoS One 2016; 11:e0154788.
10. Horst K, Eschbach D, Pfeifer R, Hübenthal S, Sassen M, Steinfeldt T, Wulf H, Ruchholtz S, Pape HC, Hildebrand F. Local inflammation in fracture hematoma: results from a combined trauma model in pigs. Mediators Inflamm 2015; 2015:126060.
11. Horst K, Hildebrand F, Pfeifer R, Hübenthal S, Almahmoud K, Sassen M, Steinfeldt T, Wulf H, Ruchholtz S, Pape HC, et al. Impact of haemorrhagic shock intensity on the dynamic of alarmins release in porcine poly-trauma animal model. Eur J Trauma Emerg Surg 2016; 42:67–75.
12. Debus F, Lefering R, Frink M, Kühne CA, Mand C, Bücking B, Ruchholtz S. Numbers of severely injured patients in Germany. Dtsch Arztebl Int 2015; 112:823–829.
13. Mutschler W, Mutschler M, Graw M, Lefering R. Long-term survival after severe trauma. Der Unfallchirurg 2016; 119:546–553.
14. Purins K, Enblad P, Wiklund L, Lewén A. Brain tissue oxygenation and cerebral perfusion pressure thresholds of ischemia in a standardized pig brain death model. Neurocritical Care 2012; 16:462–469.
15. Visser T, Hietbrink F, Groeneveld KM, Koenderman L, Leenen LPH. Isolated blunt chest injury leads to transient activation of circulating neutrophils. Eur J Trauma Emerg Surg 2011; 37:177–184.
16. Halaweish I, Bambakidis T, Chang Z, Wei H, Liu B, Li Y, Bonthrone T, Srinivasan A, Bonham T, Chtraklin K, et al. Addition of low-dose valproic acid to saline resuscitation provides neuroprotection and improves long-term outcomes in a large animal model of combined traumatic brain injury and hemorrhagic shock. J Trauma Acute Care Surg 2015; 79:911–919.
17. Pinto FC, Oliveira MF, Prist R, Silva MR, Silva LF, Capone Neto A. Effect of volume replacement during combined experimental hemorrhagic shock and traumatic brain injury in prostanoids, brain pathology and pupil status. Arq Neuropsiquiatr 2015; 73:499–505.
18. Genét GF, Bentzer P, Ostrowski SR, Johansson PI. Resuscitation with pooled and pathogen-reduced plasma attenuates the increase in brain water content following traumatic brain injury and hemorrhagic shock in rats. J Neurotrauma 2017; 34:1054–1062.
19. Imam AM, Jin G, Duggan M, Sillesen M, Hwabejire JO, Jepsen CH, DePeralta D, Liu B, Lu J, deMoya MA, et al. Synergistic effects of fresh frozen plasma and valproic acid treatment in a combined model of traumatic brain injury and hemorrhagic shock. Surgery 2013; 154:388–396.
20. Hemerka JN, Wu X, Dixon CE, Garman RH, Exo JL, Shellington DK, Blasiole B, Vagni VA, Janesko-Feldman K, Xu M, et al. Severe brief pressure-controlled hemorrhagic shock after traumatic brain injury exacerbates functional deficits and long-term neuropathological damage in mice. J Neurotrauma 2012; 29:2192–2208.
21. Schütz C, Stover JF, Thompson HJ, Hoover RC, Morales DM, Schouten JW, McMillan A, Soltesz K, Motta M, Spangler Z, et al. Acute, transient hemorrhagic hypotension does not aggravate structural damage or neurologic motor deficits but delays the long-term cognitive recovery following mild to moderate traumatic brain injury. Crit Care Med 2006; 34:492–501.
22. Rehberg SC, Raum MR, Rammelt S, Schneiders W, Neugebauer EA. Impact of fluid therapy on apoptosis and organ injury during haemorrhagic shock in an oxygen-debt-controlled pig model. Eur J Trauma Emerg Surg 2013; 39:405–414.
23. Bambakidis T, Dekker SE, Sillesen M, Liu B, Johnson CN, Jin G, de Vries HE, Li Y, Alam HB. Resuscitation with valproic acid alters inflammatory genes in a porcine model of combined traumatic brain injury and hemorrhagic shock. J Neurotrauma 2016; 33:1514–1521.
24. Shein SL, Shellington DK, Exo JL, Jackson TC, Wisniewski SR, Jackson EK, Vagni VA, Bayir H, Clark RS, Dixon CE, et al. Hemorrhagic shock shifts the serum cytokine profile from pro- to anti-inflammatory after experimental traumatic brain injury in mice. J Neurotrauma 2014; 31:1386–1395.
25. Hildebrand F, Andruszkow H, Huber-Lang M, Pape H-C, van Griensven M. Combined hemorrhage/trauma models in pigs-current state and future perspectives. Shock 2013; 40:247–273.
26. Wong YC, Lai YY, Tan MH, Tan CS, Wu J, Zeng LZJ, Lu J, Moochhala S. Potential biomarker panel for preditingorgan dysfunction and acute coagulopathy in a polytrauma porcine model. Shock 2015; 43:157–165.
27. Hoque N, Liu X, Chakkarapani E, Thoresen M. Minimal systemic hypothermia combined with selective head cooling evaluated in a pig model of hypoxia-ischemia. Pediatr Res 2015; 77:674–680.
28. Rice J, Philbin N, Handrigan M, Hall C, McGwin G, Ahlers S, Pearce LB, Arnaud F, McCarron R, Freilich D. Vasoactivity of bovine polymerized hemoglobin (HBOC-201) in swine with traumatic hemorrhagic shock with and without brain injury. J Trauma 2006; 61:1085–1099.
29. Pfortmueller CA, Drexel C, Krähenmann-Müller S, Leichtle AB, Fiedler GM, Lindner G, Exadaktylos AK. S-100 B concentrations are a predictor of decreased survival in patients with major trauma, independently of head injury. PLoS One 2016; 11:e0152822.
30. Cavus E, Meybohm P, Doerges V, Hoecker J, Betz M, Hanss R, Steinfath M, Bein B. Effects of cerebral hypoperfusion on bispectral index: a randomised, controlled animal experiment during haemorrhagic shock. Resuscitation 2010; 81:1183–1189.
31. Chien JC, Jeng MJ, Soong WJ, Hwang B. Effects of fluid resuscitation on cerebral tissue oxygenation changes in a piglet model of hemorrhagic shock. J Chin Med Assoc 2011; 74:448–454.
32. Purins K, Lewén A, Hillered L, Howells T, Enblad P. Brain tissue oxygenation and cerebral metabolic patterns in focal and diffuse traumatic brain injury. Front Neurol 2014; 5:64.
33. Orakcioglu B, Sakowitz OW, Neumann JO, Kentar MM, Unterberg A, Kiening KL. Evaluation of a novel brain tissue oxygenation probe in an experimental swine model. Neurosurgery 2010; 67:1716–1722.
34. Dengl M, Jaeger M, Renner C, Meixensberger J. Comparing brain tissue oxygen measurements and derived autoregulation parameters from different probes (Licox vs. Raumedic). Acta Neurochir Suppl 2012; 114:165–168.
35. Yokobori S, Hosein K, Burks S, Sharma I, Gajavelli S, Bullock R. Biomarkers for the clinical differential diagnosis in traumatic brain injury--a systematic review. CNS Neurosci Ther 2013; 19:556–565.
36. Krohn M, Dreßler J, Bauer M, Schober K, Franke H, Ondruschka B. Immunohistochemical investigation of S100 and NSE in cases of traumatic brain injury and its application for survival time determination. J Neurotrauma 2015; 32:430–440.
37. Yao B, Zhang LN, Ai YH, Liu ZY, Huang L. Serum S100β is a better biomarker than neuron-specific enolase for sepsis-associated encephalopathy and determining its prognosis: a prospective and observational study. Neurochem Res 2014; 39:1263–1269.
38. Pelinka LE, Jafarmadar M, Redl H, Bahrami S. Neuron-specific-enolase is increased in plasma after hemorrhagic shock and after bilateral femur fracture without traumatic brain injury in the rat. Shock 2004; 22:88–91.
39. Pelinka LE, Bahrami S, Szalay L, Umar F, Redl H. Hemorrhagic shock induces an S 100 B increase associated with shock severity. Shock 2003; 19:422–426.
40. Ohrt-Nissen S, Friis-Hansen L, Dahl B, Stensballe J, Romner B, Rasmussen LS. How does extracerebral trauma affect the clinical value of S100B measurements? Emerg Med J 2011; 28:941–944.
41. Routsi C, Stamataki E, Nanas S, Psachoulia C, Stathopoulos A, Koroneos A, Zervou M, Jullien G, Roussos C. Increased levels of serum S100B protein in critically ill patients without brain injury. Shock 2006; 26:20–24.
42. Tang Y, Le W. Differential roles of M1 and M2 microglia in neurodegenerative diseases. Mol Neurobiol 2016; 53:1181–1194.
43. Seo J-W, Kim J-H, Kim J-H, Seo M, Han H, Park J, Suk K. Time-dependent effects of hypothermia on microglial activation and migration. J Neuroinflammation 2012; 9:164.
44. Lee JH, Wei ZZ, Cao W, Won S, Gu X, Winter M, Dix TA, Wei L, Yu SP. Regulation of therapeutic hypothermia on inflammatory cytokines, microglia polarization, migration and functional recovery after ischemic stroke in mice. Neurobiol Dis 2016; 96:248–260.
45. Wenzel V, Lindner KH, Krismer AC, Voelckel WG, Schocke MF, Hund W, Witkiewicz M, Miller EA, Klima G, Wissel J, et al. Survival with full neurologic recovery and no cerebral pathology after prolonged cardiopulmonary resuscitation with vasopressin in pigs. J Am Coll Cardiol 2000; 35:527–533.
46. Xanthos T, Bassiakou E, Koudouna E, Rokas G, Goulas S, Dontas I, Kouskouni E, Perrea D, Papadimitriou L. Combination pharmacotherapy in the treatment of experimental cardiac arrest. Am J Emerg Med 2009; 27:651–659.

Biomarker; cerebral monitoring; hypothermia; microglia; porcine trauma model

Copyright © 2017 by the Shock Society