Factors influencing antithrombin activity following supplementation in sepsis-associated disseminated intravascular coagulation

Abstract

Background

See More

Antithrombin, a key regulator of the coagulation cascade, is often decreased in patients with sepsis-associated disseminated intravascular coagulation (DIC). Antithrombin is commonly supplemented when activity levels fall to 70% or below in Japan. While there is considerable interindividual variability in antithrombin activity following treatment, the factors contributing to this variability remain unclear. This study aims to identify the determinants of post-treatment antithrombin activity levels and to investigate the potential association between antithrombin activity and bleeding risk.

Methods

We conducted a retrospective analysis using data from the post-marketing surveillance of antithrombin concentrate in patients with sepsis-associated DIC. Changes in antithrombin activity were calculated as: (Day 1 activity − baseline activity [%]) divided by the daily dose (international units [IU] per kilogram of body weight). Logistic regression analysis was employed to identify factors associated with changes in antithrombin activity following supplementation and factors related to bleeding risk. Additionally, Kaplan–Meier survival curves were used to examine the relationship between antithrombin activity and 28-day survival outcomes.

Results

A total of 1,524 patients were included in the analysis. The median baseline antithrombin activity was 49%, which increased to 74% on day 1 post-treatment. The mean change in antithrombin activity was 0.99% /IU/kg and followed a normal distribution. The SOFA score ≥ 13 (p = 0.035) and FDP score ≥ 3 (≥ 25μg/mL), part of the JAAM DIC score, (p = 0.016) were significantly associated with lower antithrombin activity increase. Patients achieving ≥ 1% /IU/kg increase showed a higher 28-day survival rate (relative risk: 0.72, p = 0.004). No significant association was found between antithrombin doses or activity changes and bleeding risk.

Conclusion

A higher SOFA score and FDP level were associated with a smaller increase in post-treatment antithrombin activity. There was no clear association between antithrombin doses and bleeding risk. The present study suggests the necessity of individualized dosing beyond weight-based regimens.

Introduction

Antithrombin is a critical inhibitor of thrombin and factor Xa and plays a central role in regulating coagulation [1]. Antithrombin activity is significantly reduced in sepsis-associated disseminated intravascular coagulation (DIC) [2, 3], and marked reduction is associated with poor prognosis [4]. Thus, the Japanese Clinical Practice Guidelines for Management of Sepsis and Septic Shock 2024 (J-SSCG2024) recommend antithrombin supplementation in patients with sepsis-associated DIC whose antithrombin activity is ≤ 70% [5]. In addition, recent studies has highlighted that patients with post-treatment antithrombin activity restored to ≥ 80% have significantly higher rates of DIC resolution and improved survival compared to those with insufficient recovery [4, 6].

Regarding the dose response, previous studies have reported that administering 1 international unit (IU)/body weight (kg) of antithrombin concentrate results in an approximately 1% increase in antithrombin activity in patients with DIC [7, 8]. Therefore, to achieve a target antithrombin activity, the required dosage can be calculated as (80% − actual antithrombin activity [%]) × body weight (kg). However, these prior studies were conducted in patients with trauma and not sepsis-associated DIC [7, 8]. Moreover, substantial variability has been shown to exist in the extent of post-treatment antithrombin activity following supplementation [9]. The variation in response is thought to be influenced by several factors, including the severity of organ dysfunction, imbalance in the coagulation–fibrinolysis system, and the underlying disease [10]. Nevertheless, the factors underlying these differences in sepsis-associated DIC have not yet been fully clarified. Several studies have raised concerns about administering antithrombin concentrate due to the risk of bleeding, particularly when concomitant anticoagulation such as heparin is used [11,12,13].

This study explores factors influencing changes in antithrombin activity after supplementation in patients with sepsis-associated DIC. Additionally, the relationship between antithrombin and bleeding risk is also investigated.

Materials and methods

Study design and setting

This retrospective observational study analyzed data from a cohort of sepsis-associated DIC patients registered in the post-marketing surveillance data of antithrombin concentrate (Neuart®, Japanese Blood Products Organization, Tokyo). Surveillance was conducted in compliance with the Good Post-Marketing Study Practice from April 2013 to April 2016 across 213 domestic hospitals. Ethics approval was obtained from the Ethics Committee of the Japanese Blood Products Organization (approval number 2024–005), and the study adhered to the principles of the Declaration of Helsinki. The requirement for informed consent was waived owing to data anonymity.

Study population

Data from the post-marketing surveillance database were used to collect 2,588 cases of septic DIC based on the Japanese Association for Acute Medicine (JAAM) DIC criteria [14]. The inclusion criteria were a JAAM DIC score of ≥ 4 and antithrombin activity of ≤ 70%. Patients with missing baseline JAAM DIC scores or antithrombin activity data were excluded. Additional exclusion criteria included allergies to antithrombin, leukemia, malignant tumors, cirrhosis, or post-cardiopulmonary arrest, per the original survey exclusion criteria. The duration and dosage of antithrombin treatment were based on clinical judgment, with permitted co-administration of other anticoagulants. The standard dosage of antithrombin for DIC in Japan is 1,500 IU (30 IU/kg), with doses up to 60 IU/kg approved for emergency cases; administration is typically completed by the third day. Antithrombin concentrate is administered through drip infusion, and sample collection for activity assessment conducted the following morning. Definitions for Systemic Inflammatory Response Syndrome (SIRS), sepsis, severe sepsis, and septic shock followed the consensus of the American College of Chest Physicians/Society of Critical Care Medicine (Sepsis-1 and Sepsis-2) [15, 16]. Organ dysfunction was assessed using the Sequential Organ Failure Assessment (SOFA) score [17].

Data collection

The following baseline parameters were evaluated before antithrombin concentrate administration (day 0) for the cohort analysis: demographic data (sex, age, and body weight), site of infection, antithrombin dosage, concomitant therapies, JAAM DIC score, SOFA score, and antithrombin activity. Antithrombin activity was reassessed on day 1 (one day after treatment). The change in antithrombin activity was calculated as ‘[Day 1 antithrombin activity – baseline (Day 0) antithrombin activity] divided by the daily dose (IU/kg).’ Attending physicians assessed bleeding occurrence, severity, and causality. Major bleeding was defined according to the International Society on Thrombosis and Haemostasis (ISTH) definition [18].

Statistical analysis

Continuous variables were expressed as means with standard deviations or medians with interquartile ranges, depending on the distribution of the data. Normality was assessed based on skewness, kurtosis, and the difference between the mean and median. To identify factors associated with antithrombin response, the change in antithrombin activity per IU/kg was pre-specified to be analyzed as a binary variable, using the population mean as the cutoff. This approach was chosen to enable a statistically interpretable dichotomization based on a normally distributed variable. Risk factors associated with antithrombin activity and bleeding changes were assessed using univariate and multivariate logistic regression analyses, with the first category used as the reference for categorical variables. For the multivariate models, candidate variables were selected using a backward elimination method with a significance level for removal set at α = 0.15. Results are reported as odds ratio (OR), 95% confidence interval (CI), and p-values. Kaplan–Meier survival curves stratified by changes in antithrombin activity were generated to evaluate the cumulative mortality from days 0–28, and log-rank tests were performed. To account for potential confounding by baseline severity, a multivariate Cox proportional hazards analysis was also conducted with adjustment for baseline SOFA and JAAM DIC scores. Statistical significance was set at p < 0.05. All statistical analyses were performed using R (version 4.1.1, R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient characteristics

The patient selection process is summarized in Fig. 1. A total of 2,291 patients were included in the bleeding risk analysis. Among them, 1,524 patients were included in the analysis of antithrombin activity changes. Patient characteristics are presented in Table 1. The median age of the antithrombin activity change analysis cohort was 73 years, 58.1% (n = 886) were male, and the median body weight was 54.6 kg. The most common site of infection was the lungs, which occurred in 30.9% of the study cohort, followed by the abdomen (20.3%), urinary tract (15.6%), unknown origin (14.2%), and skin/soft tissue (8.1%). The daily dose of antithrombin concentrate administered was 1,500 IU, with a duration of administration of 3 days. Antithrombin administration was initiated on the same day as DIC diagnosis in most patients (median: 0 days). Concomitant therapies included thrombomodulin in 60.7% of patients and heparin and protease inhibitors in 14.1% and 18.6% of cases, respectively. The median JAAM DIC score was 5, and the SOFA score was 9. The median baseline (day 0) antithrombin activity was 49%, which increased to 74% on day 1. Among the 1,524 patients, 288 (18.9%) died within 28 days.

Fig. 1

Flow Chart for Patient Selection. Out of a total of 2,588 patients registered in the post-marketing surveillance of antithrombin concentrate, 2,291 patients with a JAAM DIC score ≥ 4 and baseline antithrombin activity ≤ 70% were included in the analysis of bleeding risk. Among these, 1,524 patients with available data on antithrombin activity on Day 1, body weight, and daily dose were further included in the analysis of changes in antithrombin activity. JAAM DIC: Japanese Association for Acute Medicine disseminated intravascular disease

Table 1 Patient characteristics

Antithrombin activity on day 1

The distribution of change in antithrombin activity is illustrated in Fig. 2. These changes followed a normal distribution with a mean of 0.99, a standard deviation of 0.88, and a median of 0.88 (interquartile range: 0.52–1.30).

Fig. 2

Distribution of changes in antithrombin activity. Changes in antithrombin activity were calculated as (Day 1 antithrombin activity − Baseline antithrombin activity) divided by the daily dose (IU/kg). The mean change was 0.99, with a standard deviation of 0.88 and a median of 0.88 (interquartile range: 0.52–1.30). IU: international unit

Logistic regression analysis was conducted to examine the factors associated with changes in antithrombin activity (Fig. 3). The objective variable was defined as whether the cutoff for the change in antithrombin activity was 1%/IU/kg. In univariate analysis, SOFA scores of ≥ 13 were significantly associated with a reduced change in antithrombin activity [(Fig. 3A), OR: 0.72, 95% CI: 0.53–0.98, p = 0.035]. Similarly, although not statistically significant, an increase in the JAAM DIC score was suggested to be associated with a smaller increase in antithrombin activity (Fig. 3B). In multivariate analysis of the individual components of the JAAM DIC score, higher platelet, PT-INR, and FDP scores were associated with a reduced change in antithrombin activity (Fig. 3C). Among these, an FDP score of ≥ 3 was significantly associated with a reduced change in antithrombin activity (OR: 0.66, 95% CI: 0.47–0.93, p = 0.016). Other patient-related factors evaluated, including age, sex, source of infection, and concomitant therapies, were not significantly associated with changes in antithrombin activity in both univariate and multivariate analyses (Suppl. Table 1). Kaplan–Meier survival analysis (Fig. 4) revealed that patients who achieved an increase in antithrombin activity of 1%/IU/kg had a significantly higher 28-day survival rate than those who did not (84.7% vs. 78.7%, relative risk: 0.72, p = 0.004). To account for potential confounding, a multivariate Cox proportional hazards model including SOFA and JAAM DIC scores as covariates was performed. The result showed that an increase in antithrombin activity ≥ 1%/IU/kg remained independently associated with lower 28-day mortality (hazard ratio: 0.75, 95% CI: 0.63–0.91, p = 0.003) (Suppl. table 2), consistent with the unadjusted Kaplan–Meier analysis.

Fig. 3

Association of SOFA score and JAAM DIC score with changes in antithrombin activity. Univariate and multivariate logistic regression analyses were performed to assess the impact of the severity of organ dysfunction and coagulation activation on changes in antithrombin activity. A Univariate analysis of SOFA score quartiles. B Univariate analysis of JAAM DIC score. C Multivariate analysis of JAAM DIC components (SIRS, platelet count, PT-INR, FDP score). SOFA: sequential organ failure assessment, JAAM DIC: Japanese Association for Acute Medicine disseminated intravascular coagulation, SIRS: systemic inflammatory response syndrome, PT-INR: prothrombin time-international normalized ratio, FDP: fibrin/fibrinogen degradation products

Fig. 4

Kaplan–Meier survival analysis based on changes in antithrombin activity. The patients who achieved an increase in antithrombin activity of 1%/IU/kg had a significantly higher 28-day survival rate than those who did not (84.7% vs. 78.7%, relative risk: 0.72, p = 0.004). Blue line; 1%/IU/kg or more. Red line; less than 1%/IU/kg. Shaded areas; 95% confidence intervals. IU: international unit

Antithrombin concentrate and bleeding risk

The timing and frequency of bleeding events following antithrombin concentrate administration are shown in Fig. 5. A total of 127 patients (5.5%) experienced an adverse bleeding event after antithrombin concentrate administration, of whom 10 (0.4%) were classified as having a major bleeding event. One-Quater (25.2%) of all bleeding events occurred on the first day of antithrombin administration. Logistic regression analysis was performed to evaluate the association between bleeding risk and antithrombin-related factors as well as concomitant anticoagulants (Table 2). Neither the daily dose of antithrombin, antithrombin activity on Day 1, nor the change in antithrombin activity was associated with an increased risk of bleeding. Additionally, the concomitant use of anticoagulants—including heparin, protease inhibitors, and thrombomodulin—was not significantly associated with increased bleeding risk.

Fig. 5

Time to onset and frequency of bleeding events following antithrombin concentrate administration. The histogram illustrates the frequency of bleeding events over time following antithrombin supplementation. Gray bars represent all bleeding events. Black bars denote major bleeding events

Table 2 Logistic regression analysis of bleeding risk associated with anticoagulant use (All bleeding)

Discussion

This study demonstrates an association between an increase in antithrombin activity of less than 1%/IU/kg with worse patient outcomes. Our findings demonstrate that higher SOFA scores and higher FDP levels are associated with a smaller increase in antithrombin activity.

As previously performed in trauma-associated DIC [7, 8], antithrombin activity increased by approximately 1%/IU/kg in our study. Approximately 38.2% of patients with a baseline SOFA score > 13 and 39.7% with an FDP score > 25 μg/mL showed a 1%/IU/kg or more increase in activity. Patients with greater organ dysfunction and stronger coagulation/fibrinolysis activation exhibited a smaller increase in antithrombin activity, possibly due to increased consumption, suppressed hepatic production, and vascular leakage [19]. These findings underscore the importance of considering these factors when determining the optimal dosing and timing for administering antithrombin concentrate.

The observed association between increased antithrombin activity and improved survival may partly reflect baseline severity bias, as patients with greater responses might have had more favorable clinical conditions at the outset. To address this concern, we conducted a multivariate Cox proportional hazards analysis adjusted for baseline SOFA and JAAM DIC scores. The association between increased antithrombin activity and improved survival remained statistically significant after adjustment, supporting the robustness of the observed relationship.

To improve prognosis, the cutoff of post-treatment antithrombin activity was reported to be approximately 80% [6]. Given that the baseline antithrombin activity in our cohort was approximately 49% and the average body weight was 54 kg, patients who received a standard dose of 1,500 IU/day likely reached this cutoff. This suggests that an increase of 1%/IU/kg antithrombin activity may differentiate favorable therapeutic outcomes. However, this response was not consistent across all patients—those with higher baseline SOFA scores or elevated FDP levels appeared to require larger doses of antithrombin to achieve comparable increases. Notably, the 80% target level has been suggested in retrospective studies but has not been validated in prospective trials. Therefore, further prospective investigation is warranted to determine the optimal target antithrombin activity level.

Previous reports, such as the KyberSept trial, have suggested that high-dose antithrombin administration may increase bleeding risk [11]. On the contrary, our study did not find a direct association between antithrombin activity and bleeding risk. Concomitant use of anticoagulants, including heparin, protease inhibitors, and thrombomodulin, was also not associated with increased bleeding. A sensitivity analysis limited to major bleeding events yielded similar results (Suppl. Table 3). In the KyberSept trial, approximately 70% of patients received heparin during the first 4 days of treatment, while in our cohort only 14.1% did. Moreover, the total antithrombin dose in KyberSept was 30,000 IU over 4 days, markedly higher than the standard Japanese regimen of 1,500 IU/day, typically administered for 3 days. These differences in both concomitant heparin use and total antithrombin dose may partly explain the lower bleeding incidence observed in our cohort. In our cohort, the overall incidence of bleeding was 5.5%, consistent with previous reports such as Umemura et al. [20]. Similarly, J-SSCG 2024 notes that bleeding complications may increase by 8 per 1,000 patients (95% CI: − 24 to + 89) with antithrombin administration, based on randomized controlled trial data [5]. These findings suggest that although anticoagulant therapy using antithrombin may carry a risk of bleeding, the risk appears to be lower in patients with sepsis-associated DIC.

This study had several limitations. First, because of the retrospective nature, missing data were relatively common. In particular, antithrombin activity on Day 1 was frequently missing. Second, the antithrombin dosage was not standardized and was administered at the discretion of the attending physician, potentially introducing variability in dosing and treatment decisions. Although post-administration activity levels were available, treatment adjustments based on these values were not protocolized, raising the possibility of clinician-driven bias and residual confounding. Third, although we utilized only baseline and Day 1 data, 58.5% of cases received treatment over three days, which might affect patient outcomes. Finally, because our analysis only included patients who received antithrombin, we were unable to compare the incidence of bleeding with that in patients without treatment.

Conclusions

Increased levels of post-treatment antithrombin activity were associated significantly with patient outcomes. A higher SOFA score and FDP level were associated with a smaller increase in antithrombin activity following supplementation in patients with sepsis-associated DIC. Since no clear association was found between the antithrombin dose or activity change and bleeding risk, dosing informed by baseline SOFA score and FDP level may prove impactful. It was possible that post-treatment antithrombin activity and clinical outcomes were influenced by baseline disease severity and antithrombin activity.