Coagulopathy as a predictor of the effectiveness of tranexamic acid in severe blunt trauma: a multicenter retrospective study

Abstract

Background

Tranexamic acid (TXA) reduces mortality in severe trauma cases. However, the relationships between TXA administration and coagulation/fibrinolysis abnormalities are unclear. We performed a retrospective observational study to investigate relationships between mortality and coagulation/fibrinolysis abnormalities of patients on arrival at the emergency department and whether TXA is more effective in patients with severe trauma who have coagulation/fibrinolysis abnormalities than in those who do not.

Methods

Data was collected from 15 tertiary emergency and critical care centers in Japan. Adult patients with blunt trauma and an Injury Severity Score of ≥ 16 were included in the study. Patients were categorized into two groups: the TXA group received TXA within 3 h of arrival, and the non-TXA group did not.

Results

Overall, 790 patients were included (TXA group, 276; non-TXA group, 514). In cubic spline curves for relationships between mortality and coagulation/fibrinolysis variables on arrival, odds for mortality increased and plateaued with a prothrombin time-international normalized ratio ≥ 1.2; the disseminated intravascular coagulation (DIC) score showed a marked odds increase when > 4 points. Odds increased and plateaued from an activated partial thromboplastin time (APTT) of ≥ 35 s and gradually increased as fibrinogen decreased from 250 mg/dL. Fibrinogen and fibrin degradation products (FDP) and D-dimer exhibited upward-sloping curves. In cubic spline curves for relationships between the effectiveness of TXA administration and coagulation/fibrinolysis variables on arrival, a favorable effect on mortality was observed with TXA administration when fibrinogen was ≤ 200 mg/dL or when the DIC score was ≥ 4 points; FDP, ≥ 50 µg/mL; D-dimer, ≥ 30 µg/mL; or APTT, ≥ 35 s. In each threshold subgroup, interactions between TXA administration and in-hospital mortality were observed.

Conclusions

TXA demonstrates increased effectiveness in patients with traumatic coagulation/fibrinolysis abnormalities.

Background

Severe trauma is a leading cause of morbidity and mortality worldwide, accounting for millions of deaths annually [1]. One of the most significant complications of severe trauma is the development of trauma-induced coagulopathy (TIC). TIC is a multifactorial disorder characterized by abnormal blood coagulation, hyperfibrinolysis, and the consumption of clotting factors caused by severe trauma itself, which can lead to serious complications and even death [2,3,4]. The Scientific and Standardization Committee on Disseminated Intravascular Coagulation (DIC) of the International Society on Thrombosis and Hemostasis supports the conclusion that TIC encompasses DIC and transitions into it as a pathological condition [5]. However, no clear and specific definition of TIC has yet been established. Although previous studies have used various definitions of TIC, most studies have indicated that approximately 25% of patients with severe trauma develop TIC—which has a high mortality rate that ranges between 35% and 50% [2, 6]. Furthermore, the ongoing dysregulation of coagulation and fibrinolysis may eventually lead to DIC in patients with severe trauma [3,4,5].

Tranexamic acid (TXA) is an antifibrinolytic agent that reduces mortality in patients with severe trauma [7,8,9,10]. The Clinical Randomization of an Antifibrinolytic in Significant Hemorrhage (CRASH)-2 trial indicated that TXA administration within 3 h following injury improved the survival rate of patients with a high risk of bleeding [8]. For traumatic brain injury, which is a leading cause of death and massive hemorrhage in patients with severe trauma, the CRASH-3 trial showed that early TXA treatment reduced the risk of death caused by brain bleeding in patients with mild to moderate traumatic brain injury [11]. Because TXA is a potent antifibrinolytic, the relationship between the effect of TXA administration and coagulation abnormalities has been studied [12,13,14]. However, the relationships between the effects of TXA administration and coagulation/fibrinolysis abnormalities in patients with severe trauma in clinical settings have not been reported. Furthermore, although patients with traumatic brain injury and those with hemorrhagic shock have different coagulation/fibrinolysis abnormalities [15], the difference in the relationship between the effects of TXA administration and coagulation/fibrinolysis abnormalities in these conditions has not been evaluated.

In this study, we hypothesized that TXA administration would be more effective in patients with coagulation and fibrinolysis abnormalities than in those without. Therefore, we investigated the relationships between mortality and coagulation/fibrinolysis abnormalities on arrival at the emergency department and between the effectiveness of TXA administration and coagulation/fibrinolysis abnormalities in patients with severe trauma.

Methods

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of the Ethics Committee of Tohoku University School of Medicine (No. 2013-1-41, May 17, 2013). The study was conducted in accordance with the Declaration of Helsinki, and the need for additional written informed consent was waived because of the retrospective study design involving the use of samples already in storage.

Study design and population

This retrospective study used data from the Japanese Observational Study for Coagulation and Thrombolysis in Early Trauma (J-OCTET) [15,16,17]. This multicenter observational study aimed to examine the relationships among mortality, blood transfusion requirements, baseline characteristics, and blood data in patients with trauma. The J-OCTET database comprises patients aged > 18 years with severe trauma who were admitted to 15 tertiary emergency and critical care centers between January 2012 and December 2012 with Injury Severity Scores (ISS) of ≥ 16. Pregnant individuals or patients with cardiac arrest, burns, isolated cervical spine injury unrelated to a high-energy accident, or liver cirrhosis were excluded from this study. This study analyzed a population of patients with blunt trauma registered in the J-OCTET database.

Data collection and definitions

The collected data included patient demographics, ISS, and laboratory data upon arrival at the emergency department. ISS was calculated using the Abbreviated Injury Scale (AIS) 2005. Severe shock on arrival was defined as the heart rate/systolic blood pressure ratio of ≥ 1. Severe TBI was defined as an AIS for a head score of ≥ 3 and a Glasgow Coma Scale score of < 9. We used the Japanese Association for Acute Medicine DIC scoring system [18]. The included patients were categorized into the following two groups: the TXA group included patients who received TXA within 3 h of arrival, and the non-TXA group included those who did not. The primary outcome was all-cause in-hospital mortality, and the secondary outcome was in-hospital mortality caused by massive bleeding and primary brain injury. The causes of in-hospital mortality were determined by attending physicians.

Statistical analysis

Continuous variables are expressed as medians (interquartile ranges), and categorical variables are expressed as numbers (percentages). We employed multiple imputations using chained equations to address any missing data, assuming that the missing mechanism was random, and 100 imputed datasets were created. All missing variables were imputed, and the missing data are indicated in the table detailing the patients’ characteristics. Multiple imputations included TXA administration, ISS, systolic blood pressure, prothrombin time-international normalized ratio (PT-INR), age, Glasgow Coma Scale score, respiratory rate, DIC score, platelet count, activated partial thromboplastin time (APTT), fibrinogen, fibrinogen/fibrin degradation products (FDP), D-dimer, lactate, site, sex, prehospital infusions, regular use of anticoagulant or antiplatelet drugs, AIS of each body part, revised trauma score, probability of survival, a complication of brain injury, heart rate, body temperature, white blood cell count, red blood cell count, hemoglobin, hematocrit, creatine kinase, aspartate aminotransferase, alanine aminotransferase, lactate dehydrogenase, pH, partial pressure of carbon dioxide, base excess, and discharge outcome (death). To incorporate a possible nonlinear association of mortality and coagulation/fibrinolysis variables (PT-INR, APTT, fibrinogen, platelet counts, FDP, D-dimer, and DIC scores), we used a restricted cubic spline with logistic regression to flexibly model such association, with knots placed at the 10th, 50th, and 90th percentiles [19]. Each logistic regression included all-cause in-hospital mortality as the primary outcome and in-hospital mortality caused by massive bleeding and primary brain injury as the secondary outcome. Cutoff values of coagulation/fibrinolysis variables were established based on the point where the 95% CI of the spline plot curves of relationships between coagulation/fibrinolysis variables and TXA effectiveness crossed the line for odds ratio = 1. TXA administration, coagulation/fibrinolysis variables, and the interaction between TXA administration and coagulation/fibrinolysis variables comprised the explanatory variables—which were adjusted for age, ISS, systolic blood pressure, Glasgow Coma Scale score, and respiratory rate. All analyses were performed using R version 4.2.1. All reported p-values were two-tailed, and interactions with p-values < 0.20 and differences of p < 0.05 were considered statistically significant.

Results

Patient characteristics

In the J-OCTET study, 796 patients were registered, and six with penetrating trauma were excluded. In total, 790 patients were included in this study. The TXA and non-TXA groups included 276 and 514 patients, respectively. The patient characteristics of the two groups are provided in Table 1. Overall, the patients who received TXA had more severe diseases than those who did not. Patient outcomes in the two groups are provided in Table 2. Although more massive transfusions were performed in the TXA group than in the non-TXA group, in-hospital mortality and cause of death did not differ between the two groups.

Table 1 Patient characteristics

Table 2 Patient outcomes

Relationships between mortality and coagulation/fibrinolysis variables

Figure 1 illustrates the relationship between the odds of mortality and coagulation/fibrinolysis variables using cubic spline curves. Notably, the odds gradually increased and plateaued from a PT-INR of ≥ 1.2, and the DIC score showed a remarkable increase in odds once it exceeded 4 points. Furthermore, the odds gradually increased and plateaued from an APTT of ≥ 35 s and gradually increased as fibrinogen decreased from 250 mg/dL. FDP and D-dimer levels exhibited upward-sloping curves.

Fig. 1

Relationships between coagulation/fibrinolysis variables on arrival and odds for all-cause in-hospital mortality. The odds are presented on the vertical axis of the graph using a logarithmic scale, and the values of each parameter are expressed on the horizontal axis. The gray area represents the 95% confidence interval

PT-INR, prothrombin time-international normalized ratio; APTT, activated partial thromboplastin time; FDP, fibrinogen/fibrin degradation products; DIC, disseminated intravascular coagulation

Relationships between the effectiveness of TXA administration and coagulation/fibrinolysis variables

Figure 2 demonstrates the relationship between the odds ratio for mortality attributable to TXA administration and coagulation/fibrinolysis variables using cubic spline curves. The 10th and 90th percentile values used in Fig. 2 are provided in Additional File 1. From these curves, a favorable effect on mortality rate can be observed with TXA administration when fibrinogen is < 200 mg/dL or when the values of parameters related to the coagulation/fibrinolysis system exceed the following thresholds: DIC score, 4 points; FDP, 50 µg/mL; D-dimer, 30 µg/mL; or APTT, 35 s. Appropriate cutoff values for platelet counts and PT-INR were not observed. Using the aforementioned thresholds for each parameter, a forest plot of mortality for each subgroup was created (Fig. 3). Subsequently, the interactions between TXA administration and in-hospital mortality were observed for each subgroup.

Fig. 2

Relationships among odds ratios of all-cause in-hospital mortality affected by TXA administration and coagulation/fibrinolysis variables. The odds ratio of all-cause in-hospital mortality attributable to the administration of tranexamic acid is represented on the vertical axis, and the coagulation/fibrinolysis variables are expressed on the horizontal axis. The gray area represents the 95% confidence interval. Each odds ratio indicates the ratio of the odds in the TXA group to the odds in the non-TXA group at each value of the coagulation/fibrinolysis variables

Fig. 3

Forest plot of the adjusted odds ratio of all-cause in-hospital mortality. The cutoff value for each subgroup was defined according to the cubic spline curve of the odds ratio of all-cause in-hospital mortality attributable to the administration of tranexamic acid (Fig. 2). The interactions between tranexamic acid administration and all-cause in-hospital mortality were observed for each subgroup. PT-INR, prothrombin time-international normalized ratio; APTT, activated partial thromboplastin time; FDP, fibrinogen/fibrin degradation products; DIC, disseminated intravascular coagulation; TXA, tranexamic acid

Figure 4 presents the relationship between the odds ratio for in-hospital mortality caused by massive bleeding and primary brain injury attributable to TXA administration and coagulation/fibrinolysis variables. Relationships between TXA effectiveness and coagulation/fibrinolysis variables for in-hospital mortality caused by primary brain injury were similar to those for all-cause in-hospital mortality. However, the relationships for in-hospital mortality caused by massive bleeding were different from those for in-hospital mortality caused by primary brain injury.

Fig. 4

Coagulation/fibrinolysis variables and TXA administration effectiveness for massive bleeding and primary brain injury in-hospital mortality. The upper section indicates the odds ratio of all-cause in-hospital mortality attributable to the tranexamic acid administration. The middle section indicates the odds ratio of in-hospital mortality caused by primary brain injury attributable to tranexamic acid administration. The lower section indicates the odds ratio of in-hospital mortality caused by massive bleeding attributable to tranexamic acid administration

The relationships between tranexamic acid effectiveness for in-hospital mortality caused by primary brain injury and coagulation/fibrinolysis variables were similar to those for all-cause in-hospital mortality. However, the relationships for in-hospital mortality caused by massive bleeding were different from those for in-hospital mortality caused by primary brain injury. PT-INR, prothrombin time-international normalized ratio; APTT, activated partial thromboplastin time; FDP, fibrinogen/fibrin degradation products; DIC, disseminated intravascular coagulation; TXA, tranexamic acid

Discussion

This study retrospectively analyzed data from 790 patients with severe blunt trauma. Patient mortality was observed to be associated with the coagulation/fibrinolysis variables on arrival at the emergency department. Furthermore, the improvement in the survival rate conferred by TXA administration varied widely across coagulation and fibrinolysis profiles among the patients. In particular, the positive effects of TXA administration may be greater in patients with DIC on admission. However, the effects of TXA administration for in-hospital mortality caused by primary brain injury were different from those for in-hospital mortality caused by massive bleeding.

As hypothesized, TXA administration was more effective in patients with poorer coagulation and fibrinolysis profiles. This is likely because, while TXA indeed exerts an anti-inflammatory effect, its main effect is believed to be the improvement of coagulation and fibrinolysis abnormalities in patients with severe blunt trauma—thereby acting as an antifibrinolytic agent [12]. Although the effectiveness of TXA administration did not vary significantly depending on PT or platelet counts, strong relationships were observed between the effectiveness of TXA administration and patient levels of FDP, D-dimers, and fibrinogen. Because FDP, D-dimer, and fibrinogen levels are sensitive to hyperfibrinolysis on arrival at the emergency department in patients with severe trauma [17, 20], these results reflect the antifibrinolytic effects of TXA administration on hyperfibrinolysis in patients with severe trauma. Furthermore, upon arrival at the emergency department, DIC was observed to exhibit a hyperfibrinolytic phenotype in patients with severe trauma [3]. Therefore, it is natural for TXA administration to be more effective as the DIC score increases.

Multiple studies have reported that coagulation and fibrinolysis abnormality upon arrival at the emergency department predict poor prognosis in patients with severe trauma. For example, low fibrinogen levels can be considered a predictor of massive transfusion [21], high D-dimer levels predict early death or the requirement for massive transfusion [17, 20], and DIC in the early phase of trauma leads to poor prognosis caused by massive bleeding [20]. However, to the best of our knowledge, this study is the first to demonstrate the relationship between mortality and coagulation/fibrinolysis variables using cubic spline curves as continuous variables rather than through group comparisons.

The cubic spline curves in Fig. 2 demonstrate that the relationships between the effects of TXA administration and coagulation and fibrinolysis-related variables were not linear, with significant changes occurring after a certain threshold. The forest plot created using the thresholds obtained from this cubic spline curve is perhaps the most straightforward indication that TXA is more effective in patient subgroups with coagulation/fibrinolysis abnormalities. Notably, these thresholds were similar to those that have been previously reported to predict worsening prognoses in patients with severe trauma [17, 20]. These concordances may be because the hyperfibrinolysis targeted by TXA represents a key cause of poor prognosis in patients with severe trauma.

As shown in Fig. 4, the relationships between TXA effectiveness and coagulation/fibrinolysis abnormality for in-hospital mortality caused by primary brain injury were similar to those for all-cause in-hospital mortality. However, relationships for in-hospital mortality caused by massive bleeding were different from those for all-cause in-hospital mortality. In this study, because the main cause of in-hospital death was primary brain injury (61% in all-cause in-hospital death), the relationships for in-hospital mortality caused by primary brain injury were expected to be similar to those for all-cause in-hospital mortality. Furthermore, improvements in outcomes due to TXA administration in patients with traumatic brain injury have been confirmed [11]. Conversely, in patients with traumatic hemorrhagic shock, the effects of TXA administration have been reported to be heterogeneous [12, 22,23,24]. However, in the current study, the frequency of TXA administration for patients with severe shock was 44% (55 of 125 patients), slightly higher than that for patients with severe traumatic brain injury without severe shock (39%, 57 of 145 patients). The difference in the frequency of TXA administration may be related to a difference in TXA effectiveness for in-hospital mortality in the current study.

However, the limitations of this study should be acknowledged. First, this was a retrospective study. The absence of randomization may have affected the observed associations, and unmeasured confounding variables may have influenced the validity of our conclusions. Conversely, our results were supported by novel statistical approaches that have led to new insights regarding severe trauma management. Second, this study focused on a specific Japanese population, which limits its generalizability to broader demographics and other healthcare settings. Third, this study focused on in-hospital mortality and assessed short-term outcomes. Long-term follow-up data, including potential complications or delayed effects related to TXA administration, were excluded, which limited our ability to assess the broader treatment impact. Finally, this study was based on a relatively older dataset. However, since the data were collected when TXA was not routinely administered to patients with severe trauma, it is particularly suitable for demonstrating the effectiveness of TXA administration.

Conclusions

Despite the abovementioned limitations, our study underscores the intrinsic relationships among TXA treatment, coagulation/fibrinolysis abnormalities, and mortality in patients with severe trauma, thereby offering valuable insights for clinical decision-making. Although there are no previous reports on the use of traumatic coagulation/fibrinolysis abnormalities on arrival to determine TXA administration, based on the current findings, patients with traumatic coagulation/fibrinolysis abnormalities on arrival may have to be aggressively treated with TXA as soon as possible. Further research, particularly well-designed trials, is essential to validate and refine our findings and ultimately guide the development of tailored therapeutic strategies in the challenging landscape of severe trauma care.