Characterizing The Impact Of Traumatic Brain Injury Phenotype on Coagulation Dynamics In Severely Injured Patients

Patient Sample Collection and Preparation

Between January 2021 and June 2023, after screening for inclusion and exclusion criteria, a total of 63 patients categorized at the highest level of trauma response at the Robert Wood Johnson University Hospital Level 1 trauma center were included in this study [28]. Blood samples (n=63) were collected under approval by Rutgers University Institutional Review Board (Pro2021001296) from trauma patients prior to any blood product administration or procedural intervention and processed to platelet-poor plasma as previously published [28]. Informed consent was obtained from patients or surrogates included in this study as per IRB approval. Patient samples were disposed of if any exclusion criteria were met, or approval from the patient or surrogate was not obtained in accordance with Rutgers IRB protocol [28]. Exclusion criteria include if a patient was below 18 years of age, below 110 pounds, was pregnant, or on systemic anticoagulants, was transferred from outside institutions or other facilities, a history of bleeding diathesis, or hereditary coagulopathy [28]. Plasma was stored at −80°C until testing. Clinical characteristics such as vital signs, time from injury, and laboratory values were obtained and recorded immediately for each patient upon hospital arrival, then abstracted from the institution’s trauma registry and electronic medical record for analysis (Table 1).

Healthy Donor Sample Preparation

Individual healthy donors (n=7) had 15 mL of whole blood drawn after informed consent was obtained as per Rutgers IRB approval (Pro2020001694) through venipuncture into 3.2% sodium citrate tubes (Grenier Bio, Kremsmunster, Austria). Platelet poor plasma was isolated as previously published and frozen at −80°C prior to testing and activated in the same manner as trauma samples for each testing method [28].

Plasma Rheological Testing

Platelet poor plasma (PPP) rheological properties were measured using a Malvern Kinexus Ultra Rheometer (Netzsch, Selb, Germany) after activation with 20 mM CaCl₂ and 0.6 Units/mL thrombin (final concentration). Rheological properties were tested using an oscillation strain test with a 20 mm parallel plate as previously described [29]. Clotting rate for rheological testing was defined as the increase in storage modulus from 10 to 70% of the maximum value within 30 minutes of clot activation to reduce the impact of drying.

Plasma Optical Turbidity Testing

PPP optical density was measured using a Molecular Devices SpectraMax Plus (Molecular Devices, San Jose, California, USA) at 37°C following activation of samples as described above for one hour as samples clotted and lysed [28]. Clotting rate for turbidity testing was defined as the increase in optical density from 10 to 70% of the maximum value within 30 minutes of clot activation to reduce the impact of drying. Fibrinolysis at one hour (%) was defined as the percent decrease from maximum optical density that was reached at one hour following sample activation with CaCl₂ and thrombin [28]. This measures the innate fibrinolysis occurring in patient plasma samples, identifying fibrinolytic activity occurring within patient samples at the point of blood draw across patient samples.

Confocal Microscopy

PPP clot fibrin network properties were measured using a Zeiss LSM 710 Confocal microscope (Zeiss, Oberkochen, Germany). Samples were prepared with 1% of the plasma volume replaced with 1.5mg/mL Alexa Fluor 488-labeled fibrinogen (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Clotting was initiated similar to other experiments and added to a clear bottom borosilicate glass 96-well plate, then transferred to the confocal stage. Samples were imaged using a 1024 x 1024 image resolution and a 40x water immersion objective. A 1.20 numerical aperture was used, with an image area of 213x213 micrometers, with image settings kept constant across samples. Images were taken as Z-stack images from 3 different areas for each sample within 1 hour of clot activation. For Z-stacks, 11 slices were imaged across a depth of 10 μm. Images were analyzed using maximum intensity projections, using the Java ImageJ software version 1.8.0_322. CZI files were used from the Zeiss Zen software. Fiber density was determined after making images binary and measuring total fiber area as a percent of image area. Fiber length was measured manually and calculated from the average of 36 fibers, and pore diameter was calculated from the average of 16 pores for each sample.

Plasma Tissue Plasminogen Activator Concentration Determination

Plasma tissue plasminogen activator (tPA) concentration was measured in patient and healthy donor plasma using an Abcam tPA human ELISA kit (ab1008914). A Molecular Devices Spectramax Plus was used to determine tPA concentration, measuring absorbance at 450 nm wavelength and interpolating sample concentration from the standard curve. Samples were diluted with a diluent buffer from a range of 1:8 parts plasma to diluent to 1:2000 parts plasma to diluent, with all samples run in duplicate.

Plasma D-Dimer Concentration Determination

Plasma D-dimer concentration was measured in patient and healthy donor plasma using an Abcam Human D-dimer ELISA kit (ab260076). A Molecular Devices Spectramax Plus was used to determine D-dimer concentration, measuring absorbance at 450 nm wavelength and interpolating sample concentration from the standard curve. Samples were diluted with a diluent buffer from a range of 1:10 parts plasma to diluent to 1:1200 parts plasma to diluent, with all samples run in duplicate.

Plasma Thrombin Generation Determination

Plasma thrombin generation was measured in patient and donor plasma using a Technothrombin Thrombin Generation Assay (Diapharma, OH, USA). A thrombin calibrator curve was tested prior to sample testing. Samples were prepared using 40 μL of sample, 50 μL of TGA fluorogenic substrate, and mixed with 10 μL TGA RC low trigger in duplicate. Sample fluorescence was measured at 360 nm excitation, 460 nm emission wavelength for 90 minutes at 1 minute intervals on a Molecular Devices SpectraMax M2 (Molecular Devices, San Jose, California, USA).

Plasma Fibrinogen Concentration Determination

Functional fibrinogen concentration was measured in plasma samples using the Dade Siemens Healthineers Fibrinogen Determination kit (Siemens, Erlangen, Germany) using a STAGO Start 4 coagulation analyzer [33]. Samples were incubated to 37° C and diluted 1:9 parts plasma to Owren’s Veronal buffer. 0.1 mL of Dade Thrombin Reagent was added and clotting time was measured in duplicate for each sample.

Plasma Proteomics Quantification

All plasma samples were directly digested with trypsin using SP3 method, Single-pot, solid-phase-enhanced sample preparation for proteomics experiments [34]. Patient and healthy donor plasma samples were analyzed by liquid chromatography-mass spectrometry (LC-MS) using a Nano LC-MS/MS (Dionex Ultimate 3000 RLSCnano system, Thermofisher, Waltham, Massachusetts, USA) interfaced with an Eclipse (Thermofisher, Waltham, Massachusetts, USA). Plasma samples were loaded onto a fused silica trap column Acclaim PepMap 100 (Thermofisher, Waltham, Massachusetts, USA). Following a 5 minute wash at 5 μL/min with 0.1% trifluoroacetic acid (TFA), the trap column was brought in-line with an analytical column (Nanoease MZ peptide BEH C18, 130A, 17um, Waters) for LC-MS. Peptides were fractionated at a flow rate of 300 nL/min using a segmented linear gradient: 4–15% B over 30 minutes (where A: 0.2% formic acid and B: 0.16% formic acid in 80% acetonitrile), 15–25% B over 40 minutes, 25–50% B over 44 minutes, and 50–90% B over 11 minutes. Solution B was then returned to 4% for 5 minutes before the next run. Eluted peptides were analyzed using a Data-Independent Acquisition (DIA) workflow. The MS scan range was set to 400–1200 m/z, with a resolution of 12,000, an AGC target of 3E6, and an automatic ion injection time. Eight m/z windows were used to sequentially isolate ions (AGC target 4E5, automatic ion injection time), which were fragmented in the C-trap with a relative collision energy of 30. MS/MS spectra were recorded at a resolution of 30,000. Raw data was analyzed using a predicted library from the UniProt human proteome FASTA file for a library-free search with DIA-NN 1.8.1* under recommended settings. The results were filtered with a posterior error probability (PEP) filter of <0.01 and a protein group Q value (PG.Q) filter of <0.01. Protein group MaxLFQ values were used for quantification [35]. A comprehensive analysis and description of proteomics methods used are reported [36].

Statistical Analysis

Statistical analysis was performed using GraphPad Prism 9.0. Analyzed groups were assessed for normal distribution using a D’Agostino and Pearson Normality test prior to analysis. All graphed data is shown as the mean ± standard error of the mean unless otherwise noted. Mann-Whitney and Unpaired T tests were used to compare two groups to find significant differences. When multiple comparisons were made across different variables between groups, p-values were adjusted using the Benjamini-Hochberg false discovery rate (FDR) correction in GraphPad Prism, with adjusted p-values reported and denoted as such. Brown-Forsythe and Welch One Way Analysis of Variance (ANOVA) for parametric data or Kruskal-Wallis tests for nonparametric data were used to determine group differences. Post hoc pairwise comparisons were conducted using Dunnett’s T3 or Dunn’s tests as appropriate. For Brown-Forsythe and Welch as well as Kruskal-Wallis tests, multiple pairwise comparison corrections for post hoc tests were automatically calculated with GraphPad Prism, and adjusted q-values are reported. Fisher’s exact tests were used to compare categorical variables between groups. Due to the limited number of independent comparisons, no multiple-comparison correction was applied to these comparisons. Spearman or Pearson correlation analysis was used to determine the correlation between factors measured with 95% confidence interval assessed. Multivariable linear regression was performed to identify the association between plasma D-dimer and Coagulation factor XIII alpha chain with the selected biomedical and clinical variables. Regression coefficients (β), p-value, 95% confidence intervals, and R² with other variables were reported. To assess the sensitivity of comparisons, several coagulation factors and fibrinolytic markers between patient groups were compared to identify the minimum detectable difference between groups. This was performed utilizing the observed sample sizes, means, and standard deviations, with a two-tailed significance level of (α) 0.05, and statistical power of 80%. Minimum detectable differences between key parameters are included in Table S1. Age, time from injury to blood draw, biological sex, trauma mechanism (blunt vs. penetrating), and number of injured body regions are potential confounding factors to this analysis. Post hoc power analysis was performed for several selected parameters, including clot stiffness, tPA, and thrombin generation lag time, using the observed group means and standard deviations. For each comparison, the minimum detectable effect size and corresponding statistical power were estimated using a two-sided α = 0.05. Utilizing these factors, with our focus on fibrinolytic and mechanical properties following TBI, we calculated the achieved power between key factors measured between Healthy donors and TBI patients, as well as No-TBI patients and TBI patients (Table S1).

Patient Demographics and Clinical Characteristics

We investigated the coagulation kinetics, clinical parameters, and plasma proteomics of 63 severely injured trauma patients following informed consent. Patient characteristics, injury type, and TBI parameters are shown in Table 1. Overall, 46% (n=29) of patients suffered traumatic brain injury as a part of their injury profile upon admission. Patients were retrospectively assessed for antiplatelet medication, with 56 patients (89%) receiving no antiplatelet medication prior to arrival, and 7 patients (11%) with unknown prehospital antiplatelet use. Of these, 5 were TBI patients (17% of TBI), and 2 were No-TBI patients (6% of No-TBI). Time from injury was available for 37 of the 63 patients (59%), with a median time from injury of 47 minutes (IQR 35-60) across the cohort (Table 1). Time from injury was available for 20 No-TBI patients (59%) and 17 TBI patients (59%). Median time from injury was 44 minutes for No-TBI and 50 minutes for TBI (Table 1). No differences were observed between time from injury, age, biological sex, race, and ethnicity makeup between TBI and No-TBI patients (Table 1). Patients experiencing TBI were more likely to have experienced blunt force injury than penetrating injury, as compared to No-TBI patients. Nearly 30% of the patient cohort died from their injuries after hospital admission; however, patients experiencing TBI had greater mortality (52%) than patients with No-TBI (12%), indicating a critical contribution of TBI to mortality (p<0.001)(Table 1). These TBI patients were also more severely injured, with a mean Injury Severity Score (ISS) of 26 compared to 14 in the No-TBI group (q<0.01)(Table 1). Lastly, to assess the extent of neurological impairment, patient Glasgow Coma Scale (GCS) was compared, with a low GCS indicating higher TBI severity, and a higher GCS indicating low TBI severity. The TBI cohort had the lowest mean GCS upon admission at 3, compared to a GCS of 15 in the No-TBI cohort (q<0.0001), along with a higher mean Head Abbreviated Injury Scale (AIS) at 4, compared to 0.5 in No-TBI patients (q<0.0001) (Table 1).

TBI Patients Are Predisposed to Early Prehospital Fibrinolysis

Healthy donor and patient plasma were subjected to a diverse array of testing assays. The most notable changes observed between patient groups were in their fibrinolytic activity and intrinsic coagulation, indicating unique differences in TBI-induced versus extracranial-induced coagulopathy. Both groups exhibited higher prehospital fibrinolysis than healthy donors, indicated by higher D-dimer concentration. TBI patients in particular exhibited the highest D-dimer (p<0.05) (Figure 1A). When measuring spontaneous percent reduction in maximum optical density in trauma patient samples, fibrinolysis one hour after clotting activation was similar between both patient groups, with several patients in each group exhibiting elevated fibrinolysis (p<0.05)(Figure 1B). This indicates that innate fibrinolytic potential within patient samples at the point of blood draw was similar between the groups, even though D-dimer differences indicate higher pre-hospital lysis in the TBI population. An elevated tPA concentration was measured in both patient groups compared to healthy donors, although similar to each other (p<0.001)(Figure 1C). To quantify the temporal change of fibrinolytic potential in the patient groups, D-dimer, fibrinolysis at one hour, and tPA concentration were normalized to healthy donors. This indicates fibrinolytic potential in the prehospital, at admission, and post-admission time periods (Figure 1A–D).

Activated partial thromboplastin time (aPTT) and prothrombin time (PT) were similar between patient groups, while TBI patients were more severely injured (p<0.0001) (Figure 1E,F)(Figure 2L)(Table 1). Proteomic profiling identified changes in the relative levels of intrinsic coagulation factors in the patient and donor cohorts [36]. Von Willebrand Factor level was lower in TBI patients than in No-TBI, even though both groups exhibited a moderate increase from healthy baseline levels (p<0.05)(Figure 1G). Similarly, Coagulation Factor VIII was higher in both groups than the healthy baseline, but again, TBI patients had a lower concentration than No-TBI (p<0.01)(Figure 1H). In addition, Coagulation Factor XIII alpha chain was reduced in both groups compared to healthy donors, with TBI patients showing a greater reduction from healthy donors than No-TBI, which may be due to increased consumption depleting these protein levels (p<0.001)(Figure 1I).

Kinetic, mechanical, thrombin generation assays, and fibrin network quantification assays detected only limited differences between No-TBI patients and TBI patients. Both TBI and No-TBI cohorts exhibited lower maximum clot stiffness and fibrinogen concentration determined by rheological and functional fibrinogen testing compared to healthy donors (Figure 2A–C). Patients had similar fibrin fiber density, fibrin fiber length and pore diameter (Figure 2D). No-TBI patients had lower endogenous thrombin potential (ETP) than healthy donors, whereas TBI patients did not exhibit a reduction in ETP. Both patient groups exhibited longer time to thrombin generation (lag time), lower peak thrombin, and longer time to peak thrombin than healthy donors (p<0.05)(Figure 2E–I). Lastly, several clinical factors measured upon hospital arrival were similar between groups as well, including hemoglobin, lactate, and prothrombin time (Figure 2J–L).

TBI Phenotype Influences Unique Changes in Coagulation

To determine the impact of TBI on coagulation in the context of other injuries, patients were grouped using the Abbreviated Injury Scale (AIS), an anatomic measure of injury severity that is widely used for severity stratification [37]. A head AIS of 3 or greater designates moderate TBI or worse [38]. 34 patients were categorized as No-TBI, sustaining injuries only to areas other than the head. 15 patients were categorized as Isolated-TBI (Iso-TBI), with a Head AIS ≥ 3 and all extracranial AIS < 3. 14 patients were categorized as Poly-TBI, with a Head AIS ≥ 3 and any extracranial AIS ≥ 3, representing those with polytrauma including TBI [39]. Time from injury was similar between each group, with a median time of 43.5, 45, and 50 for No-TBI, Poly-TBI, and Iso-TBI, respectively. Compared to No-TBI patients, both phenotypes of TBI exhibited increased D-dimer levels and prehospital coagulation factor consumption (p<0.05)(Figure 3A). Each group exhibited similar spontaneous fibrinolysis at one hour and tPA concentration, further indicating variations in fibrinolytic activity based on phenotype and timeline of traumatic injury, but limited differences in other kinetic factors and no significant changes to fibrin network properties (Figure 3B–D)(Figure S1A–L).

APTT was similar between all three trauma groups, and Poly-TBI had significantly higher ISS than No-TBI patients (p<0.05)(Figure 3E,F). Poly-TBI patients showed similar levels of each factor to No-TBI patients, while Iso-TBI patients had reduced levels of each factor (Figure 3G,H). Coagulation Factor XIII alpha chain concentrations were similar between each group, though Iso-TBI and Poly-TBI were moderately reduced compared to No-TBI (Figure 3I).

Severe TBI is Associated With Exacerbated Coagulation Dysfunction

The rapid detection of TBI in injured patients is challenging, as prehospital tools to determine the extent of brain injury remain limited. However, the Glasgow Coma Score has been shown to reliably predict the extent of impaired consciousness after injury and correlates to the severity of TBI in the absence of more direct analysis methods [40,41]. GCS was examined to see if it is a useful prognostic marker to determine coagulation changes prior to more quantitative testing. For use as a rapid assessment marker, GCS upon arrival was used to categorize the severity of TBI in patients upon arrival. Patients were grouped as No-TBI (n=34), Mild or Moderate TBI (n=6) with GCS between 9-15, and Severe TBI (n=23) with a GCS 3-8 (Figure 4) [42,43]. Time from injury was similar between each group, with median time from injury recorded as 43.5, 60, and 48.5 minutes for No-TBI, Moderate TBI, and Severe TBI, respectively.

D-dimer was significantly higher in Severe TBI patients compared to No-TBI (Figure 4A). Spontaneous fibrinolysis at one hour and tPA concentration were similar between all three patient groups (Figure 4B–D). Upon admission, Severe TBI patients had the highest mean aPTT, with significantly higher aPTT than Moderate TBI and No-TBI patients, paired with higher ISS compared to No-TBI patients (p<0.05)(Figure 4E,F). Severe TBI patients had similar von Willebrand Factor levels as No-TBI but reduced Factor VIII levels, which may lead to this aPTT elongation (Figure 4G,H). Furthermore, Severe TBI exhibited the lowest coagulation factor XIII alpha chain levels, significantly reduced compared to No-TBI patients, indicating further consumption after trauma (p<0.05)(Figure 4I). Similar to TBI phenotypes, no changes were observed between increasing TBI severity patient groups in regard to clot mechanical properties, thrombin generation, or fibrin network properties (Figure S2A–I), with only PT being elongated in Severe TBI compared to No-TBI (p<0.05)(Figure S2J–L).

Plasma Proteomics Unveil Intrinsic Coagulation Deficiencies in TBI Patients

Of the patient cohort included in this study, 61 patients (97%) had plasma proteomic analysis assessed in addition to all 7 of the healthy donors. Proteomic results were compared between each group as described previously. For each protein analyzed, both groups. No-TBI and TBI, had at least 6 patients in each group with detectable levels of the protein to ensure sufficient replicates per analyzed protein. Due to the availability of GCS definition upon hospital arrival, and its strong negative correlation to Head AIS in the patient cohort (r = −0.595, p<0.001)(Table S2) this was compared to measured parameters. An exploratory analysis was performed comparing the relationship between patient Glasgow Coma Score and coagulation kinetics, markers of fibrinolysis, confocal and thrombin generation parameters, clinical measurements, and coagulation-related proteomic results. Each factor was compared to GCS using a Spearman’s correlation with 95% confidence intervals assessed (Table S2–4). These results should be interpreted with caution as they are individual comparisons of proteins and biomarkers across a large number of tested associations, with a patient cohort of 63 individuals, however these associations provide novel hypothesis-generating results on TBI influences. Increasing TBI severity, as seen with a reduced GCS, exhibited significant association to vital signs, concentration of intrinsic coagulation factors, and several pro- and antifibrinolytic markers measured with ELISA and proteomic analysis (Table S2–4). A U-plot of compared factors shows the relationship between increasing and decreasing GCS and the measured factors, with the x-axis showing the correlation r value (Figure 5A). All correlations, including sample size, correlation r value, and p-value for significance, are listed in Supplemental Tables 2–4. Most notably, factors related to coagulation were highly correlated to GCS, such as aPTT (p<0.001), D-dimer (p<0.0001), and factor XIII alpha chain (p<0.05)(Figure 5A)(Table S2–4). In addition, spontaneous fibrinolysis at one hour was positively correlated with increasing GCS (lower TBI severity), indicating that at the time of blood draw, more fibrinolysis occurs spontaneously in patients with lower TBI severity (p<0.05)(Figure 5A)(Table S3).

Lastly, several protein markers that were detected showed significant positive correlation to GCS. Coagulation factor VIII, factor X, factor XIII alpha chain, and vitamin K dependent protein C all positively correlate to an increase in GCS (p<0.05)(Figure 5A)(Table S4). Taken together, these results indicate a substantial association of decreasing GCS and increasing TBI severity on decreased antifibrinolytic levels, leading to increased consumption of intrinsic coagulation factors and prolonged aPTT upon admission to the hospital. Due to the wide-ranging interactions between factors, a hypothesis-generating analysis of the relationship between fibrinolytic markers, clotting kinetics and mechanics and coagulation proteins was conducted. A Spearman correlation matrix was calculated to compare how these factors were related in our patient cohort, across the entire patient population, in only No-TBI patients and in only TBI patients (Figure 5B–D). This exploratory analysis observed unique differences in the relationship between factors such as Coagulation factor XIII alpha chain correlated to plasminogen, and between PAI-1 correlated to vitamin K-dependent protein C, which differed between No-TBI and TBI patients. Although univariate analysis showed that D-dimer and Factor XIII alpha chain were significantly correlated with GCS, negatively and positively respectively, this relationship was also observed to also be significant when correcting for patient Age, ISS, and biological sex (p<0.01)(Figure S4A,B). In this analysis, none of the other covariates showed a significant relationship to the outcome factors, D-dimer and FXIII-alpha chain (Figure S4A,B). These associations indicated larger relationships between prehospital fibrinolysis in TBI patients and their relation to reduced protein C and protein S levels than in No-TBI patients, further supporting TBI specific activation of these factors (Figure 5B–D).

Predisposition of TBI to Early Fibrinolysis

The TBI and No-TBI patient groups were demographically similar; however, TBI patients exhibited significantly higher injury severity scores and an increased mortality. Potentially contributing to this mortality is enhanced fibrinolysis seen early after injury in TBI patients. Following trauma, tissue factor abundant in the brain is exposed to blood, inducing fibrin clot formation [4,16,17,44]. After this, a subsequent state of increased fibrinolysis is triggered, with one animal study identifying tPA as the early trigger. This peaks immediately post-injury and declines rapidly, while urokinase plasminogen activator peaks later, around 9 hours post-injury, contributing to elevated D-dimer as a marker of fibrin clot degradation [45]. Elevated D-dimer observed in both TBI and No-TBI patients, with the highest levels in TBI, supports the notion that TBI patients experience early fibrinolysis, consistent with previous studies [16,18,20,46]. This early fibrinolytic response has been linked to progressive hemorrhagic injury and worse outcomes [9,47,48]. Importantly, we observed that Factor XIII alpha chain levels were lowest in TBI patients; given that factor XIII deficiency is known to increase bleeding risk and reduce clot integrity [49], its reduction in TBI patients may exacerbate the effects of early fibrinolysis to overall progressive hemorrhagic complications.

While our single time point of D-dimer collection precludes its use in assessing fibrinolysis dynamics, tPA and spontaneous fibrinolysis at one hour were not different between TBI and No-TBI patients, indicating this heightened fibrinolysis at hospital admission is consistent with a transient phase occurring early after injury in TBI (Figure 1D). This is consistent with findings from Leeper et al, in which states of low fibrinolysis (LY30<0.9%) emerged as the dominant phenotype as soon as 1 hour post-TBI [25]. Notably, our observed changes in turbidity aid in quantifying plasma-specific fibrinolysis in TBI, which may be masked in whole blood samples measured by viscoelastic assays.

Isolated TBI Exerts Distinct Coagulation Patterns

Previous studies have noted that TBI was not independently associated with coagulopathy, but rather it was the addition of other injuries and shock that led to coagulopathy [50,51]. However, by differentiating Iso-TBI from TBI with extracranial injury, we quantified several unique contributions of intracranial and extracranial injury on coagulopathy in our patient cohort. Iso-TBI patients had uniquely lower ETP, longer thrombin lag time, lower peak thrombin, and slow thrombin generation rate when compared to healthy donors, suggesting an inability to generate thrombin that is not present in other patient groups with extracranial injury (Figure S1). These findings align with previous work from Bohm et al in the CENTER-TBI study that demonstrated a reduction in activity of factors V, VIII, IX, and XIII as well as a decreased thrombin formation peak height and ETP in coagulopathic isolated TBI patients [52]. These observations from the CENTER-TBI subgroup, in conjunction with our cohort, help elucidate how isolated TBI drives coagulation factor depletion and prehospital fibrinolysis. Importantly, our study advances these prior observations by linking thrombin generation abnormalities to plasma rheology and turbidity, providing mechanistic insights into clot dynamics that may not be detected in whole blood viscoelastic testing. Our results further elaborate that extracranial injury, whether on its own in No-TBI patients, or in combination with TBI in Poly-TBI patients, is associated with a larger activation or release of procoagulant factors such as factors VIII and von Willebrand factor (Figure 3).

Along with this, the lower concentrations of circulating vWF and factor VIII observed in our Iso-TBI patients, relative to Poly-TBI, could be attributable to increased consumption, inadequate activation, or inadequate release in the absence of extracranial injury. Given that cryoprecipitate provides concentrated fibrinogen, factor VIII, vWF, and factor XIII, its use could plausibly address the deficits seen in isolated TBI. In a substudy of the FEISTY trial, cryoprecipitate transfusion restored fibrin clot structure to control levels and attenuated excessive clot lysis [53]. Despite this mechanistic potential, the CRYOSTAT-2 trial demonstrated that empirical early high-dose cryoprecipitate did not translate to mortality or adverse events benefit in trauma patients, including those with severe TBI [54]. Given the unique deficits seen in isolated TBI, evaluation of cryoprecipitate in preventing intracranial hemorrhage expansion in this specific population presents a promising area for investigation.

We observed no difference in the transient state of elevated fibrinolysis described earlier when differentiating between Iso-TBI and Poly-TBI, suggesting that this response is primarily driven by intracranial rather than extracranial injury. Samuels et al. noted no differences in fibrinolysis at 30 minutes between isolated severe TBI, severe TBI with extracranial injury, and No-TBI trauma patients, further supporting the transient nature of TBI induced fibrinolysis [55]. A heightened fibrinolytic state following TBI is well-established, driven primarily by tPA release and likely amplified by activation of the protein C pathway [45,56,57]. Reduced protein C and protein S in coagulopathic isolated TBI patients has been reported, suggesting increased activation and consumption of these anticoagulants, thereby promoting fibrinolysis [21,52]. Activated protein C complexes with active PAI-1, a fibrinolysis inhibitor, leading to PAI-1 inactivation and degradation [58,59]. In our TBI cohort, protein C and PAI-1 levels were positively correlated, while D-dimer and protein C were negatively correlated; these patterns were not observed in No-TBI patients. In addition, GCS was positively correlated with protein C and PAI-1. These relationships suggest that TBI may promote protein C activation, resulting in complexation with and inactivation of PAI-1, contributing to the concomitant reduction of both proteins that becomes more pronounced with increasing TBI severity.

The pathway from TBI to protein C activation is likely multifactorial. Hypoperfusion and shock trigger protein C activation, which inhibits thrombin generation and likely contributes to PAI-1 inactivation, further contributing to fibrinolysis [60,61]. One study found that without hypoperfusion as determined by base deficit, there was no effect of increasing brain injury severity on protein C activation [62]. While this raises the question of whether hypoperfusion and shock drive coagulopathy more than TBI itself, Wada et al observed that lactate levels in isolated TBI patients were not correlated with systolic blood pressure. They did however, show a positive correlation with DIC scores, suggesting that hypoperfusion is driven by DIC rather than low blood pressure [21]. In our cohort, protein C was negatively correlated with lactate in Iso-TBI patients, suggesting that with worsening tissue hypoperfusion, there is increased protein C activation and subsequent consumption; however, we don’t see this correlation in Poly-TBI (Figure S5). It remains unclear whether this difference is due to the direct release of coagulopathic agents from the injured brain in isolated TBI or multisystemic coagulopathy drivers in Poly-TBI.

TBI Severity is Associated With the Greatest Coagulation Dysfunction

Stratifying by head injury severity revealed that Severe TBI was associated with greater consumption of fibrinogen in the prehospital environment, without a corresponding difference in tPA or spontaneous fibrinolysis at one hour, along with reduced concentrations of factors VIII and XIII alpha chain. These findings underscore the role of intracranial over extracranial injury in driving systemic coagulopathy and suggest that the extent of early fibrinolytic activation may intensify with greater TBI severity.

Although our platelet-poor plasma assays do not capture whole-blood effects, these severity-dependent patterns provide some insight into the observations from the CRASH-3 trial, which demonstrated a mortality benefit to TXA when administered within three hours of injury in patients with mild to moderate TBI, with earlier treatment proving more effective than later treatment [15,63]. Given the time-sensitive efficacy of TXA, one might assume faster administration could improve outcomes in severe TBI. However, Bossers et al. observed an increased odds of mortality when prehospital TXA was administered in isolated severe TBI [64]. Several studies have described depressed fibrinolysis in TBI patients [23,24,65]. In our patient cohort, we observed a positive correlation between GCS and spontaneous fibrinolysis at one hour using turbidity, suggesting fibrinolytic depression proportional to TBI severity. In this context, antifibrinolytic therapy is likely redundant with endogenous fibrinolysis inhibitors. Supporting this, Moore et al observed that, in general adult trauma patients, TXA did not significantly impact mortality in those with depressed fibrinolysis [66].

We also observed the significant consumption of factor XIII alpha chains in severe TBI, which suggests that clots in these patients will be more susceptible to breakdown. FXIIIa is known to crosslink fibrin as well as fibrin with α2-antiplasmin, the latter of which has been reported to induce clot resistance to fibrinolysis [67,68]. Our findings suggest that severe TBI patients exhibit weaker clots and more suppressed fibrinolysis, which is consistent with clinical phenotypes and supports knowledge surrounding the lack of benefit from TXA administration.

Limitations

This study has several limitations. First, even though our study detects changes in fibrinolytic function following trauma based on D-dimer, we were only able to quantify time from injury to hospital admission and blood draw for 59% of our patient cohort. Although there were no significant differences in time from injury in any of the subgroup comparisons made in this study, the unrecorded patient information may present a potential confounder if these were to alter the analysis such that the time from injury difference became statistically significant. Future studies should aim to standardize injury-to-sampling times as a covariate in their analysis to better account for and develop an understanding behind these temporal hemostatic changes. As we were unable to identify time from injury to blood draw in 41% of patients, this missing information limits the strength of conclusions in temporal coagulation changes observed here. Second, since our study focuses on a single blood draw following admission, we are unable to determine the temporal changes in coagulation. This may dictate the patient’s transition into a thromboembolic state. Further heterogeneity may arise from the diverse array and severity of extracranial injury types in both the No-TBI and Poly-TBI trauma groups, which may lead to further changes in coagulation that cannot be accounted for. Moreover, grouping moderate and severe TBI together within Iso-TBI and Poly-TBI categories may obscure the role of TBI severity. Conversely, when stratifying by TBI severity, the impact of extracranial injury captured by the Poly-TBI classification may be lost. Additionally, given the variability in how similar studies establish their cutoffs can limit direct comparison.

Our study implemented rheological and turbidity testing of platelet-poor plasma clots. These assays provide mechanical and kinetic insight into the formation of fibrin fiber networks, and their resulting mechanical stiffness and fibrinolytic activity as the clots form. However, even though these assays provide analogous results to clinical viscoelastic assays, we are unable to directly compare our platelet free plasma results to whole blood assays. In addition, rheological assays have limited sensitivity in detecting subtle fibrinolytic activity that may be present in small or localized injuries. As a result, they may overlook coagulation processes occurring within the microvasculature or strictly intracranial regions in isolated TBI, which may not produce measurable systemic effects at the time of blood draw [69,70]. Along with this, our testing did not include the impact of platelets, red blood cells and other whole blood components, all of which are known to substantially influence clot mechanics and formation. These rheological studies remain mechanistically informative, but are less directly translatable to clinical assays such as TEG and ROTEM as it removes the contribution of other cellular components to the coagulation process. Further studies should be conducted utilizing these standardized viscoelastic tests, with larger patient cohorts to increase the statistical rigor of analysis and patient representation in each group.

Following patient sample collection and analysis, we calculated minimum detectable differences between vWF, FXIII-alpha chain and D-dimer, to assess the study’s statistical power for comparisons. These analyses are constrained by variability in the trauma cohort and unequal group sizes. Due to this, several smaller but biologically relevant differences in investigated factors may not have been detected in our study. Proteomic analysis generated substantial information on the relative concentration of different coagulation factors in our patient cohort; these relative levels, however, do not directly measure protein activity itself and therefore limit the direct comparison of these levels to coagulation function. Lastly, when comparing more specific forms of TBI, this reduces the patient population in each respective group, leading to relatively small sample sizes among these patients, overall limiting the generalizability of these results to the overall TBI population.