A porcine polytrauma model with two different degrees of hemorrhagic shock: outcome related to trauma within the first 48 h

There are several animal models for trauma and organ failure research, but a model mimicking real polytrauma, including simultaneous organ and musculoskeletal injuries, has not yet been described. Recently we described an animal experimental model of multi-system injury, consisting of blunt chest trauma, penetrating abdominal trauma and pressure-controlled hemorrhagic shock in pigs with an observational period of 16 h [1]. This model lacked a musculoskeletal injury component, and the study period was short by clinical standards. Most models lack information on the latter, i.e., 15-h-plus, post-traumatic phase [2]. Indeed, most studies of new early phase treatments do not adequately reflect the clinical situation in humans.

The aim of the present experiment was to establish a porcine model of hemorrhagic shock, multi-organ injury and musculoskeletal trauma with a prolonged observational period of 48 h. To standardize treatment during the post-trauma period, management of temperature, cardiovascular complications, volume therapy, and the prevention and treatment of ventilator or trauma-associated lung complications were defined prior to the study. Moreover, we compared two different severities of hemorrhagic shock to identify the most appropriate trauma intensity with respect to the occurrence of trauma-related organ failure.

We have devised a porcine polytrauma model that reflects the mortality and organ failure of polytrauma patients during shock and intensive care. We included two different severities of hemorrhagic shock. The severity of the trauma and its effects appeared to be reproducible.

Pre-clinical models are needed to develop new therapies for trauma. These are mostly porcine because of the similarity to humans [8‐10]. A discussion of current models took place at a Military Medicine Workshop on Animal Models in Hemorrhage and Resuscitation Research [11], in particular the question of a reproducible trauma/hemorrhage model. The key points are as follows: (a) the need for volume- or pressure-controlled models that can allow uncontrolled bleeding, (b) surgical manipulation, e.g., soft tissue injury coincident with hemorrhage as in clinical situations, (c) significant soft tissue injury to better represent the post-injury inflammatory state, (d) a controllable degree of hypotension to lead to poor outcomes and (e) a model simulating battlefield trauma rather than one designed to meet a specific scientific goal [2, 11]. Furthermore, it was also agreed that reproducible trauma and hemorrhage are key factors and that the duration of hemorrhagic shock has to mimic the clinical situation. The pre-clinical period is 60–80 min in Germany [12].

In 2012, we presented a porcine multi-system injury model of combined chest and abdominal trauma with hemorrhagic shock, including an observation period of 15.5 h [1]. It was shown to be reproducible and fulfilled many of the above criteria. Because of the need for studies including a longer observation period under intensive care conditions [13‐15], we included a fracture of the right hind leg, and extended the study period to 48.5 h under intensive care conditions. Two grades of hemorrhage were achieved during our investigation. As no previous studies used a similar time period combined with simultaneous trauma of similar severity, we gradually tested the limits of feasibility. Since we already knew, based on previous investigation that it worked at least for 15 h, we started with a hemorrhage of 45 % of total blood volume for 90 min and extended this hemorrhage to a 50 % loss of total blood volume and 120 min of shock, as described above. In addition to survival rates and resuscitation frequencies, we used lactate and BE during the trauma period as outcome parameters for trauma severity. In the early trauma period metabolic acidosis is accompanied by increased plasma lactate, reflecting inadequate tissue oxygen delivery [16]. Previous studies assessed the physiological plasma lactate concentration in Landrace/Pietrain crossbred animals at 1.1 mmol/L [17]. These correlate to the plasma lactate concentrations in group C, suggesting that the increased lactate levels in group L/H realistically represent severe trauma. There is a strong correlation between early serum lactate and mortality in trauma patients [18]. Lactate values similar to ours, ranging from about 2.5–4.0 mg/dl, were associated with a mortality of 6.4 %. Our mortality was similar, although data have to be carefully interpreted because of the inter-species differences. Both trauma groups showed significantly decreased BE during the whole trauma and reperfusion period, suggesting a significant degree of metabolic acidosis. Animals with a BE less than −8 died within 3 h. It is described, that base excess less than −6 during the first 24 h of admission after trauma is associated with a higher risk of massive transfusion, coagulopathy, ARDS and mortality [16]. Accordingly, return towards baseline values obtained by the sham group after 10–12 h was interpreted as a result of successfully performed resuscitation in trauma groups.

Oxygenation during the acute trauma period did not differ significantly between groups, perhaps because we adjusted CO₂ levels to about 45 ± 5 mm/Hg during observation period by regulating inspiration rate and pressure.

Some independent risk factors for the development of ARDS after blunt trauma include an injury severity score (ISS) >25, the presence of pulmonary contusion, a large transfusion requirement, hypotension on admission, and age >65 years [19]. Other studies have shown specific injury patterns, such as long bone fracture, to be independently associated [20]. Post-traumatic ARDS may occur early (<48 h) or late (>48 h). Early development is also associated with presentation in hemorrhagic shock; our trauma model provides the majority of these risk factors. Three animals of each trauma group showed some degree of respiratory impairment, but no ARDS. There were no differences in lung function between groups L and H.

Porcine hemorrhage experimental models have focused on volume-controlled [18, 21, 22] or pressure-controlled [22‐24] blood loss; however, both have their disadvantages. In a volume-controlled investigation, the degree of hypotension depends on many factors, at least in terms of the individual animal’s response. Most volume-controlled studies deal with a loss of 30–50 % of TBV, with a mortality rate of up to 20 % [25‐28]. In most pressure-controlled studies, an MAP between 30 and 45 mmHg was generally maintained for 30–45 min. This artificial control of blood pressure does not mimic the clinical situation, and the volume of exsanguination can be very variable. Therefore, a combination of volume and pressure control may best mimic the clinical situation. Induction of a well-defined and reproducible trauma that includes significant hemorrhage is a significant factor in the establishment of a relevant trauma model. No published data deal with a withdrawal greater than 45 % of blood volume and survival beyond 24 h without retransfusion of shed blood in simple porcine hemorrhage models. Furthermore, no model of combined lung and liver hemorrhage dealing with a withdrawal greater than 45 % and survivals beyond 15.5 h is described. Models with greater blood loss did not provide observation periods longer than 6 h. In a further study, we proved that withdrawal of 45 % of TBV combined with liver laceration and lung contusion is safe and reproducible for 15.5 h [1]. Since we added a lower leg fracture and aimed to triple the investigation period, we first decided to keep the hemorrhage and the shock period similar to that of a previous study. This worked reliably for the first 15 pigs. These pigs were proven to react very sensitively, showing arrhythmia and blood loss of more than 45 % TBV or an MAP less than 25 mmHg. We slowly extended the borders of hemorrhage to 50 % TBV or an MAP 25 ± 5 mm/hg. According to cross-sectional guidelines for the transfusion of red blood cells based on clinical observation and risk factors, a hematocrit value of approximately 15 % (hemoglobin concentration 5.0–4.5 g/dL = 3.1–2.8 mmol/L) has to be assumed as the critical threshold value for RBC substitution. Blood loss greater than 50 % would have surpassed the recommended levels, indicated by the minimal Hb of 4.6 g/dL in group L. In addition, an MAP falling below 20 mmHg for some minutes was shown to promote arrhythmia and frequency of resuscitation. Therefore, we decided to extend the duration of hemorrhagic shock from 90 to 120 min to additionally aggravate ischemic injury.

As previously mentioned, animal models including withdrawal of blood until a certain endpoint for a defined period of time, followed by retransfusion of shed blood in the subsequent observation period have been described [29]. Cross-sectional recommendations for the transfusion of red blood cells define a hemoglobin concentration of 4.5–5 g/dl as the threshold value for transfusion of RBC in young and healthy subjects. Additionally, a recent publication dealing with the recommendations for red blood cell transfusion in septic patients could not prove a benefit in the transfusion of red blood cells beyond a hemoglobin value of 7 g/dl in patients experiencing septic shock. Holst and colleagues provide definitive evidence that a restrictive approach to blood transfusion not only reduced blood use by half, but also did not cause further harm to 998 critically ill patients [30]. In the presented polytrauma model, numerous factors interfere with immune response, inflammation parameters and the circulatory system. Among these are pre-existing medical conditions, iatrogenic trauma due to invasive monitoring, ventilation, and environmental factors. To improve data quality and traceability and to reduce possible interactions, we abstained from RBC transfusions. Furthermore, our pre-existing model which describes similar blood loss without severe complications due to hypovolemia, confirms our approach.

In our model this extended controlled component of bleeding and reperfusion strategy worked reliably.

Liver injury is often used to produce uncontrolled bleeding in animal models. Many combine hemorrhage and grade II–IV liver injuries, but most have a very short shock period of about 5–30 min and a maximum observation time of 7 h [31‐34]. Several studies use fluid resuscitation including blood transfusion [35, 36] or surgical treatment, e.g., suturing of liver injury or the use of haemostyptic agents [37]. We decided to produce a grade II (Moore et al.) liver injury to be able to provide a 48.5-h observation period with liver trauma and impaired liver function. ALT was not elevated and there were no great differences between groups. A tripling of AST indicates severe damage; six animals of 13 in group L and 5 of 7 animals in group H showed such values at the end of study period. The AST/ALT ratio also suggests severe liver cell damage (Table 2).

However, according to creatinine clearance and urinary output, even the prolonged hemorrhagic shock in group H did not lead to clinically significant renal ischemic trauma.

We caused a proximal lower limb fracture because fractured extremities are present in more than 60 % of multiple-trauma patients [12]. This tissue trauma led to significant increases in CK and myoglobin. The differences lacked statistical significance at the late post-traumatic course, perhaps because of small group numbers and large standard deviations; shivering following anesthesia was frequent and may explain elevated CK levels in group C. The impact of muscle relaxation has not yet been evaluated.

Our study has several limitations. The animal model is limited in usefulness because of inter-species differences, and the animal welfare committee guidelines restrict sample size. Further, we were unable to achieve a great degree of post-traumatic lung injury. We thought of two ways to improve the lung injury method in advance. On the one hand, we thought of using a more widespread lung trauma at the beginning; bilateral hits were discussed, but this idea was ignored because we wished to have an uninjured lung for data comparison. On the other hand, we tried an aggravation of force by reducing the upper layer of plumb and using a more powerful munition for the captive bolt stunner. Following this, we observed a great amount of tension pneumothoraxes and cardiac arrhythmias due to concomitant heart contusions, resulting in cardiac arrest. Beyond this, a combined repetitive surfactant washout could be promising, but it has to be considered that it would provide a kind of artificial (and not trauma-induced) lung injury. According to the National Trauma Databank, incidence of trauma-related ARDS in trauma patients requiring mechanical ventilation for greater than 48 h is 6.5 %. Hence, to achieve a sufficiently large number of ARDS cases, increase not only in the observation period but also in the number of animals is needed.

As a further limitation, we conclude that any ischemic renal injury was too short to impair the outcome of our study, although changes in renal function may become significant later in the post-traumatic course. Additionally, we have taken into account that creatinine measurement is perhaps not appropriate for detecting early hypoxic-induced renal failure, with the consequence that more sensitive biomarkers for renal injury, such as TIMP-2, IGFBP-7 cystatin C, and NGAL have to be considered in further investigation.

Furthermore, we decided to use Ringer’s acetate solution for resuscitation. Some trials show that there is evidence that balanced salt solutions are beneficial towards NS as a means of preventing development of hyperchloremic metabolic acidosis. Until now there is no high-quality randomized controlled trial comparing different balanced crystalloids to each other and, therefore, no consensus exists on a single preferred solution [38, 39]. Nevertheless, even balanced Ringer’s acetate solution contains some chloride that can induce a metabolic acidosis.

At last, the trauma groups were too close to each other (in terms of blood loss and duration of shock period) to achieve significant differences. Nevertheless, a higher blood loss of about 60 % with a comparable shock period led to a need for vasopressors and a higher frequency of CPR, which led to a high mortality in previous feasibility tests. As we aimed to obtain 46 h of observation following trauma, a balance between mortality and reaching a sufficient severity of trauma was hard to find; this was also the case because of differences in shock tolerance between individual animals.

Last but not least it can be assumed that a further extension of the observation period may have revealed greater severity of trauma.

Our model of polytrauma appears to reflect the mortality and organ failure of polytrauma patients during shock and the intensive care period. It was possible to manage fluids, ventilation and complications according to our goals. Further aggravation of hemorrhagic shock led to increased organ injury and resuscitation rates, thus the degree of trauma must be chosen to reflect the aims of the investigation. This experimental protocol could be of use for the assessment of therapeutic approaches during the post-traumatic period.