Activated Protein C Drives the Hyperfibrinolysis of Acute Traumatic Coagulopathy

TRAUMA is a major global public health issue, causing nearly 6 million deaths worldwide each year.¹  Hemorrhage is responsible for approximately half of these deaths and is the leading cause of preventable death after injury.^2,3  Bleeding is exacerbated by acute traumatic coagulopathy (ATC), which is present immediately on hospital admission in up to 25% of severely injured patients.^4–6  ATC is associated with a four-fold increase in mortality, more severe organ injury, and higher transfusion requirements.^5–7  Resuscitation strategies targeted at ATC have shown potential to dramatically reduce death from injury, possibly mediated by early reversal of coagulopathy.⁸  However, controversy exists over the precise mechanism of ATC and the balance between inhibition of procoagulant pathways (anticoagulation or consumption), fibrinogen loss, and fibrinolysis.^9–13  Failure to define the pathophysiology of ATC has prevented identification of the optimal therapeutic intervention for early traumatic coagulopathy and remains an important research priority for trauma care.

ATC is a clinical syndrome¹⁴  evident soon after injury,^4,5,15  characterized predominantly by functional reductions in clot generation and clot strength with only minor prolongations in clotting times (CTs).¹⁶  Biomarker^7,17,18  and animal studies¹⁹  have highlighted the pathogenic contributions of shock and tissue injury in ATC,^17,20  suggesting an endogenous process resulting in systemic anticoagulation, hyperfibrinolysis,^7,10,21,22  and fibrinogen depletion.^23,24  It is postulated that thrombin is switched from a pro- to anticoagulation function with diversion from fibrin generation toward increased production of activated protein C (aPC)^7,17,25  and subsequent cleavage of factors Va (FVa) and VIIIa (FVIIIa). In excess, aPC consumes plasminogen activator inhibitor-1 (PAI-1) and releases fibrinolysis from inhibitory control with consequent rise in tissue plasminogen activator (tPA) and plasmin.^7,26  Tissue injury itself is associated with increased thrombin generation,²⁷  release of tPA from the endothelium, and activation of the fibrinolytic system.²⁸  Some hypothesize that these early derangements after trauma hemorrhage represent disseminated intravascular coagulation (DIC) with a fibrinolytic phenotype¹³  and propose that the early fibrinogen loss is a direct result of fibrinogenolysis.¹⁰  With a lack of mechanistic confirmation, the interplay between thrombin generation, anticoagulation, fibrinogen availability, and hyperfibrinolysis in ATC remains unclear. Shock and tissue injury both independently promote fibrinolysis, but in combination, through the release of aPC, could, in theory, provide the stimulus for massive systemic fibrinolytic activity and potentially unify the apparent opposing theories of ATC.

The overall objective of this combined prospective human cohort study of trauma patients and experimental model of trauma hemorrhage (TH) was to determine the role of aPC in ATC and specifically quantify the balance between inhibition of procoagulant pathways (anticoagulation), fibrinogen loss, and fibrinolysis. This study builds on our previous evaluation of the downstream elements of fibrinolysis in ATC,²¹  with emphasis on the mechanistic study rather than clinical outcome, in a detailed examination of functional coagulation and biomarkers of both procoagulant and fibrinolytic pathways. The first aim was to functionally characterize ATC with measurement of aPC levels in trauma patients to determine the relationship between aPC, procoagulant factors, thrombin generation, and fibrinolysis. Second, we wished to characterize the relationship of plasma aPC levels with markers and mediators of fibrinolytic activation in humans. Third, we aimed to mechanistically confirm the role of aPC in a murine model of TH to understand the effects of genetic modulation of the protein C (PC) pathway on clot formation, fibrinolysis, and fibrinogenolysis. Finally, we report outcomes for trauma patients with ATC stratified by aPC levels soon after injury.

This is a single-center, prospective, cohort study of trauma patients presenting directly to a level 1 trauma center (January 2007 to June 2009). The study is part of the Activation of Coagulation and Inflammation in Trauma (ACIT) research program. We have previously published data from the same cohort on functional measures of coagulation¹⁶  and fibrinolysis.²¹  ACIT is approved by the U.K. National Research Ethics Committee (East London, London, United Kingdom). Patient consent was in accordance with the Mental Capacity Act (United Kingdom; 2005) for inclusion of incapacitated patients into emergency medical research using a professional consultee (independent physician).

All adult trauma patients (more than 15 yr) meeting local criteria for full trauma team activation were eligible for enrollment and recruited when research personnel were present (08:00 to 22:00 daily). ACIT inclusion/exclusion criteria, data collection, blood sampling, assays, and outcome measures have been previously reported.²¹  A 20-ml research sample of blood was drawn from either the femoral vein or antecubital fossa, and the standard trauma laboratory tests were performed within 20 min of arrival in the ED. Blood for rotational thromboelastometry (ROTEM) analysis was drawn into a 2.7-ml citrated vacutainer (0.109 M buffered sodium citrate, 3.2%; Becton Dickinson, United Kingdom). Samples for prothrombin determination were collected in 4.5-ml glass vacutainers (0.109 M buffered sodium citrate, 3.2%; Becton Dickinson), 9:1 (v/v). The sample for hemostatic assays and thrombin generation was placed in a citrated tube and spun down within 2 h of blood draw. The sample was first spun at 1,750g for 10 min; the supernatant was then extracted and respun at 1,750g for a further 10 min. The extracted plasma was stored in aliquots at −80°C. Arterial blood analysis for base deficit (BD) was performed simultaneously with the research sample collection. Assays for coagulation factors were complete in 99 to 100% of patients. Plasma samples for biomarkers requiring enzyme-linked immunosorbant assay (ELISA) analysis were incomplete in 3% of cases except for aPC that could not be quantified in 7% of patients.

Samples were analyzed at the conclusion of the study with an automated analyzer (Sysmex CA-CS2100i System; Siemens AG, Germany) to measure coagulation factor activity (normal range): II (78 to 117), V (66 to 114), VII (50 to 150) VIII (52 to 153), IX (58 to 138), X (50 to 150), XI (50 to 150), XIII (70 to 140), von Willebrand factor (vWF; 50 to 160), PC (75 to 134), and antithrombin (80 to 130). ELISAs were used to quantify tPA (Asserachrom tPA; Diagnostica Stago, France; normal range, 2 to 12 ng/ml), prothrombin fragments 1 + 2 (Enzygnost F1 + 2 monoclonal; Siemens Healthcare Diagnostics, Germany; normal range, 69 to 229 pmol/l), plasmin–α2-antiplasmin complex (plasminantiplasmin; DRG Plasmin-Antiplasmin micro, Germany; normal range, 120 to 700 µmol/l), and aPC (Surgical Research Laboratory, University of California, San Francisco General Hospital, San Francisco, California [1 to 3 ng/ml]). Prothrombin and Clauss fibrinogen were processed by the central hospital laboratory along with a standard full blood count. Prothrombin ratio (PTr) was calculated as observed prothrombin divided by mean normal prothrombin for the reagent used.

ATC was defined using ROTEM (EXTEM)—amplitude (of clot) 5 min (A5) after CT based on a previous study demonstrating that low clot amplitude at 5 min and maximum clot firmness (MCF) were the viscoelastic hallmarks of ATC.¹⁶  Additional EXTEM parameters reported were EXTEM assays, clot formation time, α angle, and MCF. Prolongation of prothrombin is associated with coagulopathy clinically, and PTr greater than 1.2 was used for comparison between human and animal studies.^16,20  Hypoperfusion was defined as BD greater than 6 mEq/l.^29,30  FVIII:vWF ratio was calculated to reflect the differential inhibition by aPC. Blood for aPC analysis was collected in protease inhibitors (P100; Becton Dickinson) and measured using ELISA.³¹  PC circulates in plasma at 70 nM as the zymogen of the anticoagulant serine protease, aPC, which averages 40 pM (approximately 2.3 ± 0.2 ng/ml) in normal plasma.³²  Kaiser et al.³³  reported in vitro inhibition of clotting at 10 ng/ml aPC, and for convenience, we divided groups into four with the aPC less than 3 ng/ml representing the normal range. Thrombin generation was measured via calibrated automated thrombogram using standard protocol³⁴  in duplicate in a Fluoroskan Ascent^® reader (Thermo Labsystems, Finland; filters, 390-nm excitation and 460-nm emission). Thrombin generation curves and the area under the curve (endogenous thrombin potential [ETP]) were calculated using Thrombinoscope software (Thrombinoscope BV, The Netherlands).

Patients were followed up until hospital discharge or death. For mortality analysis, patients surviving to hospital discharge were assumed to be still alive. Outcome measures were recorded for 28-day ventilator-free days, blood transfusion requirements in the first 12 h, and length of critical care and overall hospital stay.

To mechanistically verify the clinical findings in humans, we analyzed coagulation profiles of experimental models of TH. Wild-type (WT) mice and transgenic thrombomodulin knockin (TMKI) mice with reduced capacity to activate PC were compared. In addition, we evaluated ATC in homozygous mice for factor V Leiden (FVL) that are resistant to aPC-mediated cleavage. A novel murine model for ATC was developed based on the experimental procedure previously reported in a rat model of TH.²⁰  All mice were cared for in accordance with the U.K. Home Office Guidance in the Operation of the Animals (Scientific Procedures) Act 1986. General anesthesia was induced and maintained with spontaneous respiration of isoflurane carried in medical air. Animals were placed in the supine position on a heated anesthesia platform, and body temperature was maintained at 37° ± 1°C by means of a rectal probe attached to a homeothermic blanket (Harvard Apparatus, United Kingdom). The left carotid artery and external jugular vein of all animals were catheterized with polyethylene tubing (Portex, United Kingdom) connected to a pressure transducer (Capto SP 844; AD Instruments, New Zealand) and syringe pump (PHD 22/2000; Harvard Apparatus). Fluid was continuously infused at a rate of 50 µl/h through the carotid catheter with supplementary flushes as required in order to maintain patency.

Mice in the trauma hemorrhage group received a 2-cm paramedian laparotomy, closed in one layer with 7-mm surgical clips (Harvard Apparatus) and bilateral midshaft closed tibia and fibula fractures. The experimental period commenced immediately after traumatic injury or 7 min after completion of carotid catheterization (the mean time taken to complete traumatization). Animals were then bled from an average mean arterial pressure of 91 mmHg to a target of 25 to 30 mmHg during a period of 60 min, and this pressure was maintained by further withdrawals as necessary.

Study randomization and blinding was ensured by using a predefined alternated assignment of animals to experimental groups and the use of identification codes with respect to the WT and the TMKI groups throughout the experimental intervention and data analysis (both with a C57/B6 phenotype). For FVL studies, the genetic background was revealed at the end of the study.

During the experimental period, the genetic background was unknown. V Leiden mice (Strain B6.129S2-F5tm2Dgi/J; Jackson Laboratory, USA) are knockout in a C57BL/6 background, expressing factor VR504Q, resistant to aPC proteolysis. When performing the data analysis, the background was revealed: 14 mice were WT, 15 were V Leiden heterozygous, and 15 were homozygous (mutant).

Homozygote TMPro knockin (TMKI) mice with 1000-fold reduced capacity to activate PC (Blood Research Institute of Wisconsin, Milwaukee, Wisconsin) bred from C57/B6 mice after targeted mutagenesis with a single amino acid substitution (Glu404Pro) for the thrombomodulin receptor. Twenty-six TMKI mice underwent laparotomy and TH to mean arterial pressure 25 to 30 mmHg for 60 min (TMKI TH). The same protocol was performed in 26 experimental controls (C57/B6 TH). Further mice of each genotype were anesthetized and monitored for 60 min to act as sham controls (anesthetized/monitored): TMKI sham (n = 12) and C57/B6 sham (n = 12).

Fifty microliter blood/saline was aspirated from the carotid catheter of all animals and discarded. Two hundred microliter aspirated into 1-ml syringe prefilled with 20 µl sodium citrate, 3.2% (Sigma, United Kingdom) to achieve 1:9 concentration. Lactate was measured (Accutrend; Roche, Switzerland), and the remainder was centrifuged at 3,500g for 15 min. Plasma supernatants were aspirated and immediately frozen in liquid nitrogen before storage at −80°C. Citrated blood was analyzed using pediatric ROTEM cups and ELISA for d-dimer (USCN Life Sciences Inc., China) and fibrinogen (Genway Biotech, USA).

Patients with ATC were compared with those presenting without ATC for clinical characteristics and coagulation profiles. The relationship between aPC, functional coagulation (ROTEM), and key coagulation factors was then evaluated. Next, key biomarkers of the fibrinolytic pathway were examined to determine their relationship with aPC and known control mechanisms, e.g., PAI-1 and tPA. Finally, associations between the admission aPC level and clinical outcomes were explored. In the experimental models, the effect of genetic modulation of the PC pathway was determined through analysis of aPC and functional parameters of coagulation (ROTEM). Finally, the role of the PC pathway in the murine model and impact of impaired thrombomodulin generation on fibrinolysis, fibrinogen, and mortality were examined.

The human study was based on the existing available data from a convenience sample. For the experimental models, sample size was calculated with statistical power of 90% and significance level α = 0.05 to detect a 25% difference in mortality from a baseline mortality rate of 20% in a murine model of trauma hemorrhage³⁵  and 20% difference in ROTEM parameters based on a previous animal study in ATC.²⁰  Normal quantile plots were used to test for normal distribution. Parametric data are expressed as mean ± 95% CIs. All nonparametric data are presented as median (interquartile range) and were analyzed using a two-tailed Mann–Whitney U test. Univariate linear regression was performed using the best-fit model for calculation of R² values. Two-group analysis was performed using a two-tailed, unequal variance Student’s t test. Multigroup analysis was performed using one-way ANOVA with Dunnett post hoc correction or two-way ANOVA with Bonferroni posttest between subjects for determining the interaction with elevated levels of aPC. Chi-square test was used for dichotomous data analysis. Survival analysis in the murine model was performed by deriving Kaplan–Meier plots and curves compared using the log-rank test (Mantel–Haenszel method). Significance level was set at P < 0.05, and all statistical analysis was performed using GraphPad Prism version 5.00 for Windows, GraphPad Software, USA.

There were 325 patients enrolled into the human study during the 19-month period. ROTEM sample analysis was incomplete in three patients, consent was not obtained in 15 cases, and there were seven retrospective exclusions, leaving 300 patients available for analysis. Median age was 33 (23 to 48) yr, 82% were males, and 21% presented after penetrating trauma. Patients had sustained significant trauma with a median injury severity score of 10 (4 to 25), and 41% were severely injured (injury severity score, more than 15). Average time from injury to blood sampling was 86 (69 to 112) min, 9 (5 to 14) min from arrival in the emergency department. On arrival, 15% of patients were severely shocked (BD, more than 6 mEq/l), 17% coagulopathic by ROTEM definition (A5, less than or equal to 35 mm), and 9% with PTr more than 1.2 (median prothrombin, 11.2 [10.8 to 11.9] s). Minimal intravenous fluid was administered before baseline sample collection was 0 (0 to 400) ml, and no patient received artificial colloid or vasoactive agents before venepuncture. Overall mortality was 8%. We retrospectively analyzed the deaths of 25 of 26 patients from electronic archives–seven patients died from exsanguination, six from multiple organ dysfunction (MODS) or sepsis, and 12 from traumatic brain injury. The mean admission aPC level in survivors was 1.8 ± 0.6 ng/ml. Patients who died early (less than 24 h) from hemorrhage had significantly higher aPC levels on admission compared to those who died after 24 h from MODS or sepsis (early hemorrhagic deaths: aPC, 30 ng/ml vs. MODS/sepsis deaths: aPC, 8 ng/ml; P < 0.05).

In this study, patients with ATC (A5, less than or equal to 35 mm) had aPC levels more than five times higher than those with normal coagulation (table 1). Elevated aPC was associated with a significant concentration-dependent decrease in ROTEM parameters of clot strength (fig. 1, A and B) and prolongation of PTr (aPC, less than 3 ng/ml: PTr, 1.0 [1.0–1.0] vs. aPC, more than 9 ng/ml: PTr, 1.3 [1.2 to 1.4]; P < 0.001). ATC has been shown to be associated with early PC depletion,⁷  and we now demonstrate that a reduction in PC activity is associated with high aPC (aPC, less than or equal to 3 ng/ml: PC activity, 91% vs. aPC, more than 9 ng/ml: PC activity, 65%; P < 0.001). Activated protein C was elevated only in the presence of shock (BD, more than 6 mmol: aPC, 9 [5 to 13] ng/ml vs. BD less than 3 mmol/l: aPC, 1 [0 to 1] ng/ml; P < 0.001).

Activated protein C is known to inactivate coagulation FV and FVIII and can result in systemic anticoagulation. Factor V has the greatest reduction in activity in ATC (average, 32% reduction; table 1). High aPC levels in trauma patients were associated with a 50% reduction in FV activity (fig. 1C). Eighty-five percent of patients with FV less than 50% had A5 less than or equal to 35 mm (P < 0.001), and progressive loss of FV was associated with a concentration-dependent reduction of ROTEM clot strength (Supplemental Digital Content 1A, https://links.lww.com/ALN/B329: figure showing the association of falling factor V level with reduced A5). Similarly, the FVIII:vWF ratio was significantly reduced as aPC levels increased, suggesting a relative inhibition of FVIIIa (fig. 1D). However, preservation of factor V activity did not fully protect against coagulopathy in trauma. Reduced factor V is not sufficient to inhibit procoagulant pathways as ETP is maintained regardless of aPC concentration (fig. 1E). Consequently, in trauma patients, aPC is not associated with any significant increase in CT (fig. 1F), and prolongation in PTr was only observed when 3× normal (more than 9 ng/ml). This suggests that the reduced FV/VIII levels associated with aPC cannot fully explain the observed functional coagulation changes in ATC. Together these data suggest that systemic aPC-induced anticoagulation is not the primary mechanism in ATC.

Levels of procoagulant factors that are not targets of aPC-mediated cleavage were preserved in ATC. FII, FVII, FIX, and FX activities were all maintained at more than 80% (Supplemental Digital Content 1B, https://links.lww.com/ALN/B329: figure showing aPC levels and FII, FVII, FIX, and FX activities), and thrombin generation was in fact two-fold higher in the ATC cohort (table 1). Widespread consumption of procoagulant factors and reduced thrombin-generating capacity do not appear to be major causes of ATC. Procoagulant factor activity (sampled in pH buffered assays) was minimally affected by the degree of metabolic acidosis with levels preserved above 80% at BD more than 6 mmol/l (Supplemental Digital Content 1C, https://links.lww.com/ALN/B329: figure showing coagulation factor levels stratified by BD). Hemodilution with intravenous fluids produced minimal effects on coagulation, and low hematocrit was not associated with any significant changes in clot strength (Supplemental Digital Content 1D, https://links.lww.com/ALN/B329: figure showing no change in A5 with respect to changing hematocrit). In patients receiving more than 1,000 ml fluid before sampling, procoagulant factors were all maintained above 80% with relative preservation of fibrinogen levels (0 ml: fibrinogen, 2.17 g/dl vs. more than 1,000 ml: fibrinogen, 1.82 g/dl; P < 0.05; Supplemental Digital Content 2, https://links.lww.com/ALN/B330: table showing coagulation factor levels stratified by the amount of fluid administered). Patients given intravenous fluids had only minor changes in clotting parameters, and hemodilution in isolation does not appear to be an important mechanism for ATC.

Average fibrinogen levels were significantly lower in the ATC cohort (A5, less than or equal to 35 mm; 1.35 [1.19 to 1.51] g/l vs. A5, more than 35 mm: 2.23 [1.53 to 2.93] g/l; P < 0.05), and this 40% reduction in fibrinogen levels was a greater reduction than that in any other procoagulant factor (table 1). Despite very high thrombin generation, fibrinogen levels were maintained above critical levels, while aPC was normal, suggesting that consumption is not the primary mechanism of fibrinogen loss (fig. 2A). High levels of aPC were associated with low fibrinogen levels (fig. 2B) and reduced functional fibrinogen activity (FIBTEM; fig. 2C). There was a trend toward fibrinogen reduction to critical levels (less than 1.5 g/dl) in association with high thrombin generation, only in the presence of elevated aPC (P = 0.08; fig. 2A). Currently, it is not possible to accurately quantify fibrinogenolysis independent of fibrinolysis; however, plasmin–antiplasmin levels were inversely correlated with fibrinogen (R² = 0.24; P < 0.01). In patients with low fibrinogen (less than 1.5 g/l), plasmin–antiplasmin levels were 15 times the upper limit of normal and significantly higher than those in patients with normal fibrinogen (fibrinogen, less than 1.5 g/l: plasmin–antiplasmin, 10,739 μg/l vs. fibrinogen, more than 1.5 g/l: plasmin–antiplasmin, 2,734 μg/l; P < 0.01).

High aPC was associated with increased fibrinolysis as indicated by elevated plasmin–antiplasmin levels (aPC, less than or equal to 3 ng/ml: plasmin–antiplasmin, 2,221 ± 392 μg/l vs. aPC, more than 9 ng/ml: plasmin–antiplasmin, 18,955 ± 3,861 μg/l; P < 0.001). This tPA-dependent process was completely inhibited at normal aPC levels but was massively activated when aPC was high (fig. 2D). Mechanistically, aPC is known to consume PAI-1 when in excess, and there was a trend toward reduced PAI-1 in trauma patients with high aPC (more than 9 ng/ml; fig. 2E). Fibrinolysis is known to coactivate with thrombin generation through its direct induction of tPA release. However, we were clearly able to show that, as with fibrinogenolysis, thrombin generation–associated fibrinolytic activity (d-dimer production) in ATC is significantly higher in the presence of elevated aPC levels (fig. 2F).

Patients with ATC had increased mortality: PTr more than 1.2 (52%) and A5 less than or equal to 35 mm (26%). Consistent with our previous study,⁷  high aPC on admission was strongly associated with increased all-cause mortality (Supplemental Digital Content 3A, https://links.lww.com/ALN/B331). Patients with aPC more than 9 ng/ml had a mortality of 68% compared to 2% for those with normal aPC levels (P < 0.001). There was a concentration-dependent relationship between aPC and transfusion requirements for both packed red blood cells (PRBCs) and fresh frozen plasma (Supplemental Digital Content 3B, https://links.lww.com/ALN/B331). Patients requiring a massive transfusion (10+ PRBC units) had significantly higher aPC levels compared with those receiving PRBC less than 10 units: aPC, 20 (15 to 25) ng/ml versus 3 (3–3) ng/ml (P < 0.001). High aPC was associated with longer hospital median length of stay (aPC, more than 9 ng/ml: 30 days vs. aPC, less than or equal to 3 ng/ml: 5 days; P < 0.001). Patients with elevated aPC on admission required significantly more days in the intensive care unit (Supplemental Digital Content 3C, https://links.lww.com/ALN/B331) and had fewer ventilator-free days at 28 days (aPC, more than 9 ng/ml: 2 days vs. aPC, less than or equal to 3 ng/ml: 5 days; P < 0.001). Increased aPC is associated with excess mortality and morbidity in ATC.

Trauma hemorrhage was associated with significantly higher aPC levels in WT mice compared to sham animals: 10.3 (7.8 to 12.7) ng/ml versus 1.5 (1.2 to 1.8) ng/ml (P < 0.001; fig. 3A). TMKI mice had a minor increase in aPC levels and were relatively protected from ATC after TH for 60 min. In comparison to WT, clot strength (MCF) was well preserved (fig. 3B and Supplemental Digital Content 4, https://links.lww.com/ALN/B332) with significantly less marked prolongations of prothrombin and activated partial thromboplastin time (Supplemental Digital Content 4, https://links.lww.com/ALN/B332). Homozygous mice for FVL when subjected to TH had similar prolongations in CT (fig. 3C) and clot generation (Supplemental Digital Content 3, https://links.lww.com/ALN/B331) compared with WT animals. In support of the findings in humans, there was only a minor prolongation in CT (20%) in WT mice and no change in TMKI animals subjected to TH (fig. 3D). As with WT animals, FVL mice after TH had significant delays in clot formation and reductions in clot strength albeit to a slightly lesser extent than WT (Supplemental Digital Content 5, https://links.lww.com/ALN/B333). In this experimental model, a reduction in endogenous capacity to activate PC was protective against ATC in mice while preserved factor V function did not fully prevent ATC.

Confirming the observation seen in humans of apparent aPC-related fibrinolytic activity, WT mice after 60 min of TH exhibited a three-fold increase in d-dimer levels, which was completely abolished in the TMKI variants (fig. 4A). In the murine model, fibrinogen was observed to fall by over a third after TH in WT mice (T0: 1.86 g/l vs. T60: 1.25 g/l; P < 0.05), but TMKI animals were protected from this fibrinogen depletion with only a nonsignificant reduction in fibrinogen after TH (fig. 4B). Mortality rates in TMKI mice were almost half that of WT mice subjected to TH (WT: 42.3% vs. TMKI: 23.1%; P < 0.05) with longer median survival times (fig. 4C).

In both clinical and experimental studies, we have demonstrated that aPC may be a unifying link in the current opposing hypotheses of early traumatic coagulopathy. Building on previously published data from our human cohort study,²¹  this work completes a detailed evaluation of the procoagulant and fibrinolytic systems in ATC, highlighting a potential central mechanism for aPC with mechanistic confirmation in an experimental model of TH. Patients with ATC had elevated aPC levels, fibrinolytic activity, and early depletion of fibrinogen yet with only minimal inhibition of procoagulant pathways (factor loss) and preserved thrombin generation. Fibrinogen was significantly reduced in association with elevated levels of aPC with no evidence to support a systemic consumptive process to explain this early depletion. ATC, in particular both fibrinolysis and fibrinogen depletion, was significantly attenuated after TH in transgenic mice with reduced capacity to activate PC. These results add further evidence to previous work highlighting the importance of the PC pathway^7,18,35–37  and, in a murine model, provide mechanistic confirmation of aPC with respect to fibrinogen loss and fibrinolysis in ATC.

In the early stages after injury, generation of aPC appears to be of greater importance than other classical mediators of traumatic coagulopathy (hemodilution, acidosis, and clotting factor consumption). Dilution of plasma alters the dynamics of thrombin generation by relative reductions in both pro- and anticoagulant factors.³⁸  Paradoxically, dilution actually increases thrombin generation until plasma proteins are reduced below 40% of normal.²⁷  Consistent with this observation, we have shown that relatively small volumes of intravenous fluids had some minimal effect on clot amplitude. However, in vitro measurements of clot strength in the presence of hemodilution may not be true representation of clot integrity in vivo. An increased proportion of plasma in the ROTEM cup produces an apparent increase in clot strength although hemodilution will simultaneously reduce the platelet count and clot strength. Platelet counts were well preserved above 150 despite hemodilution, but aPC levels were significantly higher in patients receiving large amounts of crystalloids. The effects of dilution may, therefore, be confounded by the degree of shock, which, we have shown, is associated with increased aPC production. APC was associated with selective depletion of FV and FVIII but with no reduction in the capacity to generate thrombin and hence with only minimal effects on PTr. These findings are consistent with a recent study demonstrating no difference in ETP between severely injured patients and healthy controls.³⁹  In FVL mice, resistant to aPC-mediated cleavage of factor V, ATC was still observed after TH; therefore, inhibition of procoagulant pathways (anticoagulation) cannot fully account for functional changes of hemostasis after major trauma. Thrombin generation and platelet count in patients with ATC were normal with no clear evidence of widespread consumption of clotting factors. This is supported by Johansson et al.¹¹  who in a study of 80 severely injured and shocked trauma patients were unable to identify any overt DIC according to International Society on Thrombosis and Haemostasis criteria. ATC is defined predominantly by diminished clot strength but preservation of procoagulant pathways and thrombin generation.

Procoagulant factors are maintained at near-normal levels and are unlikely to be responsible for the observed prolongation of prothrombin or reductions in A5 and MCF. The mechanism for diminished clot strength and increased CTs must, therefore, involve fibrinogen, consistent with other recent investigations of alternative mechanisms for early traumatic coagulopathy.^10,40,41  We have previously shown that fibrinogen reduction is a principle component of ATC^16,24  with both retrospective data^42,43  and a pilot randomized controlled trial of cryoprecipitate⁴⁴  demonstrating an association between early replacement and improved outcome. Fibrinogen levels were reduced significantly in the human study to levels below transfusion guidelines for replacement (less than 1.5 g/dl).⁴⁵  In the experimental model, TMKI mice were protected against significant fibrinogen depletion in contrast to WT mice, suggesting that an aPC-related process is responsible for early loss. Currently, the critical level for hemostatic function is unknown, and potential mechanisms for the loss of fibrinogen remain to be elucidated^24,45  although indirect evidence for fibrinogenolysis has been demonstrated in alternative models of TH.¹⁰  In this study, activation of the fibrinolytic system was marked and occurred in association with elevated aPC and appeared to be a tPA-related process with consumption of PAI-1. Genetic modulation of the PC pathway was protective in the TMKI mice with significant attenuation of fibrinolysis after TH. Plasmin can directly induce fibrinogenolysis,⁴⁶  and together these novel findings suggest that production of aPC early after TH may release both fibrinolysis and fibrinogenolysis from inhibitory control to give rise to systemic clot lysis and direct breakdown of fibrinogen.

High levels of aPC on admission were associated with increased mortality, longer hospital stay, and increased transfusion requirements. Seventy percent of patients who died early did so from exsanguination and had significantly higher aPC levels on admission compared to those who died later from MODS or sepsis. APC levels in these patients dying from MODS were three times normal, suggesting early activation of PC, with subacute depletion a risk factor for later sepsis and organ failure.¹⁷  Low PC levels have been reported several hours or days after injury,^36,37,47  and patients who develop ventilator-acquired pneumonia have persistently low plasma levels of PC in the immediate period after trauma.³⁶  PC depletion after sepsis is well reported^48,49  with thrombotic microvascular organ injury⁵⁰  and may contribute directly to cell dysfunction and death.⁵¹  In trauma, early depletion of PC would result in a procoagulant state with insufficient cytoprotective mechanisms and theoretically would predispose to septic complications and organ injury as reported by Cohen et al.³⁶ 

There are several limitations to this study. First, the endothelium and thrombocytes are fundamental to both coagulation and inflammation with the platelet membrane of central importance to clot assembly. We did not measure platelet function in this study, and our results are, therefore, based on plasma protein levels. At present, the role of platelets in ATC pathophysiology remains unknown. In light of the results from the Pragmatic, Randomized Optimal Platelet and Plasma Ratios study and potential outcome benefits of early platelet transfusion, endogenous function and aggregation capacity of transfused platelets should be a key focus of research. Second, our definition of ATC (A5, less than or equal to 35 mm) is based on previous work that functionally characterized it by reduced clot strength rather than prolonged CTs.¹⁶  Patients meeting this definition will, therefore, more likely have lower fibrinogen levels and possibly reduced platelet function and/or count. Third, we were unable to measure activated FV and FVIII to assess direct effects of aPC. However, increased loss of FVa and FVIIIa by aPC-mediated cleavage will produce reciprocal reductions in FV and FVIII through alterations in the pharmacokinetics of the enzymatic reaction. Fourth, it is not possible to quantify separately fibrinogenolysis secondary to lysis by plasmin or cleavage by thrombin, and therefore, we are only able to show by process of elimination, the likely mechanism of fibrinogen depletion. The coagulation response to prolonged shock, continued hemorrhage, and the effects of transfusion form part of an ongoing study. In the early phase of the ACIT study that to date has recruited over 2,500 patients internationally, we did not capture the mode of death although this is now embedded in the study protocol. Finally, the human cohort study was completed more than 5 yr ago during which time the fields of trauma-induced coagulopathy and transfusion science have advanced greatly. The robust and early blood sampling in this study does, however, ensure validity of the study objectives to examine the early endogenous changes in coagulation.

The findings of this study have important clinical implications. There are currently no therapeutic options for ATC beyond procoagulant administration, i.e., plasma, which is known to be poorly effective at correcting coagulopathy^52,53  and the antifibrinolytic tranexamic acid.⁵⁴  But if the predominant pathophysiologic model converges on activation of PC, this may suggest a potential disadvantage to early administration of plasma, which provides a source of plasminogen and PC, both substrates for fibrinolytic pathways. However, this assumption must be balanced against the role of plasmin inhibitors, e.g., α2 antiplasmin, contained within plasma and their beneficial effects in ATC. We have previously reported preliminary results, which demonstrate that early high-dose fresh frozen plasma elicits a variable response in the coagulation system.⁴¹  If aPC-related fibrinolysis and fibrinogen depletion are the primary problems, then augmenting thrombin generation may further drive PC activation with negative effects on clot function and exacerbate bleeding. Although plasma contains important inhibitors of plasmin, i.e., α2 antiplasmin, resuscitation with products that lack plasminogen and PC, i.e., cryoprecipitate or fibrinogen concentrate, requires further study as they may have a more favorable therapeutic profile than those that contain protein C (plasma and prothrombin complex concentrates).

In summary, ATC is defined predominantly by increased fibrinolytic activity and rapid depletion of fibrinogen. We have now shown that the PC pathway provides a mechanistic link between the two main theories of early trauma coagulopathy—DIC with fibrinolytic phenotype¹³  and primary aPC-driven fibrinolysis (and possible fibrinogenolysis).⁷  Patients with raised aPC had a dose-dependent reduction in clot strength and evidence of increased fibrinolytic activity yet normal thrombin generation. The mechanistic confirmation of fibrinolysis as a central component of ATC supports clinical evidence from cohort studies^55–57  and randomized controlled trials⁵⁴  that early empiric administration of an antifibrinolytic improves outcomes in trauma. Fibrinogen was depleted early through a nonconsumptive (systemic) process, theoretically by fibrinogenolysis through the direct action of plasmin. Elevated aPC was associated with worse outcomes and increased transfusions after major trauma. A unifying hypothesis defined by aPC generation and widespread fibrinolytic activation provides new translational opportunities for treating this important global disease. Effective reversal of ATC may require novel treatment strategies directed at ameliorating activation of PC and subsequent depletion of PC stores with targeted replenishment of fibrinogen.

Transgenic thrombomodulin knockin mice were a kind gift from Hartmut Weiler, Ph.D., M.S., Human and Molecular Genetics Center, Medical College of Wisconsin, Milwaukee, Wisconsin. Frances Seeney, B.Sc., M.Sc., NHS Blood and Transplant, Bristol, United Kingdom, provided statistical review of data analysis.

Supported in part by the National Institute for Health Research Programme Grant for Applied Research (RP-PG-0407-10036), London, United Kingdom. TEM Innovations (Munich, Germany) provided ROTEM reagent and equipment on an unrestricted basis.

The authors declare no competing interests.