Circulating extracellular vesicle tissue factor activity is associated with disseminated intravascular coagulation in 2 cohorts of patients with sepsis

Abstract

Background

Disseminated intravascular coagulation (DIC) occurs in a variety of diseases and syndromes. Platelet count, prothrombin time, and levels of fibrinogen and D-dimer are used to determine if a patient has DIC. The tissue factor (TF)/factor (F)VIIa complex plays a central role in the activation of coagulation in sepsis. In Gram-negative sepsis, lipopolysaccharide induces TF expression in monocytes and release TF-positive extracellular vesicles (EVs).

Objective

This study determined whether EV-TF activity is associated with DIC in patients with sepsis.

Methods

We studied 2 cohorts of patients with sepsis: a cohort of sequential patients that included 143 patients without DIC and 45 patients with DIC, and a cohort of selected patients that included 39 patients without DIC and 36 patients with DIC. The first cohort had longitudinal samples taken, and the second cohort had samples taken from pre-DIC patients (who developed DIC during the study). Healthy subjects were used as controls. Plasma EV-TF activity was measured using an in-house assay. We used the updated 2025 International Society on Thrombosis and Haemostasis (ISTH) DIC score.

Results

EV-TF activity was significantly higher in patients with sepsis than in healthy controls. There was no significant difference in the level of EV-TF activity between pre-DIC patients and patients without DIC. Importantly, EV-TF activity was associated with DIC in both cohorts of patients with sepsis. We observed a time-dependent decrease in EV-TF activity in patients with DIC.

Conclusion

Our study suggests that increased levels of circulating EV-TF activity contribute to the activation of coagulation in patients with sepsis.

Essentials

• •Tissue factor (TF) plays a central role in the activation of coagulation in sepsis.
• •We assessed whether extracellular vesicle (EV)-TF activity is associated with coagulopathy in sepsis.
• •EV-TF activity was associated with disseminated intravascular coagulation (DIC).
• •We observed a time-dependent decrease in EV-TF activity in patients with DIC.

1. Introduction

Disseminated intravascular coagulation (DIC) occurs in various diseases and syndromes, including sepsis, and can be thrombotic or hemorrhagic [1]. In 2001, the International Society on Thrombosis and Haemostasis (ISTH) Scientific and Standardization Committee (SSC) on DIC described DIC as “an acquired syndrome characterized by the intravascular activation of coagulation that can produce organ dysfunction” [2]. In 2025, the committee updated the definition to “an acquired, life-threatening intravascular disorder characterized by systemic coagulation activation, dysregulated fibrinolysis, and endothelial injury, resulting in microthrombosis.” DIC arises from various underlying etiologies, such as sepsis, and progresses from an asymptomatic early phase to an advanced phase with thrombosis/hemorrhage and/or organ dysfunction and death [3,4].

Sepsis is a severe syndrome where the body’s excessive response to an infection can lead to organ failure, shock, and death [5]. Bacterial infection is the most common cause of sepsis. Despite the treatment with broad-spectrum antibiotics and other supportive care, the mortality rate remains high (15%-30%), with sepsis accounting for 20% of all deaths globally [6,7]. Activation of the coagulation cascade occurs in sepsis as part of the host response to the infection [8]. Importantly, up to 40% of patients with sepsis develop DIC [9].

The ISTH DIC score uses platelet count, prothrombin time (PT), and levels of fibrinogen and D-dimer to determine whether a patient has DIC [2]. A score of ≥5 is defined as DIC. In 2025, the ISTH DIC score was modified to better define the cutoffs for the increase in D-dimer [10]. A modified ISTH score does not include fibrinogen [11]. Other scores have been developed to be specifically designed for patients with sepsis, such as the sepsis-induced coagulopathy (SIC) score and the sepsis-associated coagulopathy score [12,13]. These scores do not use D-dimer or fibrinogen. The SIC score uses a modified sequential organ failure assessment (SOFA). It has been proposed that the SIC score detects early-phase DIC [10]. However, 1 study concluded that the SIC score was not better than the ISTH DIC score at diagnosing DIC [11].

The tissue factor (TF)-factor (F)VIIa complex is the physiological activator of the coagulation protease cascade [14]. Under normal conditions, TF is expressed by perivascular cells and at organ and body surfaces where it plays an essential role in hemostasis [15]. However, under pathological conditions, TF contributes to thrombosis [14]. For instance, lipopolysaccharide from Gram-negative bacteria is a potent activator of monocytes and induces TF expression and the release of TF-positive extracellular vesicles (EVs) [[16], [17], [18], [19]]. TF plays a central role in the activation of coagulation in endotoxemia and sepsis. In a human endotoxemia model in which healthy subjects are injected with lipopolysaccharide, TF mRNA expression is increased in monocytes, and there is an increase in plasma EV-TF activity that coincides with the activation of coagulation [[20], [21], [22], [23]]. Furthermore, administration of an anti-TF antibody inhibited the increase in thrombin generation in a chimpanzee endotoxemia model [24]. In a mouse model of endotoxemia, either a genetic reduction in TF gene expression or administration of an anti-TF antibody reduced the activation of coagulation [25,26]. Finally, administration of an anti-TF antibody attenuated the coagulopathy and protected baboons against an LD100 dose of Escherichia coli [27].

We and others have shown elevated levels of circulating EV-TF activity in a human endotoxemia model and in patients with sepsis compared with those in healthy controls [22,23,[28], [29], [30]]. Two small studies compared the levels of EV-TF activity in patients with sepsis with and without DIC using the 2001 ISTH DIC score [31,32]. Tobiasch et al. [31] found a nonsignificant increase in EV-TF activity in the patients with DIC compared with that in patients without DIC. They also reported a time-dependent increase in EV-TF activity in patients with sepsis with DIC. Meng et al. [32] found significantly higher levels of EV-TF in patients with DIC than in patients without DIC. In contrast to the study by Tobiasch et al. [31], Meng et al. [30] observed a time-dependent decrease in the level of EV-TF in patients with sepsis and DIC.

In this study, we evaluated levels of EV-TF activity in 2 independent cohorts of patients with sepsis: Marseille, France [33], and Hamilton, Ontario, Canada [34]. The Marseille cohort also had longitudinal samples taken, whereas the Hamilton cohort had samples taken from pre-DIC patients (patients with sepsis who did not present with DIC on the first day but subsequently developed DIC during the study). The novelty of our study is that it compares levels of EV-TF activity in sepsis patients with or without DIC (measured using the 2025 ISTH DIC score) or SIC in 2 independent cohorts of patients. Previously, we used the Marseille cohort samples to show that elevated levels of EV-TF activity were associated with increased mortality in patients with sepsis [33,34]. In contrast, we used the Hamilton cohort samples to identify levels of protein C and antithrombin as biomarkers of a pre-DIC state. In the current study, we calculated the 2025 ISTH DIC scores and SIC scores in both cohorts and measured levels of plasma EV-TF activity in the Hamilton cohort. We evaluated the association of EV-TF activity with DIC and SIC in samples from both cohorts.

2. Methods

2.1. Patients and blood collection

2.1.1. Marseille cohort

Whole blood was collected from 188 patients with sepsis from a multicenter prospective study (MISS study; NCT02062970) [33]. The study was approved by the local medical ethics committee. Written informed consent was obtained from the patients or their relatives. Patients were followed up for 90 days after their inclusion on the study. Patients were eligible if they were older than 18 years and met the sepsis and septic shock criteria. Patients were excluded if they were treated with anticoagulants at curative dose, pregnant or breastfeeding, or immunosuppressed.

2.1.2. Hamilton cohort

Whole blood was collected from 75 intensive care unit patients with sepsis [34]. These patients were a subset of 357 subjects enrolled in a previous study (DYNAMICS study; NCT01355042) [34]. The study was approved by the research ethics boards of the participating centers. Patients were eligible if they were older than 18 years and had a confirmed or suspected infection, had at least 1 dysfunctional organ system, showed at least 3 signs of systemic inflammatory response syndrome, and were expected to remain in the intensive care unit for 72 hours or more. Patients were excluded in they were pregnant or breastfeeding or were receiving palliative care.

2.2. Blood collection and plasma preparation

2.2.1. Marseille cohort

Whole blood was collected within the first 12 hours (day 0), at 24 hours (day 1), and at 48 hours (day 2) after inclusion. Blood was collected into 4.5-mL Vacutainer tubes containing 0.105 mol/L sodium citrate (BD Diagnostics). The first few milliliters were discarded. Platelet-free plasma (PFP) was prepared by 2 successive centrifugations (2500 × g for 15 minutes at room temperature). Plasma was frozen and stored at −80 °C.

2.2.2. Hamilton cohort

Whole blood was collected from patients daily for the first 7 days and then weekly between day 7 and 28. Patients were identified as pre-DIC if they did not present with DIC on the first day but developed DIC during the first 7 days. Blood was collected from existing arterial or venous lines (or by venipuncture with a 20-gauge needle) into 4.5-mL Vacutainer tubes containing 0.105 mol/L sodium citrate (BD Diagnostics) and processed within 2 hours of collection. Blood was centrifuged at 1700 × g for 10 minutes at 20 °C to generate platelet-poor plasma (PPP). Plasma was frozen and stored at −80 °C.

Liver disease in the 2 cohorts was mild, and therefore, we did not exclude these patients from the analysis. The number of patients with liver disease in patients with and without DIC were not significantly different (Marseille cohort, 11% vs 19%; Hamilton cohort, 10% vs 8%). Only 2 of the 29 patients with liver disease in the 2 cohorts had significantly decreased fibrinogen levels, and these 2 patients already had a higher DIC score (≥5) due to the other parameters.

2.2.3. Healthy subjects

Blood was collected from 10 healthy subjects and PPP prepared using the Hamilton protocol. We did not attempt to match cases and controls.

2.3. DIC score

The DIC scores were calculated for both cohorts based on the 2025 ISTH DIC score guidelines [10]. Only patients who had all items for the DIC score were included for these analysis (n = 188 for Marseille cohort and n = 75 for Hamilton cohort).

2.3.1. Marseille cohort

Routine blood parameters and hemostatic tests included blood counts, PT, fibrinogen, and D-dimer (STA-Liatest D-Di Plus; Stago). The range of the D-dimer assay was 0.27 to 20 μg/mL fibrinogen equivalent units (FEU), and a cutoff was 0.5 μg/mL FEU for the upper limit of healthy subjects.

2.3.2. Hamilton cohort

Platelet counts and PT–international normalized ratio (INR) values were extracted from clinical charts. Plasma levels of D-dimer and fibrinogen were measured by ELISAs (RayBio and Affinity Biologicals, respectively). The range of the D-dimer ELISA was 0.08 to 60 pg/mL FEU. Plasma samples were diluted. A cutoff was 0.5 μg/mL for the upper limit of healthy subjects.

2.4. SIC score

SIC scores were calculated for both cohorts as described [12]. The SIC score includes a modified SOFA score based on only 4 systems (respiratory, cardiovascular, hepatic, and renal). Only patients who had all items for the SIC score were included for these analyses (n = 147 for Marseille cohort and n = 72 for Hamilton cohort).

2.5. EV-TF activity

2.5.1. Marseille cohort

EV-TF activity was measured as described [35]. Briefly, EVs were isolated from 500 μL of plasma by centrifugation at 24,000 × g for 60 minutes at room temperature, washed with buffer and resuspended in buffer. Samples were preincubated for 30 minutes at 37 °C with either an anti-TF antibody (10 μg/mL final; clone SBTF1; BioCytex) or a control antibody (10 μg/mL final; clone 1-DNP 2H11-2H12; BioCytex). Next, a mixture of FVII, FX, and calcium was added, and the samples were incubated for 2 hours at 37 °C. The reaction was stopped by addition of EDTA. Factor Xa generation was determined as described earlier. After incubation with a FXa chromogenic substrate (Pefachrome FXa 8595; Pentapharm; Cat. 085-27) for 15 minutes at 37 °C, the absorbance at 405 nm was measured. TF-dependent FXa generation was determined by subtracting the FXa generation in the presence of anti-TF antibody (TF-independent FXa) from the FXa generation in the presence of IgG control (total FXa generation). EV-TF activity was expressed as pictograms per milliliter.

2.5.2. Hamilton cohort

EV-TF activity was measured using the Chapel Hill assay as described [36]. Briefly, EVs were isolated from 100 μL of plasma diluted in 1 mL of wash buffer by centrifugation at 20,000 × g for 60 minutes at 4 °C. The pellet was washed once with 1 mL of wash buffer and centrifuged again at 20,000 × g for 60 minutes at 4 °C. The pellet was resuspended in 100 μL of wash buffer and used for the assay. EVs samples were added to a 96-well microplate and incubated with an anti-TF inhibitory antibody (HTF-1; BD Biosciences; Cat. 550252) or a control IgG antibody (Sigma Aldrich; Cat. 15381). All other steps of the assay are as described earlier.

2.6. Statistical analysis

Results are presented as median (interquartile range) for continuous variables and frequencies for categorical variables. Normal distribution and homoscedasticity of the results were analyzed using the Shapiro–Wilk test and Levene’s test, respectively. The Mann–Whitney U-test was used to compare values between groups for continuous variables. For categorical variables, Fisher test was used. For longitudinal data, comparison of values from the same patients on days 0 and 2 was performed using Wilcoxon matched-pairs rank test. Multiple logistic regression receiver-operating characteristic (ROC) area under the curve (AUC) was calculated to determine the predictive power of different parameters for the DIC and SIC scores. Correlations were determined using Spearman’s test. Data were considered statistically significant when P < .05 and analyzed with Prism version 10.6.1 (GraphPad Software).

3. Results

3.1. Baseline characteristics of the 2 cohorts of patients with sepsis

Table 1 shows the baseline characteristics of the patients with sepsis for the Marseille and Hamilton cohorts. Information on race/ethnicity of the patients was not available. No differences were observed between the 2 cohorts for sex, age, occurrence of diabetes, cancer, cardiac failure, renal insufficiency, liver disease, lactate levels, and mortality. However, a few differences were observed between the cohorts. EV-TF activity was higher in the Marseille cohort than in the Hamilton cohort. Importantly, the Hamilton cohort had significantly higher PT, D-dimer, white blood cell count, DIC score, and SOFA score and significantly lower fibrinogen and platelet count than the Marseille cohort. This suggests that the patients in the Hamilton cohort had more severe sepsis than those in the Marseille cohort.

Table 1.

Baseline characteristics of the 2 cohorts of patients with sepsis

Characteristic	Marseille (n = 188)	Hamilton (n = 75)	P
Sex			NS
Female	70	35
Male	118	40
% Female	37.25	46.67
Age (y)	67.00 (58.25-73.75)	65.00 (53.00-77.00)	NS
Diabetes (%)	28.41	25.33	NS
Cancer (%)	6.22	6.67	NS
Cardiac failure (%)	3.41	4.00	NS
Renal insufficiency (%)	9.66	10.67	NS
Liver disease (%)	13.07	9.33	NS
WBC (×109/L)	14.20 (8.53-21.48)	17.10 (12.50-24.90)	.0244
Platelet count (×109/L)	190.00 (119.25-300.50)	167.00 (110.00-223.00)	.0451
PT (s)	15.60 (13.33-17.83)	18.50 (15.80-21.10)	<.0001
PT-INR (ratio)	1.30 (1.10-1.50)	1.40 (1.20-1.60)	.0034
Fibrinogen (g/L)	5.60 (3.83-7.28)	3.69 (2.54-5.69)	<.0001
D-dimer (mg/L)	3.50 (2.20-7.80)	6.60 (5.20-8.10)	<.0001
EV-TF activity (pg/mL)	0.69 (0.19-1.52)	0.34 (0.12-0.82)	.0002
Lactate (mM)	2.20 (1.50-3.50)	2.25 (1.50-4.65)	NS
SOFA score	7.00 (5.50-9.00)	9.00 (7.00-11.00)	<.0001
DIC score	3.00 (3.00-4.00)	4.00 (3.00-5.00)	<.0001
SIC score	3.00 (2.00-4.00)	4.00 (3.00-5.00)	NS
28-d mortality (%)	19.14	26.67	NS

Results are presented as median (interquartile range) for continuous variables and frequencies for categorical variables. The Mann–Whitney test was used to compare values between groups for continuous variables. For categorical variables, Fisher test was used. NS denotes not significant (P > .05).

DIC, disseminated intravascular coagulation; EV, extracellular vesicle; INR, international normalized ratio; PT, prothrombin time; SIC, sepsis-induced coagulopathy; SOFA, sequential organ failure assessment; TF, tissue factor; WBC, white blood cell.

3.2. Contribution of the different items to the DIC and SIC scores in the 2 cohorts

As expected, we observed that more patients had a positive SIC score compared with the DIC score in both cohorts of patients with sepsis (Hamilton, 58% vs 48%; Marseille, 50% vs 25%). This is consistent with the SIC score detecting an early-phase DIC [10]. We analyzed the levels of the different items used to calculate the DIC and SIC scores (Table 2). Patients with sepsis with DIC had significantly lower platelet counts, higher PT, and higher D-dimer than patients with sepsis but without DIC in both cohorts. However, fibrinogen was only significantly lower in patients with DIC than in patients without DIC in the Marseille cohort. The mortality was nonsignificantly higher in the patients with DIC than in patients without DIC in both cohorts. For both cohorts, patients with sepsis and SIC had significantly lower platelet counts, higher PT-INR, and higher SOFA score than patients with sepsis but without SIC. The mortality was nonsignificantly higher in the patients with SIC than in patients without SIC in both cohorts.

Table 2.

Levels of different markers for the 2 cohorts of patients with sepsis.

Markers	Marseille (n = 188)			Hamilton (n = 75)
	Non-DIC (n = 143)	DIC (n = 45)	P	Non-DIC (n = 39)	DIC (n = 36)	P
Platelets (109/L)	208.00 (141.00-308.00)	120.00 (47.00-204.00)	<.0001	181.00 (132.00-250.00)	129.00 (82.25-188.50)	.0061
PT (s)	14.80 (13.20-16.00)	19.00 (17.55-23.30)	<.0001	15.80 (14.50-17.20)	21.75 (19.80-32.68)	<.0001
Fibrinogen (g/L)	5.98 (4.30-7.60)	3.70 (2.73-2.85)	<.0001	3.90 (2.43-5.70)	3.56 (2.56-5.60)	NS
D-dimer (mg/L)	3.10 (2.00-4.70)	8.00 (4.90-16.10)	<.0001	5.40 (4.60-6.50)	8.00 (7.18-9.10)	<.0001
Mortality (%)	16.78	26.67	NS	17.95	36.11	NS

	Marseille (n = 147)			Hamilton (n = 72)
	Non-SIC (n = 77)	SIC (n = 70)	P	Non-SIC (n = 30)	SIC (n = 42)	P
Platelets (109/L)	234.00 (185.50-319.75)	110.00 (56.00-182.50)	<.0001	210.00 (169.25-296.00)	120.50 (83.75-180.25)	<.0001
PT-INR (ratio)	1.20 (1.20-1.30)	1.50 (1.32-1.70)	<.0001	1.20 (1.10-1.30)	1.60 (1.48-2.33)	<.0001
SOFA score	7.00 (5.75-8.00)	9.00 (6.00-10.00)	.0020	8.00 (6.00-10.00)	10.00 (8.00-11.00)	.0086
Mortality (%)	12.99	24.29	NS	13.33	33.33	NS

Results are presented as median (interquartile range). The Mann–Whitney test was used to compare values between groups. NS denotes not significant (P > .05).

DIC, disseminated intravascular coagulation; PT, prothrombin time; INR, international normalized ratio; SIC, sepsis-induced coagulopathy; SOFA, sequential organ failure assessment.

3.3. Predictive power of individual items and combinations for the DIC and SIC scores in the 2 cohorts of patients with sepsis

The predictive power of the different items used for the DIC and SIC scores were analyzed individually and in combination using receiver-operating characteristic curve analysis (Table 3). PT and D-dimer were good to excellent discriminators for DIC in both cohorts. However, platelet counts and fibrinogen were poor to fair discriminators for DIC in both cohorts. As expected, the combination of all items showed a good to excellent predictive power for DIC in both cohorts.

Table 3.

ROC curves of the predictive power of individual items and combinations for the DIC and SIC scores in the 2 cohorts of patients with sepsis.

Score	Item	Marseille			Hamilton
		AUC	95% CI	P	AUC	95% CI	P
DIC	Platelet	0.739	0.647-0.831	<.0001	0.653	0.526-0.780	.0259
	PT	0.848	0.767-0.928	<.0001	0.978	0.952-1.00	<.0001
	Fibrinogen	0.726	0.640-0.812	<.0001	0.516	0.381-0.652	.8128
	D-dimer	0.807	0.737-0.876	<.0001	0.832	0.729-0.934	<.0001
	Platelet + PT + fibrinogen +D-dimer	0.862	0.798-0.926	<.0001	0.996	0.988-1.00	<.0001
SIC	Platelet	0.820	0.750-0.890	<.0001	0.808	0.710-0.907	<.0001
	PT	0.833	0.759-0.906	<.0001	0.947	0.900-0.993	<.0001
	SOFA	0.691	0.606-0.775	<.0001	0.680	0.555-0.805	.0097
	Platelet + PT + SOFA	0.920	0.877-0.962	<.0001	0.984	0.964-1.00	<.0001

Results are presented as area under the curve (AUC) and 95% confidence interval (CI). The Mann–Whitney test was used to compare values between groups. NS denotes not significant (P > .05). AUC values of 0.5 to 0.7 (red) represent poor discrimination, >0.7 to 0.8 (yellow) represent fair discrimination, >0.8 to 0.9 (blue) represent a good discrimination, and >0.9 (green) represent excellent discrimination.

AUC, area under the curve; DIC, disseminated intravascular coagulation; PT, prothrombin time; SIC, sepsis-induced coagulopathy; SOFA, sequential organ failure assessment.

PT and platelet count were good to excellent discriminator for both cohorts. On the contrary, the SOFA score was a poor discriminator for SIC in both cohorts. The combination of items showed an excellent predictive power for SIC in both cohorts.

3.4. Levels of EV-TF activity in patients with sepsis compared with healthy subjects

We observed significantly higher levels of EV-TF activity in patients with sepsis than in healthy subjects (Figure 1A).

Figure 1.

Levels of extracellular vesicle (EV) tissue factor (TF) activity in healthy individuals and patients with sepsis. (A) Levels of EV-TF activity in healthy individuals (controls) and patients with sepsis (Hamilton cohort). (B) Levels of EV-TF activity in patients with sepsis without disseminated intravascular coagulation (DIC; non-DIC) and with pre-DIC (Hamilton cohort). Data are shown as median ± interquartile range and individual values. The Mann–Whitney test was used to compare values between groups. ∗P < .05.

3.5. Levels of EV-TF activity in patients who developed DIC during the study

Identification of patients with sepsis who develop DIC could increase the effectiveness of targeted therapies. In a recent study, we found that patients with sepsis who subsequently developed DIC during the study (defined as pre-DIC patients) had significantly lower levels of the anticoagulants protein C and antithrombin than patients with DIC who never developed DIC [34]. Therefore, we measured levels of EV-TF activity in patients with sepsis with pre-DIC and without DIC to determine whether it is a predictor of DIC. We observed higher levels of EV-TF activity in pre-DIC patients than in non-DIC patients but the difference was not significant (Figure 1B). Only 6 pre-DIC patients were identified in the Marseille cohort because samples were taken only for 2 days.

3.6. Levels of EV-TF activity in patients with sepsis with or without DIC or SIC

We observed a stepwise increase in EV-TF activity with increasing DIC score for both cohorts (Figure 2A, B). It is notable that patients in the Hamilton cohort had higher DIC scores than patients in the Marseille cohort. In addition, patients with sepsis and DIC had significantly higher EV-TF activity than patients without DIC in both cohorts (Figure 2C, D). EV-TF activity was a poor discriminator for DIC in the Marseille cohort (AUC, 0.609 [0.510-0.707]; P = .0281) and a fair discriminator for DIC in the Hamilton cohort (AUC, 0.730 [0.615-0.845]; P = .0008). We also compared EV-TF activity in patients with sepsis with or without SIC. Similar to the DIC results, we observed a stepwise increase in EV-TF activity with increasing SIC score for both cohorts (Figure 3A, B). Patients with sepsis and SIC had higher EV-TF activity than patients without SIC in both the Marseille and the Hamilton cohorts (Figure 3C, D). Spearman’s correlation analysis demonstrated that EV-TF activity positively correlated with the SIC score (r = 0.212; P = .011) and the SOFA score (r = 0.274; P = .001). In the Hamilton cohort, EV-TF activity positively correlated with PT (r = 0.393; P = .0005), D-dimer (r = 0.314; P = .006), the DIC score (r = 0.370; P = .004), the SIC score (r = 0.307; P = .009), and the SOFA score (r = 0.307; P = .009) and negatively correlated with the level of protein C (r = −0.305; P = .008). EV-TF activity was a poor discriminator for SIC in the Marseille (AUC, 0.648 [0.557-0.739]; P = .002) and Hamilton (AUC, 0.595 [0.488-0.702]; P = .085) cohorts. These results show that EV-TF activity is associated with DIC and SIC in patients with sepsis in 2 independent cohorts.

Figure 2.

Comparison of the levels of extracellular vesicle (EV) tissue factor (TF) activity in patients with sepsis with or without disseminated intravascular coagulation (DIC). (A) Levels of EV-TF activity in patients with sepsis with different DIC scores (Marseille cohort). (B) Levels of EV-TF activity in patients with sepsis with different DIC scores (Hamilton cohort). (C) Levels of EV-TF activity in patients with sepsis but without DIC (non-DIC) and with DIC (Marseille cohort). (D) Levels of EV-TF activity in patients with sepsis but without DIC (non-DIC) and with DIC (Hamilton cohort). Data are shown as median ± interquartile range and individual values. The Mann–Whitney test was used to compare values between groups. ∗P < .05; ∗∗P < .01.

Figure 3.

Comparison of the levels of extracellular vesicle (EV) tissue factor (TF) activity in patients with sepsis with or without sepsis-induced coagulopathy (SIC). (A) Levels of EV-TF activity in patients with sepsis with different SIC scores (Marseille cohort). (B) Levels of EV-TF activity in patients with sepsis with different SIC scores (Hamilton cohort). (C) Levels of EV-TF activity in patients with sepsis without SIC (non-SIC) and with SIC (Marseille cohort). (D) Levels of EV-TF activity in patients with sepsis without SIC (non-SIC) and with SIC (Hamilton cohort). Data are shown as median ± interquartile range and individual values. The Mann–Whitney test was used to compare values between groups. ∗P < .05; ∗∗P < .01.

3.7. Longitudinal analysis of EV-TF activity in patients with sepsis

EV-TF activity was analyzed in the Marseille cohort between days 0 and 2. Patients with sepsis but without DIC showed similar levels of EV-TF activity over time (Figure 4A). On the contrary, patients with sepsis and DIC showed a decrease in EV-TF activity on day 2 compared with that on day 0 (Figure 4B), returning to levels observed for patients without DIC. Similar results were observed for patients with or without SIC (Figure 4C, D).

Figure 4.

Time course of levels of extracellular vesicle (EV) tissue factor (TF) activity in plasma from patients with sepsis with or without disseminated intravascular coagulation (DIC) or sepsis-induced coagulopathy (SIC; Marseille cohort). Levels of EV-TF activity in (A) patients with sepsis but without DIC (non-DIC; n = 106), (B) with DIC (n = 36), (C) without SIC (n = 58), and (D) with SIC (n = 57). EV-TF activity was measured at days 0, 1, and 2. Wilcoxon matched-pairs rank test was used for comparison of values from the same patients between days 0 and 2. ∗P < .01.

4. Discussion

In this study, we used samples from 2 cohorts of patients with sepsis. Based on levels of the different items in the ISTH DIC score and the DIC and SOFA scores, patients in the Hamilton cohort had more severe sepsis than those in the Marseille cohort. Both cohorts had higher numbers of patients with SIC than those of DIC, which is consistent with the SIC score detecting an early-phase DIC [10]. We evaluated the ability of individual items of the DIC and SIC scores to discriminate between patients with sepsis with DIC or SIC compared with patients without DIC or SIC. For the DIC score, PT and D-dimer were good to excellent discriminators, whereas platelet count and fibrinogen were poor discriminators between patients with or without DIC. A recent study found no difference in fibrinogen levels between patients with sepsis and SIC and patients with sepsis but without SIC [37]. For the SIC score, PT and platelets were good to excellent discriminators, whereas the SOFA score was a poor discriminator between patients with and those without SIC.

In our study, we showed that the level of EV-TF activity in patients with sepsis was higher than that in healthy controls, which is consistent with previous studies [28,30]. EV-TF activity was associated with DIC and SIC in 2 independent cohorts of patients with sepsis. A major source of TF-positive EVs is likely to be monocytes because these cells are activated during sepsis and activated monocytes have been shown to express TF and release TF-positive EVs [17,18,20,38]. Three other studies have analyzed levels of EV-TF activity in patients with sepsis with DIC in comparison with that of patients with sepsis but without DIC [31,32,39] (Table 4). Importantly, 2 of these studies used the ISTH 2001 DIC score [2], and 1 study did not state which ISTH DIC score was used. Tobiasch et al. [31] observed a nonsignificant increase in EV-TF activity in patients with sepsis and DIC compared with that in patients with sepsis but without DIC. EV-TF activity was strongly correlated to the EV concentration. Meng et al. [32] found a significantly higher levels of EV-TF activity in patients with sepsis and DIC than in patients with sepsis but without DIC. EV-TF activity correlated with the inflammatory markers C-reactive protein, interleukin-1β and tumor necrosis factor α. Van Dreden et al. [39] also showed significantly higher levels of EV-TF activity in patients with sepsis and DIC than in patients with sepsis but without DIC. EV-TF activity correlated with D-dimer, fibrin degradation products, soluble thrombomodulin, the DIC score, and the SOFA score. Finally, a recent study observed higher levels of TF antigen in serum from patients with sepsis and SIC than in serum from patients with sepsis but without SIC [37].

Table 4.

Comparison of different studies on levels of EV-TF activity in patients with sepsis with or without DIC.

Non-DIC (n); EV TF, median (IQR)	DIC (n); EV TF, median (IQR)	P	Method	Reference
52; 0.28 (0.06-1.42)	35; 0.69 (0.25-1.60)	NS	ZYMUPHEN MP-TF kit	Tobiasch et al. [31], 2024
31; ∼0.1	31; ∼0.5	<.01	Chapel Hill assay	Meng et al. [32], 2024
49	65	<.01	CY-Quant MV-TF Activity kit	Van Dreden et al. [39], 2025

DIC, disseminated intravascular coagulation; EV, extracellular vesicle; NS, not significant; TF, tissue factor.

TF activity of EVs isolated from either PPP or PFP by centrifugation or captured onto plates containing an anti-TF antibody can be measured using several assays [40]. These assays have different sensitivities. In our study, we used PPP in the Hamilton cohort and PFP in the Marseille cohort. We and others have observed higher levels of TF activity in EVs isolated from PPP vs PFP [41]. EV-TF activity in the Hamilton samples was measured using the Chapel Hill assay, whereas EV-TF activity in the Marseille cohort was measured using a modified Chapel Hill assay that has increased sensitivity compared with the original Chapel Hill assay [35]. This may explain why we observed higher levels of EV-TF activity in the Marseille samples than in the Hamilton samples. However, it is important to note that we observed an association between EV-TF activity and DIC in patients with sepsis in both cohorts despite the differences in preanalytical variables and assays. Tobiasch et al. [31] used the Zymuphen MP-TF Activity Kit. We have shown that this kit has lower sensitivity than the Chapel Hill assay [42]. Meng et al. [32] used the Chapel Hill assay. Van Dreden et al. [39] used the CY-Quant MV-TF Activity assay. We have shown that this kit has lower sensitivity than the Chapel Hill assay [43].

We analyzed the level of EV-TF activity over time in both patients with sepsis with or without DIC or SIC. We found that EV-TF activity decreased in a time-dependent manner between day 0 and day 2 in the DIC and SIC groups but not in the non-DIC or non-SIC groups. Similar to our results, Meng et al. [30] observed a time-dependent decrease in the level of EV-TF activity in patients with sepsis between day 1 and day 5. In contrast, Tobiasch et al. [31] reported a time-dependent increase in EV-TF activity in patients with sepsis and DIC. This is somewhat surprising because one would expect the level of EV-TF activity to decrease as DIC resolves.

The association between EV-TF activity and DIC has been reported in other diseases, such as cancer. A study of 5 patients with cancer with DIC, including 1 with acute promyelocytic leukemia (APL), found high levels of plasma EV-TF activity [44]. DIC resolved in 2 of the patients, and this was associated with a reduction in EV-TF activity. Another study reported elevated levels of EV-TF activity in patients with prostate cancer with DIC compared with that in patients without DIC [45]. Increased levels of EV-TF activity have been found in patients with APL compared with those in healthy controls [46]. Indeed, the APL cell line NB4 expresses high levels of TF and releases TF-positive EVs [47,48]. In patients with acute myeloid leukemia (AML), TF procoagulant activity of peripheral blood mononuclear cells, which includes blast cells, was associated with decompensated DIC (fibrinogen levels ≤ 1 g/mL). TF procoagulant activity of peripheral blood mononuclear cells correlated with circulating blast cells and D-dimer and serum lactate dehydrogenase levels [49]. In addition, EV-TF activity was more frequently detected in patients with leukemia with DIC than in patients without DIC. A small study of 7 patients with acute leukemia reported an association between EV-TF activity and DIC in 2 patients but no association with venous thromboembolism in 3 patients [50]. Interestingly, the level of EV-TF activity decreased to baseline levels after resolution of the DIC. We observed higher levels of EV-TF activity in patients with APL or AML with DIC but not patients with acute lymphoblastic leukemia than in patients without DIC [51]. EV-TF activity was also associated with bleeding in patients with acute leukemia [51]. In APL and AML, blast cells are likely the major source of TF-positive EVs.

The strengths of our study are the analysis of EV-TF activity in 2 independent cohorts of patients with sepsis and the use of the ISTH 2025 DIC score. Importantly, we found that EV-TF activity was associated with the 2025 ISTH DIC score and the SIC score in both cohorts. Our longitudinal analysis showed that the level of EV-TF activity decreased over time in patients with sepsis and DIC or SIC. Limitations of the study include that the inclusion and exclusion criteria were not standardized for the 2 cohorts, there were differences in preanalytical variables (PPP vs PFP) and the assays used to measure EV-TF activity, and some of the group sizes were small. In summary, we found that EV-TF activity is associated with DIC and SIC in patients with sepsis.