Combination of antithrombin and soluble thrombomodulin for early prediction of sepsis-Induced disseminated intravascular coagulation

Abstract

Background

See More

To identify early diagnostic biomarkers for sepsis-induced disseminated intravascular coagulation, we investigated the relationship between the novel coagulation biomarkers and antithrombin with the development of Disseminated intravascular coagulation post-admission, as well as the prognosis of patients with sepsis.

Methods

We retrospectively collected data from septic patients admitted to the Emergency Intensive Care Unit (EICU) of a teaching hospital between October 2021 and September 2023. Multivariate logistic regression analysis was performed to identify risk factors, and receiver operating characteristic (ROC) curve analysis was used to assess the performance of the predictive model. In addition, non-parametric bootstrap analysis with 1,000 replications was conducted to evaluate the internal stability and empirical power of the predictive models, particularly given the limited sample size.

Results

Among 91 septic patients, 15 were diagnosed with DIC. Soluble thrombomodulin (OR: 1.28, 95% CI: 1.033–1.586, P = 0.024) and antithrombin activity (OR: 0.887, 95% CI: 0.792–0.994, P = 0.039) were identified as independent risk factors for the development of DIC in septic patients. The area under the curve (AUC) for soluble thrombomodulin and antithrombin was 0.788 and 0.757, respectively. When combined, the AUC increased to 0.858. Prothrombin Time (HR: 1.058, 95% CI: 1.016–1.102, P = 0.007) and APACHE II score (HR: 1.071, 95% CI: 1.005–1.141, P = 0.035) were identified as independent risk factors for 28-day mortality in septic patients. When combined, the AUC increased to 0.834. Bootstrap validation demonstrated strong discriminatory performance of both models, with a mean bootstrap AUC of 0.865 (empirical power = 0.994) for the DIC prediction model, and 0.836 (empirical power = 0.996) for the 28-day mortality model, further supporting the robustness and reliability of the findings despite the small sample size.

Conclusion

Elevated soluble thrombomodulin and decreased antithrombin may be associated with the early onset of disseminated intravascular coagulation in sepsis, but showed limited predictive value for 28-day mortality.

Introduction

Disseminated intravascular coagulation (DIC) is an acquired coagulopathy syndrome characterized by sustained, systemic coagulation activation and microthrombosis [1]. The hallmark of DIC is the widespread generation of thrombin and extensive activation of the coagulation system, which ultimately results in organ dysfunction [2]. Sepsis is a common risk factor for the development of DIC, with an incidence rate ranging from 30 to 50% [3]. Furthermore, DIC serves as an independent risk factor for poor prognosis in septic patients, significantly increasing mortality rates [3,4,5]. Consequently, the early diagnosis and prompt treatment of DIC in septic patients are crucial for reducing mortality.

Endothelial damage and immunothrombosis lead to an interplay between coagulation and inflammation that occurs early in sepsis and manifests as sepsis-induced coagulopathy (SIC) [2]. The clinical manifestations of SIC are diverse, ranging from mild coagulation marker abnormalities to severe coagulation disorders, such as DIC [6]. In contrast to the enhanced-fibrinolytic type DIC observed in other conditions, lipopolysaccharide (LPS) and inflammatory cytokines exert their effects on vascular endothelial cells, leading to an enhanced production of plasminogen activator inhibitor-1 (PAI-1) and α₂-plasmin inhibitor (α₂PI) [7]. This interaction culminates in a pronounced state of fibrinolytic suppression [1, 8]. The risk of organ failure is heightened due to microcirculatory collapse, although symptoms of bleeding are relatively uncommon. At this stage, fibrin degradation products, such as D-dimer, may only exhibit a mild elevation. Concurrently, the inflammatory response typically prevents a decrease in fibrinogen levels [1]. Diagnostic scoring systems for overt DIC, such as the International Society on Thrombosis and Haemostasis (ISTH) and the Japanese Association for Acute Medicine (JAAM), rely on conventional coagulation markers, including fibrinogen, fibrin degradation products, and platelet counts. These systems may take longer to detect overt DIC [9, 10]. Previous studies [11] have demonstrated that mortality rates in patients with non-overt DIC are nearly identical to those in patients with overt DIC, highlighting the critical importance of early identification of DIC to reduce mortality in septic patients.

Endothelial injury, abnormal coagulation activation, and dysfunction of fibrinolysis are critical factors in the development of sepsis-induced DIC [1]. Consequently, novel coagulation biomarkers that reflect these abnormalities—such as soluble thrombomodulin (sTM), tissue plasminogen activator-inhibitor complex (t-PAIC), thrombin-antithrombin complex (TAT), and α2-plasmin inhibitor-plasmin complex (PIC)—may prove to be valuable in predicting the early stages of sepsis-induced DIC [12, 13]. Antithrombin (AT) plays a crucial role in maintaining the equilibrium between coagulation and anticoagulation by irreversibly inhibiting thrombin and other coagulation factors [13]. Deficiency of AT may arise from various factors, including coagulation activation, prolonged anticoagulation therapy, or impaired synthesis [14].In addition, microvascular leakage of plasma proteins is a common mechanism contributing to acquired AT deficiency [15]. Acquired AT deficiency is particularly prevalent in patients experiencing excessive coagulation activation. Previous studies [16] have demonstrated that the combination of AT and Prothrombin Time (PT) can facilitate the early diagnosis of sepsis-induced DIC. Furthermore, plasma AT levels have been shown to predict the onset of DIC in patients undergoing ECMO treatment [17]. Nonetheless, there is a paucity of reports addressing the predictive value of novel coagulation biomarkers in conjunction with AT for sepsis-induced DIC.

This study aimed to evaluate the predictive value of novel coagulation biomarkers (sTM, TAT, t-PAIC, PIC) and antithrombin (AT) in the early diagnosis of DIC and the assessment of 28-day mortality in patients with sepsis. We hypothesized that abnormalities in these coagulation markers would manifest in the initial stages of sepsis and correlate with the subsequent development of overt DIC following hospital admission. Accordingly, we investigated the relationship between plasma biomarkers measured at admission and the onset of overt DIC within the subsequent seven days, as well as their association with 28-day mortality.

Methods

Study design and setting

A retrospective, observational, single-center study was conducted on septic patients admitted to the Intensive Care Unit (ICU) of the Emergency Department at Tongji Hospital, affiliated with Tongji University in Shanghai, China, from October 2021 to September 2023. The inclusion criteria specified septic patients aged 18 years or older. Sepsis was diagnosed based on the Sepsis-3 criteria [18], defined by a confirmed infection and a Sequential Organ Failure Assessment (SOFA) score ≥ 2. Exclusion criteria were: (I) age < 18 years, (II) massive hemorrhage or hemorrhagic shock due to trauma, (III) pregnancy, (IV) heparin-induced thrombocytopenia (HIT), (V) thrombotic thrombocytopenic purpura (TTP), (VI) decompensated liver cirrhosis (Child–Pugh B or C), (VII) prior anticoagulant therapy (warfarin, dabigatran, rivaroxaban, heparin/low molecular weight heparin), (VIII) a history of malignancy.

This study was approved by the Ethical Committee of Tongji Hospital, which granted a waiver for informed consent from patients and their legal representatives (Approval No: B2021-090R). All sepsis treatments followed the Surviving Sepsis Campaign: International Guidelines for the Management of Sepsis and Septic Shock: 2016 [19].

Data collection and patient grouping

The following demographic and clinical information was extracted from electronic medical records: patient sex, age, comorbidities, infection site, laboratory tests within 24 h after admission, Sequential Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation (APACHE II) score and ISTH score based on International Society on Thrombosis and Haemostasis (ISTH) criteria. Laboratory tests include the following 22 specific biomarkers: Hematologic Markers: White blood cell count (WBC)(Reference ranges:3.5–9.5 × 10⁹/L), platelet count(PLT)(125–350 × 10⁹/L). Routine Coagulation Markers: Prothrombin time (PT)(9.8–14.5 s), fibrinogen (Fib)(2–4.5 g/L), fibrin degradation products (FDP)(< 5 μg/L), D-dimer(< 0.55 mg/L), and antithrombin(AT)(75–125%). Inflammatory Markers: C-reactive protein(CRP)(< 10 mg/L), procalcitonin (PCT)(< 0.05 ng/mL), interleukins-6(IL-6)(< 5.4 pg/mL), interleukins-8(IL-8)(< 20.6 pg/mL), and interleukins-10(IL-10)(< 12.9 pg/mL). Novel Coagulation Biomarkers: Soluble thrombomodulin (sTM)(3.82–13.35TU/mL), thrombin-antithrombin complex (TAT)(0–4.08 ng/mL), tissue plasminogen activator-inhibitor complex (t-PAIC)(0–17.13 ng/mL), and plasmin-inhibitor complex (PIC)(0–0.85 μg/mL). Organ Function Markers: Total bilirubin (TB)(5.1–20.5 μmol/L) and creatinine (Cr)(49-90 μmol/L).

Patients were categorized into the DIC group and non-DIC group based on the development of DIC within 7 days of admission. DIC was diagnosed if the ISTH score was ≥ 5 on any day during this period [8]. Additionally, patients were further classified into the survival group and non-survival group based on their 28-day survival outcomes.

Statistics

Statistical analyses were conducted using SPSS version 27.0 and GraphPad Prism 10.0 software. All data were subjected to normality and homogeneity of variance tests. Continuous variables with a normal distribution were expressed as mean ± standard deviation (SD), while non-normally distributed variables were reported as median (interquartile range: 25th and 75th percentiles). Categorical data were presented as numbers (percentages). For non-normally distributed continuous variables, the Mann–Whitney U test was used to compare differences between groups. The Pearson chi-square test was employed for the comparison of categorical variables. Logistic regression analysis with forward stepwise selection was applied to calculate odds ratios (ORs) and 95% confidence intervals (CIs) to identify biomarkers predictive of DIC, based on variables found to be significant in univariate analysis. For variables showing significant differences between survivor and non-survivor groups, Cox proportional hazards regression analysis, using forward stepwise selection, was conducted to estimate hazard ratios (HRs) and 95% CIs to identify prognostic biomarkers. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated. The optimal cut-off values were determined using the Youden index to maximize the sum of sensitivity and specificity. Kaplan–Meier survival curves were generated to depict 28-day mortality trends, and the log-rank test was employed to compare survival curves between groups. A two-tailed P-value of less than 0.05 was considered to indicate statistical significance.

To address concerns related to limited sample size and to evaluate the robustness and empirical power of our predictive models, we performed a non-parametric bootstrap resampling procedure using R software (version 4.4). Specifically, we resampled the original dataset 1,000 times with replacement. For each bootstrap replicate, a binary logistic regression model was refitted, and the area under the receiver operating characteristic curve (AUC) was computed to assess discriminative performance. Empirical power was estimated as the proportion of bootstrap replicates in which the AUC exceeded a pre-specified clinically meaningful threshold (θ = 0.70). This threshold was selected to reflect the minimum acceptable discriminative ability, with AUC values ≥ 0.80 generally considered to indicate good to excellent model performance [20, 22, 32].

Results

Demographic and baseline characteristics of the study population

This study began in October 2021 and concluded in September 2023. Over the two-year period, a total of 213 patients were initially enrolled (Fig. 1). Based on the exclusion criteria, 112 patients were excluded from the study. Ultimately, 91 patients were included in the final analysis. Of these, 15 patients were diagnosed with DIC, and 31 patients died within 28 days of admission.

Fig. 1

Study flow diagram of the patient selection process

The demographic and baseline characteristics of the patients, based on DIC groupings, are shown in Table 1. Among the 91 septic patients, 15 (16.48%) developed overt DIC within 7 days of ICU admission. Compared to the non-DIC group, patients in the DIC group had significantly more severe conditions, with higher APACHE II score (24, IQR 16–29 vs. 15, IQR 8–19.75, P = 0.005), SOFA score (8, IQR 4–10 vs. 4, IQR 3–6, P = 0.007), and a significantly higher 28-day mortality rate (80% vs. 24.68%, P < 0.001). The length of hospital stay was significantly shorter in the DIC group (5.5, (IQR 3.75–9.5) vs. 11, (IQR 7–14), P = 0.005) compared to the non-DIC group. Table 1 also illustrates the clinical and laboratory characteristics of the population based on DIC groupings. Age, gender, and comorbidities were similar between the DIC and non-DIC groups. There were no significant differences between the groups in WBC count, IL-6, IL-8, CRP, or PCT levels. However, the DIC group exhibited significantly lower platelet counts (74 × 10³/μL (IQR 40–113 × 10³/μL) vs. 130 × 10³/μL (IQR 87.25–200.5 × 10³/μL), P = 0.002), lower fibrinogen levels (3.09 mg/dL (IQR 1.64–5.65 mg/dL) vs. 4.94 mg/dL (IQR 3.91–6.48 mg/dL), P = 0.004), and reduced AT activity (44.3% (IQR 37.2–55.7%) vs. 61.35% (IQR 52.25–70.18%), P = 0.002). In contrast, levels of IL-10, PT, D-dimer, FDP, sTM, TAT, TB and Cr were significantly higher in the DIC group compared to the non-DIC group (P < 0.05). These findings highlight the potential for more severe coagulopathy and organ dysfunction in septic DIC patients. Specific biomarkers associated with these conditions, particularly sTM, PLT, Fib, TB, Cr and AT, may serve as valuable predictors for the early diagnosis and prognosis of sepsis-induced DIC.

Table 1 Patient demographics and clinical characteristics (n = 91) with sepsis

The demographic and baseline characteristics of the patients, categorized by 28-day mortality, are presented in Table 2. There were no significant differences in gender, age and comorbidities between non-survivors (n = 31) and survivors (n = 60). However, non-survivors had a significantly higher incidence of DIC (38.71% vs. 5.00%, P < 0.001) and septic shock (74.19% vs. 18.33%, P < 0.001). Additionally, non-survivors exhibited markedly higher APACHE II score (21 (IQR 16–27) vs. 13 (IQR 7.25–19), P < 0.001) and SOFA score (6 (IQR 4–10) vs. 3.5 (IQR 3–6), P < 0.001) compared to survivors. Non-survivors had significantly elevated levels of IL-10, PT, D-dimer, FDP, TAT, and creatinine (P < 0.05) compared to survivors. These results underscore the greater severity of illness in non-survivors, with pronounced coagulopathy and organ dysfunction evident from the early stages of ICU admission.

Table 2 Patient demographics and clinical characteristics (n = 91) based on outcomes

Early prediction role of sTM and AT for sepsis-induced DIC

Patients in the DIC group exhibited a higher severity of illness. Univariate analysis revealed that levels of sTM, TAT, PT, D-dimer, Cr, TB, as well as APACHE II and SOFA scores were significantly higher in the DIC group compared to the non-DIC group. In contrast, AT, PLT, and Fib levels were significantly lower in the DIC group. Multivariate logistic regression analysis revealed that sTM (OR = 1.28, 95% CI: 1.033–1.586, P_adj = 0.024) and AT (OR = 0.887, 95% CI: 0.792–0.994, P_adj = 0.039) were independent predictors for the occurrence of DIC in septic patients (Table 3). ROC curve analysis was performed to assess the predictive performance of sTM and AT for DIC, yielding area under the curve (AUC) values of 0.788 (95% CI: 0.664–0.912) and 0.757 (95% CI: 0.597–0.918), respectively, with cut-off values of 24.68 TU/mL for sTM and 48.05% for AT. When sTM and AT were combined, the AUC improved to 0.858 (95% CI: 0.748–0.968), with corresponding sensitivity and specificity values of 0.933 and 0.711, respectively. (Table 4 and Fig. 2A).

Table 3 Multivariate logistic regression analyses for the DIC

Table 4 Area under the ROC curves of biomarkers at baseline for prediction of DIC

Fig. 2

(A) ROC curves for the prediction of sepsis-induced DIC; (B) comparison of Kaplan–Meier survival curves between patients with high and low APACHE II score; (C) ROC curves for the prediction of 28-day mortality comparison of Kaplan–Meier survival curves between patients with high and low t-PAIC levels; (D) comparison of Kaplan–Meier survival curves between patients with high and low PT level

Predictive value of APACHE II score and PT for 28-day mortality in septic patients

In the univariate analysis between the survivor and non-survivor groups, plasma levels of sTM, IL-10, PT, D-dimer, FDP, Cr, as well as APACHE II and SOFA score were significantly higher in the non-survivor group compared to the survivor group. Further multivariate Cox regression analysis identified IL-10 (HR = 1.002, 95% CI: 1.000–1.003, P_adj = 0.015), APACHE II score (HR = 1.071, 95% CI: 1.005–1.141, P_adj = 0.035), and PT (HR = 1.058, 95% CI: 1.016–1.102, P_adj = 0.007) as independent predictors of 28-day mortality in septic patients (Table 5).

Table 5 Multivariate COX regression for 28-day mortality

Subsequently, ROC curve analysis was performed to further evaluate the predictive performance of IL-10, PT, and APACHE II score for 28-day mortality, resulting in AUC values of 0.613 (95% CI: 0.491–0.735), 0.782 (95% CI: 0.676–0.888), and 0.742 (95% CI: 0.632–0.852), respectively. Combining APACHE II score and PT further increased the AUC to 0.834 (95% CI: 0.734–0.925, P_adj < 0.001). The cut-off values for APACHE II score and PT were determined to be 17.5 and 15.25 s, respectively (Table 6 and Fig. 2C). Kaplan–Meier curves, plotted based on these cut-off values, demonstrated that septic patients with PT and APACHE II score above the respective thresholds had significantly higher 28-day mortality rates. (As shown in Fig. 2B and D.)

Table 6 Area under the ROC curves of biomarkers for prediction of survival

Statistical power assessment using bootstrap resampling

For the model incorporating AT activity and sTM, the distribution of bootstrap AUCs yielded a mean AUC of 0.865. Under the null hypothesis of AUC = 0.70, the empirical power was 0.994, indicating strong evidence for reliable discriminative performance. These findings align with the underlying biological rationale: AT serves as a marker of endogenous anticoagulant capacity, while sTM reflects endothelial injury and activation in sepsis-associated coagulopathy. (Fig. 3A).

Fig. 3

(A) Distribution of bootstrap AUCs (B = 1,000) for the model incorporating AT activity and sTM, (B) Distribution of bootstrap AUCs (B = 1,000) for the model incorporating APACHE II score and PT. Vertical dashed line indicates the null threshold AUC = 0.70

For the model using APACHE II score and PT as predictors, the mean bootstrap AUC was 0.836. The empirical power under the same null hypothesis (AUC = 0.70) was 0.996, suggesting a near-certain ability to detect at least clinically acceptable discrimination. (Fig. 3B).

Discussion

In this retrospective, observational study, we observed that the incidence of sepsis-related DIC was 17.64%, with an overall mortality rate of 32.33%. The incidence and mortality rates observed in our study were largely consistent with those reported for DIC diagnosed according to the ISTH criteria [22]. There were significant differences in the levels of sTM, TAT, and AT between patients with sepsis-related DIC and those without DIC within 24 h of admission. We concluded that elevated sTM levels and decreased AT are independent risk factors for the development of sepsis-related DIC. Furthermore, we did not observe any role of novel coagulation markers and AT in assessing short-term mortality.

A multicenter retrospective study involving 681 patients [12] demonstrated that four novel coagulation biomarkers, including sTM, TAT, t-PAIC and PIC, exhibited significant diagnostic and prognostic value across various underlying diseases that can lead to DIC. Notably, the combined use of these biomarkers outperformed individual biomarkers in terms of predictive efficacy. Research conducted by Zhang J et al. [6] identified elevated levels of sTM and t-PAIC as independent predictors of poor 60-day outcomes in septic patients, highlighting sTM as a sensitive biomarker for the early prediction of septic shock and sepsis-associated DIC. Similarly, Yin Q et al. [23] underscored the prognostic significance of sTM in sepsis, demonstrating its capability to enhance the predictive accuracy of the Mortality in Emergency Department Sepsis(MEDS) score for severe sepsis and 30-day mortality. These findings further corroborate sTM's role in risk stratification and early clinical assessment. In our study, we observed that, compared to non-DIC patients, the levels of sTM and TAT were significantly elevated in the DIC group, while no significant differences were noted in t-PAIC and PIC. Notably, sTM (cut-off value > 24.68) emerged as an independent risk factor for DIC development in septic patients, demonstrating an area under the curve (AUC) of 0.788, with a sensitivity of 0.667 and a specificity of 0.886. The combination of sTM with AT resulted in a significant enhancement of the AUC, sensitivity, and specificity. These findings underscore the critical role of endothelial injury and coagulation activation in septic patients, particularly concerning the early detection of sepsis-induced DIC.

In 2016, the Japanese Society on Thrombosis and Hemostasis (JSTH) DIC Diagnostic Standards Committee proposed a draft for provisional DIC diagnostic standards [1]. These new criteria introduced several commendable innovations to reduce misdiagnosis, including the selective application of diagnostic criteria based on underlying conditions and the integration of molecular biomarkers alongside antithrombin. Notably, this initiative marked the first inclusion of AT activity in DIC diagnostic criteria, recognizing its direct relevance to anticoagulant therapy decisions and its potential to enhance both the sensitivity of DIC diagnosis and prognostic evaluation in patients with infections. AT is a pivotal anticoagulant in plasma, functioning primarily by inhibiting thrombin, inactivating coagulation factors, and suppressing platelet aggregation, which collectively account for 70% to 80% of total plasma anticoagulant activity [24]. Previous studies have identified AT as an independent risk factor for the development and progression of DIC, with its predictive value significantly enhanced when assessed in conjunction with prothrombin time (PT) [16, 17, 25, 26]. Furthermore, Wada et al. [27] observed that among a cohort of 3,008 DIC patients, the 28-day survival rate was significantly lower in patients with severe AT deficiency compared to those without such deficiency. In our study, we found that AT activity was significantly reduced in patients with DIC compared to those non-DIC. Importantly, AT activity (cut-off value < 48.05) was identified as an independent risk factor for the development of DIC in septic patients, with an area under the curve (AUC) of 0.757, sensitivity of 0.855, and specificity of 0.667. The combination of sTM and AT activity demonstrated enhanced predictive value, yielding an AUC of 0.858, sensitivity of 0.933, and specificity of 0.711. Consequently, AT may serve as a valuable diagnostic marker for sepsis-induced DIC in clinical practice. The diagnostic accuracy of AT is significantly enhanced when used in conjunction with novel coagulation biomarkers such as sTM, providing valuable insights for clinical decision-making. For septic patients with elevated plasma sTM levels and decreased AT activity, early initiation of anticoagulation therapy should be considered, assuming there are no contraindications related to bleeding. Furthermore, it is essential to closely monitor coagulation function in these patients.

The APACHE II score is a widely utilized scoring system for evaluating patient prognosis in the intensive care unit(ICU) [28]. The PT serves as a marker indicative of organ failure, with prolonged PT values often correlating with poor prognosis [29, 30]. In this study, TAT levels were significantly elevated in the non-survivor group compared to the survivor group, whereas there were no significant differences were observed between the two groups regarding sTM, PIC, t-PAIC, or AT activity levels. PT (cut-off value > 15.25) and APACHE II (cut-off value > 17.5) were identified as independent risk factors for 28-day mortality in septic patients. The combination of these two factors demonstrated enhanced predictive performance for 28-day adverse outcomes, with an area under the curve (AUC) of 0.834, a sensitivity of 0.933, and a specificity of 0.9. Kaplan–Meier survival curves revealed that patients with PT and APACHE II score exceeding the established the cut-off values experienced significantly higher mortality rates compared to those with lower score. However, no correlation was observed between novel coagulation markers, and prognosis in septic patients. Similarly, a multicenter retrospective cohort study conducted by Yarimizu et al. [31] reported that AT activity serves as an independent prognostic predictor of in-hospital mortality. When AT activity is utilized to forecast mortality in disease groups with mortality rates ranging from 20 to 50%, it exhibits enhanced specificity and predictive value. However, in high-risk mortality cases, no significant difference in AT activity were observed between survivors and non-survivors. In our cohort, DIC patients exhibited a notably elevated mortality rate of 72.22%. This increased mortality may explain the limited predictive value of AT activity identified in our study.

Although the relatively small sample size in this study may raise concerns regarding statistical robustness and the generalizability of the findings, we employed a non-parametric bootstrap approach to assess the empirical power and stability of the predictive models. Bootstrap is a robust internal validation technique that estimates model performance variability through repeated resampling, which is particularly valuable in small datasets. The high empirical power obtained from 1,000 bootstrap replications suggests that both models are highly likely to detect true predictive signals rather than spurious associations [20, 22, 32].

The results demonstrated excellent discriminative ability: the logistic regression model incorporating sTM and AT activity yielded a mean AUC of 0.865, with an empirical power of 0.994 under the null hypothesis of AUC = 0.70. Similarly, the model including APACHE II score and PT for predicting 28-day mortality yielded a mean AUC of 0.836, with an empirical power of 0.996. Overall, these findings enhance the credibility of our results and support the clinical relevance of AT and sTM in the early prediction of DIC, as well as the utility of APACHE II and PT in mortality risk stratification among septic patients. Future studies with larger cohorts and external validation across multiple centers are warranted to further confirm these observations.

Limitations

This study has several limitations. First, it was a single-center study with a relatively small sample size, and the majority of participants were elderly individuals aged over 60 years. The mortality rate among patients with DIC was relatively high, which may limit the external validity of the findings. Second, there was some variability in the timing of AT and sTM measurements among patients, and the dynamic trends of these biomarkers were not evaluated. Finally, this study primarily assessed predictive performance using the area under the curve (AUC), without establishing a scoring system or clinical decision support tool, which limits its practical applicability. Consequently, further validation through multicenter studies with larger sample sizes is essential to confirm the generalizability of these findings.

Conclusion

Our findings suggest that elevated levels of sTM and reduced AT may be associated with the early development of disseminated intravascular coagulation in patients with sepsis. Nevertheless, sTM, TAT, t-PAIC, PIC, along with AT, demonstrated limited predictive value for 28-day mortality in these patients.