Swine Model
The animal model satisfactorily produced a severe state of supply-dependent hemorrhagic shock by oxygen transport and metabolic markers which became significantly abnormal when OD = 80 ml/kg. However, the severe shock state combined with injury produced only an isolated reduction in MA without overt coagulopathy by standard definitions.
One reason for the lack of overt coagulopathy during shock may be our limited level of tissue injury. We calculated the total ISS = 13, which is less than that identified by Brohi et al, as being compatible with early coagulopathy [
1]. However, the goal of the study was to isolate and examine the associations between tissue oxygen perfusion parameters and clot strength rather than to produce a significant overall coagulopathy. Increasing extremity injury would not have increased the ISS in our model per se. Thoracic injury would have likely confounded our oxygen debt measurements by impairing pulmonary oxygen exchange. Adding abdominal solid organ injury would have detracted from our ability to standardize shock severity due to uncontrolled hemorrhage. Inducing traumatic brain injury would have induced specific changes in clotting function, making interpretation of our results difficult. For these reasons, we limited ISS in order to better examine the specific associations between oxygen transport variables and TEG-MA.
Hypothermia was also prevented and plasma dilution was limited to that occurring from transcapillary refill and small aliquots of isotonic crystalloid during the hypotensive period. The 9-10% reduction noted in Hct and cell counts likely did not play a significant role in the measured significant decrease in MA from baseline. Small volume dilution of blood (less than 10% changes in Hct) with isotonic crystalloid has been shown
in vitro to instead produce procoagulant properties to the blood and increase MA in healthy humans [
28]. Overall, the animal model achieved the stated goal by providing an experimental platform in which significant changes in both oxygen transport and clot strength were achieved in the setting of traumatic shock, but should not be interpreted as producing an overt coagulopathy by current definitions.
Porcine models of coagulopathy in the setting of trauma are popular and favored because they use a large mammalian species that shares gross cardiovascular physiology with humans. Swine are amenable to precise monitoring while providing adequate sample volumes for viscoelastic testing. A review of experimental traumatic coagulopathy models found that of 33 models deemed appropriate for review, 17 were porcine [
29]. However, significant differences exist in the type of coagulation changes produced in swine in response to hemorrhage and these differences are important to consider when interpreting our results.
Standard tests of blood coagulation function typically demonstrate pro-coagulant activity in swine compared to humans, and immunologic methods are not generally comparable as illustrated in a comparison of 22 commercial assays in healthy pigs and humans by Munster et al [
30]. The authors found that PT was approximately equal between species while aPTT was shorter in pigs suggesting enhanced intrinsic coagulation cascade activity. In addition, plasma tissue factor levels were 4-fold higher in pigs, which may have special relevance in the setting of trauma since coagulopathic trauma patients have demonstrated increased plasma tissue factor activity [
31]. Using ROTEM, comparisons of porcine and human clot formation also suggest a hypercoagulable state in pigs relative to humans. Pigs tend to demonstrate shorter clot formation times, faster clot buildup, and increased maximal clot firmness with similar clot lysis profiles to humans [
26,
32]. TEG parameters correlate with ROTEM in porcine blood, with TEG demonstrating higher values for clotting angle and clot strength (MA vs. MCF) [
33]. Therefore, the native hypercoagulable state of porcine blood relative to humans may require that a greater degree of shock or increased injury severity be incurred in order to accurately reproduce the early coagulation changes seen in humans. This species difference may have contributed to our lack of overt coagulopathy during shock.
To date, no porcine model has accurately reproduced the initial hemostatic changes observed in human traumatic coagulopathy. Sapsford et al, observed no change in PT after 40% hemorrhage compared to baseline measurements using an aortic tear model [
34]. Martini et al, observed no difference in PT, R, K, Angle, and a significant, but limited, reduction in MA (approx 67 to 63 mm) measured 4 hours after 35% hemorrhage combined with crystalloid resuscitation of 3 times shed blood volume [
17,
35]. Via et al, reported in their sham resuscitation group no change in PT, PTT, or fibrinogen, and a reduction in TEG-MA from 74-71 mm at one hour of shock after a 40% blood volume hemorrhage [
36]. Using ROTEM, Haas et al, reported that clotting time and clot formation time were essentially unchanged and maximal clot firmness was reduced, but not necessarily abnormal, after a 60% blood volume hemorrhage [
37]. Cho et al, reported a multi-institute porcine model that, similar to ours, added femur fracture by captive-bolt pistol [
38]. When compared to our model, they achieved a similar injury profile, hemorrhage volume, and a similar level of lactate accumulation during shock. Their coagulation parameters measured at "End of Shock" after injury and hemorrhage, but prior to fluid resuscitation, are most likely comparable to our OD = 80 ml/kg measurements. At this particular time point, they found an INR baseline/shock ratio of only 1.1 and TEG parameters demonstrating a trend towards hypercoagulability (R, K, and Angle) with an isolated decrease in MA that was not outside the baseline reference range [
38]. Overall, the available viscoelastic porcine data demonstrates a tendency for isolated and limited decrease in clot strength as the initial response to hemorrhage. This result agrees with our own and is somewhat dissimilar to human observational studies which typically demonstrate a mixed impairment of prolonged clot onset times and decreased clot strength. This initial response may be species-specific. Alternatively, current porcine models may lack the appropriate criteria (combined shock and injury severity) to induce very early coagulopathy similar to that seen in humans. Our model is also limited in this respect since we achieved only and ISS = 13. Therefore, our results, while consistent with other porcine models, may not be directly comparable to traumatic coagulopathy observed in human studies.
Oxygen Transport and Clot Strength
Forward variable stepwise selection revealed that ScvO2, fibrinogen, Hct, and platelet count were important predictors of clot strength in this animal model. There was also evidence for an interaction between ScvO2 and platelet count in determining Delta MA during shock suggesting a specific role for platelets. Each selected linear regression model was highly predictive of both the value of MA during shock and Delta MA from baseline. The lack of a direct association between VO2 and clot strength and the importance of ScvO2 as the only significant oxygen transport associated with MA was interesting and surprising. This finding was even more surprising when considering that BE and lactate, the current metabolic markers used clinically to define tissue hypoperfusion, shared no association with clot strength in our animal model. Lactate did correlate with fibrinogen concentration during shock, but was not directly associated with MA. Therefore, it is possible that fibrinogen may have confounded an underlying association between lactate and clot strength. Alternatively, another physiologic variable (such as acidosis) mediates this relationship, but was not sufficiently pronounced in our model.
The reason why ScvO
2 was more strongly associated with clot strength when compared to other direct markers of oxygen transport or tissue hypoperfusion remains unclear. One explanation is that lactate produced in hypoperfused tissues may not have reached the central circulation by "wash out" prior to reperfusion, thus lactate may be less accurate than ScvO
2 in terms of hypoperfusion prior to fluid resuscitation. Among hemodynamic and oxygen transport measurements, ScvO2 has been found by Scalea et al, to be the best predictor of acute blood loss in experimental trauma models [
43]. The authors suggest that this sensitivity is a result of the ability of ScvO2 to reflect early increasing oxygen extraction at the blood/tissue interface in response to hemorrhage before gross hemodynamic measurements become abnormal.
In our study, the same sensitivity of ScvO2 to early changes in oxygen extraction may potentially explain its strong association with clot strength via compensatory endothelial activation in response to hypoxia. The observed fall in ScvO
2 and VO
2 with a concurrent increase in lactate confirms that oxygen delivery to the tissues was reduced below critical levels, despite maximal oxygen extraction. In addition, the disproportionately large fall in ScvO
2 from baseline levels (reduced 72%) when compared to VO2 (reduced 19%) suggests that blood oxygen extraction was actively enhanced at the blood/endothelial interface during shock. Therefore, we speculate that ScvO
2 and clot strength may be associated via activation of the endothelium as part of the local endothelial response to hypoxic conditions [
44]. While we did not directly measure biomarkers of endothelial activation, further evidence for a link between ScvO
2, protein C, and endothelial activation was recently reported by Trecziak et al. in critically ill septic patients. The authors used ScvO
2 to measure hypoxia and its effect on coagulation measurements and found that a subgroup of patients with both abnormally low ScvO
2 plus hypotension demonstrated changes in protein C, thrombomodulin, and increased endothelial activation by E-selectin expression [
45]. Therefore, our findings taken in this context may indirectly support the mechanism put forth by Brohi et al., who described a critical role for endothelial activation of protein C in the pathophysiology of trauma-induced coagulopathy [
2]. Future research on this topic should seek to include biomarkers of endothelial activation when examining associations between tissue hypoxia/hypoperfusion and clot formation.
Limitations
We acknowledge that there are distinct limitations to this study. As discussed, the relevance of the swine model to human subjects is concerning due to the native differences between porcine and human coagulation function. In addition, we calculated the coefficient of variation (CV) for swine MA measured at baseline in the study of Cho et al., and found it to range from 12-20% across centers [
38]. Our 5% change in MA from baseline to shock is well within this range, further limiting our results. In addition, tissue injury was limited and the model itself achieved only a mild reduction in clot strength without overt coagulopathy. We also did not strictly standardize the timing of TEG test performance, possibly adding variability to our results. However, when taken in the context of other similar swine models of hemorrhage, the changes in clot strength in our model were quite similar to those described by other investigators when measured during shock and prior to fluid resuscitation.
We intended to isolate the association between oxygen metabolism and clot strength so to examine the inherent relationships in detail. As a result, we can only speculate on the associations found between independent and dependent variables and cannot make any causative or mechanistic conclusions from the data. Nevertheless, the associations found suggest important areas for further focused study concerning the early detection and monitoring of hemostasis during trauma.