Trauma patients rapidly develop an acute traumatic coagulopathy (ATC) which is associated with poor outcomes including transfusion requirements, critical care utilization, and death.1–3 Hyperfibrinolysis is a central characteristic of ATC4,5 and early empiric antifibrinolytic administration has been shown to improve overall survival.6–8 There is controversy as to whether antifibrinolytics should be administered empirically or given only to patients with diagnostic evidence of hyperfibrinolysis.9,10 This controversy centers on the potential risks of inducing a thrombotic event in patients without active fibrinolysis and the role of viscoelastic hemostasis assays (VHAs) in the diagnosis of hyperfibrinolysis.11 This confusion has led to a wide variation in antifibrinolytic practice worldwide.
We have previously shown that a large number of patients with biomarker evidence of hyperfibrinolysis are not detected by rotation thromboelastometry (ROTEM) viscoelastic devices.12 Around 60% of trauma patients had an “occult” hyperfibrinolysis with elevated plasmin–antiplasmin (PAP) and D-dimer (DD) levels, increased mortality and blood product utilization, but no evidence of hyperfibrinolysis on ROTEM viscoelastic assay. Several studies have subsequently identified that high mortality is associated with a hypofibrinolytic pattern on thromboelastography and that this represents a group of patients in whom fibrinolysis has been switched off.13,14 This fibrinolysis “shutdown” pattern appears to be the most common fibrinolytic phenotype and the implication is that administration of antifibrinolytics to these patients could induce arterial or venous thrombotic events and death. It remains unclear though whether all low VHA detected lysis is abnormal; whether this is a heterogeneous group not differentiated by VHA; what are the underlying mechanisms leading to these different patterns of fibrinolysis; and which are associated with poor outcomes.
We hypothesized that not all low levels of fibrinolysis as detected by VHA were pathological. Our first aim was to determine whether low VHA-fibrinolysis represented a homogeneous or heterogeneous group of patients. If heterogeneous, we aimed to determine which patients had poor outcomes and to understand how they might be identified within the whole low VHA-fibrinolysis cohort. We further aimed to characterize the injury characteristics and outcomes of these patients, to shed light on the mechanism of this form of trauma-induced coagulopathy. We conducted an analysis of clinical data and samples from trauma patients prospectively enrolled into our platform study of trauma-induced coagulopathy.
The study was first approved by East London and The City Research Ethics Committee (07/Q0603/29) and then subsequently by each local ethics committee and conducted according to the Declaration of Helsinki. Deferred written informed consent was obtained from each patient or their next of kin. Blood samples from healthy volunteers taking no regular medications were obtained after written informed consent (07/Q0702/24).
A prospective multicenter observational cohort study, the Activation of Coagulation and Inflammation in Trauma (ACIT) study (UK CRN ID 5637), was conducted at 5 European major trauma centers in London, Oslo, Copenhagen, Amsterdam, and Cologne. All centers are members of the International Trauma Research Network. Patients were excluded from ACIT if they arrived >2 hours postinjury; were transferred from another hospital; received more than 2000 mL crystalloid prehospital; or had sustained burns of over 5% of their body surface area. Patients were retrospectively excluded if they declined to give consent to the use of their research samples, had severe liver disease, a known pre-existing bleeding diathesis, or were taking anticoagulant medication (excluding aspirin) preinjury. Of patients in the ACIT study recruited between January 2008 and July 2014, all adults (>15 yrs) who were either severely injured (ISS ≥ 15), shocked (lactate ≥ 2 mEq/L or base deficit ≥ 4 mEq/L), coagulopathic (international normalized ratio [INR] > 1.2 or ROTEM EXTEM CA5 ≤ 35 mm) or transfused ≥ 4 packed red blood cell (PRBC) units within the first 12 hours and did not receive tranexamic acid (TXA) had multiple biomarker assays performed as part of the Targetted Action for Curing Trauma Induced Coagulopathy program of work and are analyzed in this study.15
Research personnel at each center screened and enrolled patients. Data were collected prospectively and included patient demographics, time of injury, mechanism of injury (blunt or penetrating), Injury Severity Score (ISS), vital signs on-scene and on arrival in the emergency department (ED), total number of blood products and volume of intravenous fluids administered within the first 12 hours from injury. Patients were observed for 28 days from injury for the occurrence of venous thromboembolic events (deep vein thrombosis or pulmonary embolism), and overall mortality.
The baseline research blood sample was drawn within 20 minutes of the patient's arrival in the ED along with standard trauma laboratory tests. A conventional coagulation screen was performed for the determination of prothrombin time (PT) and INR and a point-of-care arterial blood gas analysis performed simultaneously for base deficit calculation.
Blood for ROTEM analysis was collected in a 2.7 mL citrated vacutainer (0.109 Molar/3.2% sodium citrate; Becton, Dickinson and Company, Plymouth, UK). Blood for coagulation and fibrinolysis protein assays was collected in a 4.5 mL glass citrated vacutainer (0.109 Molar/3.2% sodium citrate; Becton, Dickinson and Company, Plymouth, UK). The filled 4.5 mL vacutainer was centrifuged within 1 hour of collection and double-spun plasma subsequently stored at −80°C.
Functional Coagulation Analysis
Functional coagulation analysis was performed within 1 hour of blood drawn at 37°C on a ROTEM delta instrument (Tem International GmbH, Munich, Germany) using the automated electronic pipette according to the manufacturer's instructions. The methodology and parameters of ROTEM have been described previously.16 The EXTEM assay, measuring tissue-factor initiated clotting, was run for 60 minutes and the maximum lysis (ML) determined. For spiking experiments, recombinant S100A10 (Abbexa, Cambridge, UK) or vehicle was added to the ROTEM cup prior to the start of the test.
Coagulation and Fibrinolysis Plasma Protein Assays
Plasma stored at −80°C was thawed to 37°C immediately before all analyses. Prothrombin fragment 1 + 2 (prothrombin time Frag 1 + 2; Enzygnost F 1 + 2 (monoclonal); Siemens Healthcare Diagnostics Products GmbH, Marburg, Germany), tissue plasminogen activator (tPA; Asserachrom tPA, Diagnostica Stago, Asnières sur Seine, France), plasminogen activator inhibitor-1 (PAI-1; Asserchrom PAI-1; Diagnostica Stago), plasmin-α2-antiplasmin complex (PAP; PAP micro enzyme-linked immunosorbent assay; DRG Instruments GmbH, Marburg, Germany), urokinase (uPA, Abcam, UK) and S100A10 (Abbexa, Cambridge, UK) were measured using sandwich enzyme-linked immunosorbent assays. Fibrinogen levels (Siemens Thrombin reagent, Sysmex UK) and α2-antiplasmin (Siemens Berichrom α2-antiplasmin; Sysmex UK) were determined in the hospital laboratories with a Sysmex CS2100i automated analyzer (Sysmex, UK) according to standard protocols. Latex immunoassays were used to quantify the levels of D-dimer (Siemens Innovance D-dimer; Sysmex, UK) also with the Sysmex CS2100i automated analyzer.
Major hemorrhage was defined as administration of at least 4 units of PRBCs within 12 hours of admission, and massive hemorrhage as at least 10 PRBC units in 12 hours. Severe traumatic brain injury was defined as a brain abbreviated injury score of 3 or higher. Coagulopathy was defined as an INR > 1.2. We defined VHA hyperfibrinolysis as an ML >15%.12
Statistical analysis was performed using GraphPad Prism version 5 (GraphPad Software Inc, San Diego, CA) and Microsoft Excel 2013 (Microsoft Inc, Redmond, USA). Normal quantile plots were used to test for normal distribution. Nonparametric data are expressed as median (interquartile range) and analyzed using Mann–Whitney U test or Kruskal–Wallis test with Dunn multiple comparison test. Percentages were analyzed using Chi-squared or Fisher exact tests. Correlation between continuous variables was assessed with Spearman correlation coefficient. Trends in survival were compared with Log-rank (Mantel-Cox) test. A P value of < 0.05 was considered statistically significant.
A total of 2465 subjects were enrolled in the ACIT study during the recruitment period across the 5 study sites. Of these, 1318 patients met the inclusion criteria and were eligible for enrolment into this study. Twenty-five patients coenrolled in the CRASH-3 trial were excluded, along with 169 who received TXA, 196 who did not have a baseline ROTEM performed and 14 who did not have a DD measured, leaving 914 patients for analysis.
Determination of ROTEM Maximum Lysis Thresholds for hypofibrinolysis
We first determined our study threshold for low VHA fibrinolysis as this has not been previously defined for ROTEM (Fig. 1). A histogram of mortality for each % ML suggested a lower inflection point at less than 5% and below (Fig. 1A). As confirmation a receiver operating characteristic curve for mortality was generated using patients with a ML value of ≤15%. The point of ML <5.5% had the optimal performance by the Youden index with a sensitivity of 61.6% and specificity of 58.4%. Mortality was doubled in patients with ML <5% (15 vs. 7.1% for ML 5–15%, P<0.001). We therefore defined MLLOW as ML <5%; MLNORMAL as ML 5–15%; and MLHIGH as ML>15%.12 These 3 groups of patients showed a U-shaped mortality curve, similar to previous studies of hypofibrinolysis in trauma13,14 (Fig. 1B).
Characterization of Patterns Associated With Mortality in the MLLOW Cohort
To explore the heterogeneity of the MLLOW group, we began by examining the differences between MLLOW patients who lived and those who died. The survivors in this group had very similar admission characteristics, injury patterns, and outcomes to MLNORMAL patients. The MLLOW survivors were statistically older, more shocked and more severely injured, but these differences were small (Table 1). Their outcomes in terms of VTE, massive transfusion rates, critical care, and hospital lengths of stay were the same as for MLNORMAL patients. In contrast, the MLLOW patients who died were much older, critically injured (median ISS 29), had lower GCS than any other cohort (median 7) and 85% of them had a traumatic brain injury (Table 1).
MLLOW patients who died were nearly 3 times as likely to receive a massive transfusion and were 4 times as likely to be coagulopathic on admission (63% INR >1.2) than MLLOW patients who survived (Table 1). Despite the low ML, DD levels in patients who died were extremely high (median 103,170 ng/mL), compared with 13,672 ng/mL in survivors (P < 0.001, Table 1, Fig. 1C). These DD differences were replicated when analyzing coagulopathic versus noncoagulopathic patients and red cell transfusion requirements (Fig. 1D–I). For further analysis we therefore set a DD threshold at 30,000 ng/mL (approximately the upper quartile in MLNORMAL patients) and used this as a biological marker to explore the heterogeneity within the MLLOW patient cohort.
More than half (58%) of patients with MLLOW on admission had low DD levels. Patients with MLLOW and DDLOW were essentially identical to patients who were MLNORMAL in terms of injuries, admission physiology, and outcomes (Table 2). The only significant difference between the 2 groups was that MLLOW + DDLOW patients were more likely to have received prebaseline crystalloid (51 vs. 37%, P = 0.002).
Characteristics of Patients Presenting With Low ML But High DD Levels
Mortality was 30% in the MLLOW + DDHIGH cohort compared with only 3% in the MLLOW + DDLOW group (P < 0.001, Table 2). Deaths occurred later in the MLLOW + DDHIGH cohort compared with MLHIGH + DDHIGH patients, who tended to die on the day of admission (Fig. 2A, Table 2). MLLOW + DDHIGH patients were more severely injured overall (median ISS 29) and were more than twice as likely to have a severe head or torso injury as the MLLOW + DDLOW cohort. MLLOW + DDHIGH patients were also more shocked and coagulopathic (by INR) on admission and had greater fluid and blood product requirements in the first 24 hours than those with MLLOW + DDLOW. Overall MLLOW + DDHIGH patients appeared to be severely injured and bleeding and were more like the MLHIGH + DDHIGH group. The MLHIGH + DDHIGH patients had a greater severity of shock and much higher transfusion requirements, but were less likely to have sustained a severe traumatic brain injury (Table 2).
Coagulation Profiles of Patients With Low ML But High DD Levels
We examined the fibrinolytic mechanisms in these patient cohorts (Fig. 2B–F). Across all low D-dimer cohorts, prothrombin fragment (PF1 + 2), fibrinogen, plasmin-antiplasmin complex (PAP), tPA, urokinase-type plasminogen activator (uPA), and PAI-1 levels were the same.
Patients with MLLOW + DDHIGH had the highest levels of prothrombin fragments of all cohorts, associated with a reduction in fibrinogen levels (Fig. 2B, Supplementary Table 2, http://links.lww.com/SLA/B393). As expected, they also had high PAP levels indicative of fibrinolytic activity (Fig. 2C). However, surprisingly, tPA levels in patients with MLLOW + DDHIGH were low and comparable to the MLNORMAL and MLLOW + DDLOW cohorts (Fig. 2D). Plasma levels of uPA were the same across all groups, as were PAI-1 levels (Fig. 2E, F). Antiplasmin levels were low in the MLLOW + DDHIGH group (72 vs. MLNORMAL 102 u/dL, P < 0.001) and even lower in MLHIGH + DDHIGH patients (median 42.0 u/dL, P < 0.001 vs. MLNORMAL – Supplementary Table 2, http://links.lww.com/SLA/B393).
Mechanism of Occult Hyperfibrinolysis in Trauma Patients
MLLOW + DDHIGH patients therefore had high thrombin generation and high fibrinolytic activity but no elevations in the expected plasminogen activators nor increased levels of plasminogen/plasmin inhibitors. We therefore proceeded to measure levels of S100A10, a principally membrane-bound mediator of fibrinolysis in a subgroup of 179 patients. Clinical characteristics and coagulation profiles were similar to the main cohort (supplemental Tables 1 and 2, http://links.lww.com/SLA/B393). Plasma S100A10 levels were low in the DDLOW cohorts but significantly elevated in both the MLLOW + DDHIGH and MLHIGH + DDHIGH patients (Fig. 3A). Plasma S100A10 levels were strongly correlated with both PAP (r = 0.58, P < 0.001) and DD (r = 0.56, P < 0.001) levels (Fig. 3B, C). S100A10 levels were strongly associated with injury severity (Fig. 3D). In particular, patients with TBI (abbreviated injury score Head 3 + ) had much higher plasma levels of S100A10 than those without (S100A10: 437 vs. 55 pg/mL, P < 0.001). TBI patients with higher plasma S100A10 (above the median 120 pg/mL) had a 58% mortality, compared with only 6% for those with low S100A10 levels (P < 0.001), as well as a higher incidence of coagulopathy (31 vs. 0%, P = 0.009) and PRBC transfusion (38 vs. 6%, P = 0.01).
As S100A10 is known to bind tPA, we further hypothesized that S100A10 would bind tPA in the ROTEM cup, thus reducing ex-vivo plasmin generation and therefore the VHA-measured fibrinolysis. We added 10 ng/mL S100A10 to the whole blood of 6 healthy volunteers and saw a significant reduction in %ML (Fig. 3E). In the MLLOW and MLNORMAL combined cohort, plasma S100A10 levels negatively correlated with %ML (r = −0.26, P < 0.001, Fig. 3F) but were still positively correlated with PAP (r = 0.56, P < 0.001) and DD levels (r = 0.28, P < 0.001). There was no relationship between plasma levels of S100A10 and plasma tPA levels.
In this multicenter study of over 900 patients, we have shown that low levels of fibrinolysis as measured by VHAs are physiological in over half of all cases and have a mortality rate and transfusion requirement which is comparable to patients with normal VHA levels of fibrinolysis. Conversely, a subgroup of patients with low VHA detected fibrinolysis are severely injured, functionally coagulopathic, have high transfusion requirements and a 30% mortality. These patients have high levels of thrombin, plasmin, and DD generation but have low levels of plasminogen activators including tPA. We have identified S100A10 as a possible explanation for localized activation of plasmin on cell surfaces in these patients which when shed into the plasma artificially lowers the observed VHA maximum lysis parameter ex-vivo. Together these results support the existence of an occult hyperfibrinolysis in severely injured patients, not detectable by existing diagnostic tools and which is associated with very poor outcomes.
S100A10 is a membrane-bound plasminogen receptor which complexes with Annexin 2 in a hetero-tetramer formation to bind tPA and plasminogen on the cell surface in close proximity to the uPA receptor.17–19 S100A10 is widely expressed in body tissues and particularly in the brain20,21 and the complex is upregulated at times of hypoxic stress.22 S100A10 levels are detectable in the blood but the majority of cellular plasmin is formed on the endothelium with S100A10 catalyzing the tPA mediated activation of plasminogen.23,24 The classical presentation of Acute Promyelocytic Leukaemia is a hemorrhagic and fibrinolytic phenotype characterized by normal levels of tPA but high surface expression of S100A10 on Acute Promyelocytic Leukaemia cells.25 Our findings suggest that S100A10 is exposed by tissue injury leading to plasmin generation and hyperfibrinolysis without the need for high levels of circulating tPA.26 These findings suggest a mechanism that may contribute to the catastrophic coagulopathy often seen with severe traumatic brain injury, the severity of which is not apparent on existing diagnostic tests.
Plasmin has a very short half-life and S100A10 is known to increase plasmin autoproteolysis.24 Ex-vivo detection of fibrinolysis by VHA devices relies on the presence of plasmin to lyse clot forming in the measuring cup. We have shown that the addition of S100A10 to blood reduces VHA-detected lysis. In patients with low tPA levels there is no other available activator of plasminogen ex-vivo and no fibrinolysis will be detected. When tPA levels are high, the effect of S100A10 will be overcome and fibrinolysis will be revealed. These findings are entirely consistent with all observed studies of hyperfibrinolysis in trauma patients.
There has been controversy in the literature as to whether ATC patients with low maximum lysis but high D-dimer levels (MLLOW + DDHIGH) have ongoing hyperfibrinolysis or have had their fibrinolysis shutdown at some time immediately before blood sampling. These patients are severely injured and have large on-going blood product requirements. They have a high incidence of coagulopathy with an underlying biochemistry of high PAP and DD levels. Taken together these features suggest active hyperfibrinolysis. It is difficult to support a fibrinolysis shutdown model that would require explanation of how tPA levels could have suddenly become low given that the active pathological processes of injury, shock and bleeding have not yet been controlled and PAI-1 levels are not elevated. Localized hyperfibrinolysis via cell-bound activation of plasminogen is a more plausible mechanism and matches the observed on-going bleeding and transfusion requirements in these patients. This occult hyperfibrinolysis in ATC does not preclude the likely existence of a later fibrinolytic shutdown in trauma patients due to upregulation of PAI-1 and other control mechanisms.
Tranexamic acid's effective reduction of mortality in trauma patients appears to extend beyond reducing blood loss and death from bleeding to affect all-cause mortality.6,27 Our results suggest that severe tissue injury can induce hyperfibrinolysis through S100A10 and tranexamic acid may therefore reduce local as well as systemic hemorrhage. In traumatic brain injury this might have a profound effect on outcome and we found S100A10 levels correlated closely with coagulopathy, blood use and mortality in severe TBI patients. S100A10 may provide the mechanistic rationale for the use of tranexamic acid in TBI and the results of on-going clinical trials are anticipated soon.28,29 Overall, our results confirm an occult hyperfibrinolysis which is prevalent in trauma patients and supports the empiric use of tranexamic acid until new diagnostics are available. Restricting treatment only to patients with VHA detected lysis may exclude 80% of trauma patients with high DD levels who might benefit from tranexamic acid.
The inclusion criteria for this study prevent us from determining the absolute incidence of each of these subtypes of TIC. As patients with minor injuries and without shock were excluded, overall rates of hyperfibrinolysis are likely to be overestimates compared with a general trauma population. As the MLLOW + DDHIGH S100A10-driven subtype is essentially exclusive to blunt trauma, its incidence will also vary depending on rates of blunt trauma seen in specific institutions. Our exclusion of patients who had already received tranexamic acid is also likely to have led to an underestimate of the numbers in the MLHIGH subtypes and contributed to the relatively low incidence of massive transfusion in this cohort. There are several aspects of this work that require further experimental research. This clinical study cannot definitively state that the MLLOW + DDHIGH patients still have active fibrinolysis after the point of sampling. Our finding of high circulating levels of the primarily tissue-bound fibrinolytic activator S100A10 is circumstantial, but coupled with grossly elevated PAP & DD and no obvious inhibitors, is highly suggestive of active lysis, as are subsequent high blood product requirements. It will require both experimental evidence of active lysis and clinical trial evidence of the effect of TXA in this subgroup to confirm this. The mechanisms of exposure and release of S100A10 also need exploring, especially in the context of traumatic brain injury. Again, both clinical and experimental studies will be required to understand the role of S100A10 in trauma-induced coagulopathy.
We have shown that VHA-detected lysis (ML) is a poor indicator of both the fibrinolytic state and clinical outcomes of patients. Absolute DD levels were more strongly associated with outcomes. Patients with low DD levels had good outcomes regardless of ML. In contrast a specific phenotype of patients exists with elevated DD but low ML who have poor outcomes. These patients are critically injured and have a high incidence of severe traumatic brain injury. We have identified the tissue-bound S100A10 fibrinolytic mediator as a candidate mechanism for hyperfibrinolysis in these patients that also results in an artificially low ML. S100A10 exposure after critical injury represents a new precision marker and therapeutic opportunity in trauma-induced coagulopathy.
All of the author institutes are affiliated members of the International Trauma Research Network and as such this work represents a combined output resulting from this Network. The authors acknowledge Patricia Madureira and Sean Platton for technical advice.
1. Brohi K, Singh J, Heron M, et al. Acute traumatic coagulopathy
. J Trauma
2. Maegele M, Lefering R, Yucel N, et al. Early coagulopathy
in multiple injury: an analysis from the German Trauma
Registry on 8724 patients. Injury
3. MacLeod JBA, Lynn M, McKenney MG, et al. Early coagulopathy
predicts mortality in trauma
. J Trauma
4. Davenport RA, Guerreiro M, Frith D, et al. Activated protein C drives the hyperfibrinolysis of acute traumatic coagulopathy
5. Brohi K, Cohen M, Ganter M, et al. Acute coagulopathy
: hypoperfusion induces systemic anticoagulation and hyperfibrinolysis. J Trauma
6. Shakur H, Roberts I, Bautista R, et al. CRASH-2 trial collaborators. Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma
patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. Lancet
7. Morrison JJ, Dubose JJ, Rasmussen TE, et al. Military application of tranexamic acid in trauma
emergency resuscitation (MATTERS) study. Arch Surg
8. Morrison JJ, Ross JD, Dubose JJ, et al. Association of cryoprecipitate and tranexamic acid with improved survival following wartime injury: findings from the MATTERs II Study. JAMA Surg
9. Moore EE, Moore HB, Gonzalez E, et al. Rationale for the selective administration of tranexamic acid to inhibit fibrinolysis
in the severely injured patient. Transfusion
10. Roberts I. Fibrinolytic shutdown: fascinating theory but randomized controlled trial data are needed. Transfusion
11. Gall L, Brohi K, Davenport R. Diagnosis and treatment of hyperfibrinolysis in trauma
(a European perspective). Semin Thromb Hemost
12. Raza I, Davenport R, Rourke C, et al. The incidence and magnitude of fibrinolytic activation in trauma
patients. J Thromb Haemost
13. Moore HB, Moore EE, Gonzalez E, et al. Hyperfibrinolysis, physiologic fibrinolysis
, and fibrinolysis
shutdown. J Trauma Acute Care Surg
14. Moore HB, Moore EE, Liras IN, et al. Acute fibrinolysis
shutdown after injury occurs frequently and increases mortality: a multicenter evaluation of 2,540 severely injured patients. J Am Coll Surg
15. Targetted Action for Curing Trauma
(TACTIC) project. Available at: http://www.tacticgroup.dk/
. Accessed November 24, 2017.
16. Ganter MT, Hofer CK. Coagulation monitoring: current techniques and clinical use of viscoelastic point-of-care coagulation devices. Anesth Analg
17. MacLeod TJ, Kwon M, Filipenko NR, et al. Phospholipid-associated annexin A2-S100A10
heterotetramer and its subunits: characterization of the interaction with tissue plasminogen activator, plasminogen, and plasmin. J Biol Chem
18. Kassam G, Choi KS, Ghuman J, et al. The role of annexin II tetramer in the activation of plasminogen. J Biol Chem
19. Madureira PA, Surette AP, Phipps KD, et al. The role of the annexin A2 heterotetramer in vascular fibrinolysis
20. Egeland M, Warner-Schmidt J, Greengard P, et al. Co-expression of serotonin 5-HT(1B) and 5-HT(4) receptors in p11 containing cells in cerebral cortex, hippocampus, caudate-putamen and cerebellum. Neuropharmacology
21. Kwaan HC, Wang J, Weiss I. Expression of receptors for plasminogen activators on endothelial cell surface depends on their origin. J Thromb Haemost
22. Luo M, Hajjar KA. Annexin A2 system in human biology: cell surface and beyond. Semin Thromb Hemost
23. Kassam G, Le BH, Choi KS, et al. The p11 subunit of the annexin II tetramer plays a key role in the stimulation of t-PA-dependent plasminogen activation. Biochemistry
24. Choi KS, Fitzpatrick SL, Filipenko NR, et al. Regulation of plasmin-dependent fibrin clot lysis by annexin II heterotetramer. J Biol Chem
25. O’Connell PA, Madureira PA, Berman JN, et al. Regulation of S100A10
by the PML-RAR-α oncoprotein. Blood
26. Madureira PA, O’Connell PA, Surette AP, et al. The biochemistry and regulation of S100A10
: a multifunctional plasminogen receptor involved in oncogenesis. J Biomed Biotechnol
27. Cole E, Davenport R, Willett K, et al. Tranexamic acid use in severely injured civilian patients and the effects on outcomes: a prospective cohort study. Ann Surg
28. Dewan Y, Komolafe EO, Mejía-Mantilla JH, et al. CRASH-3 - tranexamic acid for the treatment of significant traumatic brain injury: study protocol for an international randomized, double-blind, placebo-controlled trial. Trials
29. Prehospital Tranexamic Acid Use for Traumatic Brain Injury (TXA). November 21, 2013. Available at: https://clinicaltrials.gov/show/NCT01990768
. Accessed November 24, 2017.