Antithrombin Activity Is Associated with Persistent Thromboinflammation and Mortality in Patients with Severe COVID-19 Illness

According to the Centers for Disease Control and Prevention (CDC), COVID-19, caused by SARS-CoV-2 virus infection, has been attributed as the cause of death for over 1 million people in the USA alone (as of 09/28/2022) [1]. COVID-19 patients have been shown to be at significant risk of developing severe thrombotic complications including venous thromboembolism (VTE) and strokes; however, the precise pathophysiology is unresolved [2]. Recent studies have shown that patients with severe COVID-19 exhibit pronounced coagulation abnormalities such as increased fibrinogen and D-dimer levels, both of which are indicative of a hypercoagulable state [2] and can lead to thrombotic complications. With the rise of new and more infectious SARS-CoV-2 variants and only 67.8% of the US population fully vaccinated (as of 09/28/2022), there is an urgent need to research and understand COVID-19 disease pathogenesis [3] for treatment and prevention of these thrombotic complications and their sequelae.

Antithrombin (AT) is an endogenous anticoagulant that inhibits several coagulation factors in the intrinsic and extrinsic coagulation pathways to prevent blood clot formation, most potently factor Xa (FXa) and thrombin [4]. Its inhibitory activity is accelerated in the presence of heparin, making AT an essential determinant of responsiveness to standard-of-care in-hospital VTE chemoprophylaxis. In addition, evidence indicates that AT possesses anti-inflammatory activity through its ability to modulate endothelial signaling upon binding to heparin-like proteoglycans expressed on the endothelial surface, an interaction that induces intracellular prostacyclin production and inhibition of nuclear factor kappa-B-mediated gene transcription [5‒7].

Due to these unique characteristics, AT plays a role in the pathogenesis of thromboinflammatory diseases, such as sepsis and acute respiratory distress syndrome (ARDS) [8]. Past research from our group has found that transient acquired AT deficiency during critical illness is linked to poor responsiveness to heparin-based thromboprophylaxis and strongly associated with VTE [9‒11]. In COVID-19, studies have found that patients have significantly reduced AT levels, similar to that seen in sepsis, ARDS, and trauma, indicating that AT deficiency could play an important pathophysiologic role in COVID-19 disease progression by repressing optimal VTE prophylaxis and facilitating aberrant thrombosis and thromboinflammation [12].

The goal of this study was to examine the link between AT activity and markers of hypercoagulability and inflammation among hospitalized patients with severe COVID-19. We hypothesized that AT deficiency would be associated with poor responsiveness to thromboprophylaxis, increased markers of hypercoagulability, increased markers of inflammation, and poor outcomes.

This prospective observational study was performed at the Memorial Hermann-Texas Medical Center and the University of Texas Health Science Center at Houston (UTHealth) from June 11, 2020, to April 18, 2021. This study protocol was approved by the UTHealth Committee for the Protection of Human Subjects (Institutional Review Board) and Memorial Hermann (HSC-MS-20-0539).

Patients admitted to the intensive care unit (ICU) who were at least 18 years old and had SARS-CoV-2 infection confirmed by PCR testing were included; pregnant women and prisoners were excluded. Patients with a known history of hematologic or immunologic disorders were also excluded. This study used an initial waiver of consent procedure due to the minimal risk posed by the study methodology to the patients involved. Delayed written consent or verbal consent via telephone was obtained to minimize influence from investigator-patient interactions and SARS-CoV-2 virus infection risk.

Standard-of-care thromboprophylaxis at our institution is low-molecular-weight heparin (enoxaparin) as recommended by the International Society on Thrombosis and Haemostasis (ISTH) for prophylaxis of thromboembolic events in all patients who are hospitalized for COVID-19 [13]. Enoxaparin is administered initially as 30–40 mg twice daily based on weight. Failure to achieve an anti-FXa level of 0.6 IU/mL prompts enoxaparin dose escalation by 10-mg increments every 12 h until the target anti-FXa is achieved. In addition, patients are transitioned to unfractionated heparin if D-dimer levels exceed 5 µg/mL, VTE is positively identified, the patient is placed on extracorporeal membrane oxygenation (ECMO), or upon diagnosis of acute kidney injury.

Hospital laboratory test results, patient demographics, and patient outcomes were collected from patient medical records. Demographic data collected included age, sex, race and ethnicity, and body mass index (BMI). Collected outcome data included thromboembolic events (deep vein thrombosis, pulmonary embolism, stroke), respiratory failure, ARDS, ECMO use, lowest daily oxygen saturation, length of stay, ventilator days, ICU days, and mortality. The study observation period was 7 days from ICU admission.

Blood samples were collected in conjunction with standard-of-care daily ICU laboratory draws; no separate blood samples were collected for research purposes only. All samples were analyzed in the hospital laboratory. Laboratory values evaluated specifically for this study included AT (measured in % activity) and anti-FXa levels. International normalized ratios (INRs) and D-dimer, fibrinogen, and C-reactive protein (CRP) levels were measured as part of routine laboratory testing in the ICU as these values are recommended for use in tracking coagulation status in patients with COVID-19 by the ISTH [13] and were obtained from the patient medical records. Laboratory values were evaluated daily for 7 days from ICU admission. If a patient was discharged prior to 7 days, the last available data were used for analysis.

Responsiveness to thromboprophylaxis was defined as achieving an anti-FXa level of greater than or equal to 0.6 IU/mL, and unresponsiveness was defined as an anti-FXa level of less than 0.6 IU/mL [14]. Spearman correlation coefficients were computed to assess correlation levels between laboratory markers. Longitudinal laboratory markers, including INR, fibrinogen, D-dimer, CRP, and AT, are presented as median with interquartile range. Comparisons between groups at distinct time points were evaluated by Mann-Whitney tests. Bonferroni adjustment was used to adjust for multiple comparisons from day 1 to day 7 for each marker.

The issue of missing values arose in the longitudinal measurements of laboratory markers. The dropout (missing value) rates were 33%, 36%, 31%, 39%, and 58% for INR, fibrinogen, D-dimer, CRP, and AT, respectively. Survivors had a higher proportion of dropout events compared to deceased patients. To test random dropouts, we fitted a logistic regression model for dropout events at each time that dropouts occurred. The model included 2 covariants, mortality and marker measurement at the prior time [15]. This model tested whether abnormal marker values caused immediate dropout and whether dropout rates significantly differed between survivors and deceased patients. Random dropouts were tested to be satisfactory for all markers. As a result of this information, we deemed it proper to perform a 2-sample comparison at later time points.

Multivariable analyses were performed to evaluate the effect of the AT activity level on mortality, adjusting for important demographic factors. We considered AT activity at the end of the 7-day observational window or at the day of discharge if the patient was discharged prior to the 7th day since ICU admission. The AT activity level was dichotomized according to the median, which was 100%, and included in the logistical regression model for mortality that also included age, sex, and BMI. BMI was dichotomized into obesity and non-obesity categories based on the threshold value of 30. We performed the Hosmer-Lemeshow test to assess goodness-of-fit for the logistic regression model. The test result suggested that the model fitted the data adequately. We reported 2-sided p values. A p value <0.05 was considered statistically significant. All statistical analyses were performed using the SAS software (version 9.4, SAS Institute).

Due to institutional regulations during the pandemic on patient enrollment in COVID-19 studies, we were only able to enroll 36 patients. However, prospective power calculations indicated that a sample size of 24 patients would provide 90% power to detect a 10% difference in AT activity between those who survived and died, based on an assumed mean AT activity in healthy individuals of 112% with a standard deviation of 16.8%.

A total of 36 patients were enrolled. All patients were confirmed to have SARS-CoV-2 infection by PCR testing. Patient characteristics are provided in Table 1. Patients had a median (interquartile range) age of 60 (40.25, 65) years, and 16 (44%) were female. In general, outcomes were poor as 32 (89%) patients developed ARDS and 18 (50%) patients died. Causes of death included acute respiratory failure (N = 16; 89%) and hemorrhagic stroke (N = 2; 11%). VTE screening by duplex ultrasound was performed on 9 patients (25%). Three patients developed thrombotic complications (8.3%), including 2 VTEs (1 pulmonary embolism and 1 combined deep vein thrombosis and pulmonary embolism in the same patient) and 1 stroke event. One unit of convalescent plasma was administered to 18 patients (50%) for treatment of COVID-19.

All patients were initially placed on prophylactic enoxaparin for thromboprophylaxis. Five patients (13.9%) received enoxaparin dose escalations due to failure to achieve an anti-FXa level of ≥0.6 IU/mL. In addition, 5 patients (13.9%) received dual thromboprophylaxis with a combination of enoxaparin and aspirin (81 mg once daily) due to underlying cardiovascular disease prior to hospitalization. Finally, 13 patients were transitioned to unfractionated heparin during hospitalization due to elevated D-dimer, the presence of VTE or acute kidney injury, or placement on ECMO. There were no significant differences in enoxaparin dosing between patients who survived or died, and both anti-FXa levels and rates of achieving a prophylactic anti-FXa (≥0.6 IU/mL) were similar between patient groups (p > 0.05).

Standard coagulation laboratory values (INR, D-dimer, fibrinogen, CRP) were compared between patients who survived and died. Greater ongoing hypercoagulability was evident in the patients who died, as demonstrated by significant increases in circulating fibrinogen levels on ICU days 4 and 5, as well as significantly elevated D-dimer levels on ICU day 4 compared to those who survived (Fig. 1a, b). No significant differences in INR, a marker of coagulopathy, were observed between groups (Fig. 1c). In addition, CRP values were available on 18 patients; in this subgroup, 9 patients died. We observed significant increases in CRP levels on ICU days 4 and 5 among patients who died in this subgroup (Fig. 1d), indicating a more pronounced proinflammatory state.

AT activity between patients who survived and died was compared over the sampling period (Fig. 2). No differences in AT activity were observed between patients who did and did not receive convalescent plasma therapy. Patients who died had significantly lower AT activity on ICU day 5 compared to those who survived (p < 0.05). In addition, the AT activity on the 7th day or day of discharge since ICU admission was significantly lower among patients who died compared to those who survived (92%; 85, 104 vs. 114%; 99, 120, respectively; p < 0.05). To examine the independent association between AT activity and mortality, patients were dichotomized according to the median AT activity level (100%). AT ≥100% versus AT <100% was evaluated in a logistic regression model for death, controlling for age and sex (Table 2). Compared to AT activity <100%, the odds of death for AT activity ≥100% was only 12% (95% CI = 0.02–0.65), which translates to an 8.1-fold increased odds of death in COVID-19 patients with AT activity levels <100% (95% CI = 1.53–42.5).

AT activity had a significant negative correlation with D-dimer levels on ICU days 1 and 7 (Table 3). The small sample size led to very wide confidence intervals for correlation coefficients. The finding about correlation of AT activity with D-dimer level needs to be validated by large-scale studies. No significant relationships were identified between AT activity and anti-FXa levels, fibrinogen levels, or INRs. In the subgroup analysis of 18 patients with available CRP data, we identified a moderate inverse relationship between AT activity and CRP levels, particularly on ICU days 3 and 6; however, these data were not statistically significant due to the small sample size.

COVID-19 patients are at significant risk of developing life-threatening thrombotic complications [16‒18]. The goal of this study was to assess the relationships between laboratory markers of thromboinflammation, responsiveness to VTE prophylaxis, and mortality with time-dependent changes in AT activity. Our prospective, single-center study found that in spite of sufficient anticoagulation and a low VTE rate in our population, patients who died exhibited a persistent hypercoagulable and proinflammatory state compared to patients who survived. AT activity was significantly reduced on ICU day 5 among COVID-19 patients who survived and was associated with both markers of hypercoagulability and inflammation. Additionally, logistic regression analysis showed that less than 100% AT activity was associated with an 8.1-fold increase in risk of death. Together, these data indicate that patient death due to COVID-19 is associated with a persistent thromboinflammatory state in spite of sufficient thromboprophylaxis and that AT plays an important role in regulating thromboinflammation during COVID-19 illness.

The observed link between mortality and AT is supported by the findings of other studies. Anakli et al. [12] showed that AT activity levels in patients with ARDS and COVID-19 could potentially serve as a prognostic marker for survival and organ failure; however, this finding was limited by a retrospective approach. In addition, Joshi et al. [19] demonstrated significant reductions in AT activity among COVID-19 patients who died compared to those who survived. Finally, Ranucci et al. [20] showed that time-dependent increases in AT activity were associated with coagulation normalization among recovering COVID-19 patients. Our findings validate and build upon these reports in a larger, prospective study using multivariable modeling to identify the independent relationship between AT activity and COVID-associated mortality.

Our study identified a strong inverse relationship between AT activity and D-dimer levels, indicating that low AT levels contribute to increased activation of the coagulation system. Interestingly, we did not identify a significant correlation between AT activity and anti-FXa levels, which are used clinically to assess responsiveness to heparin-based anticoagulants. This finding could be due to the fact that, in general, AT activity remained well within the normal range of 80–120% in our patient population. Logistic regression analyses identified a significant increase in the risk of death below an AT activity of 100%. However, our past work in other critically ill populations has identified AT activity at a threshold within a normal range is linked to poor outcomes. Specifically, our group demonstrated that the risk of VTE following trauma and hemorrhagic shock is significantly greater for patients with an AT activity <90%, indicating that the definition of “normal” AT activity during critical illness may require redefining [10].

Part of the reason that high AT activity is necessary to prevent death during COVID-19 illness could be due to its anti-inflammatory action. Although our small sample size precluded robust statistical significance, subgroup analyses did identify a moderate correlation between AT activity and the inflammatory marker CRP. AT regulates the host inflammatory response to infection and injury both directly, through its modulation of endothelial-mediated anti-inflammatory activities, and indirectly, through its inhibition of proinflammatory coagulation enzymes [5, 21]. Past research has demonstrated AT administration significantly improves survival in rodent sepsis models but only at supraphysiologic levels [22]. This could be reflective of the greater amount of AT necessary to counteract the pronounced procoagulant and proinflammatory response induced by bacteremia or could indicate that AT’s anti-thromboinflammatory function is dependent on its interaction with the vascular endothelium. Further in vitro and in vivo studies are necessary to conclusively determine the mechanistic basis for AT’s protective effects during COVID-19 disease.

Several important limitations of the study should be noted. This study was conducted at a single center with a small sample size as patient enrollment was limited at our institution during this phase of the pandemic. This limits the generalizability of our observations and our ability to detect significant differences in some analyses. In addition, our observation period was limited to 7 days; therefore, we could have failed to capture some additional laboratory results that could influence data interpretation. Also, the overall incidence of thrombotic complications was low. This may be due to the consistent adherence to the aforementioned ISTH guidelines on recommending prophylactic low-molecular-weight heparin for all hospitalized COVID-19 patients. Further, this does not exclude the possibility of undetectable microvascular thrombotic events that could have contributed to organ injury in these patients. Due to the pragmatic nature of the study, pre-planned VTE screening, such as duplex ultrasound, was not performed and represents a limitation. Additionally, the target anti-FXa thresholds were higher, 0.6 IU/mL, than for other hospitalized patients [11]. Finally, this study is associative in nature. Examination of the role of AT as an anti-inflammatory agent in severe COVID-19 and the use of maintaining AT greater than 100% as a clinical goal in the treatment of COVID-19 patients should be explored in the future on a larger scale to fully demonstrate the depth of the association between mortality and AT.

AT activity is associated with mortality in hospitalized COVID-19 patients. Low AT activity is connected with the persistent hypercoagulable and proinflammatory state apparent in patients who died. The anti-thromboinflammatory properties of AT could make it an appealing therapeutic target for future investigations toward improving outcomes during severe COVID-19 illness.

The authors would like to thank Kimberly A. Mankiewicz, PhD, UTHealth, for writing and editing assistance.

This study protocol was reviewed and approved by the Committee for the Protection of Human Subjects (Institutional Review Board) of the University of Texas Health Science Center at Houston and the Memorial Hermann Health System, approval number HSC-MS-20-0539, and conducted in accordance with the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Written informed consent or verbal consent via telephone was obtained within 72 h of admission from either the patient (or legally authorized representative if the patient was unable to provide consent). Written informed or verbal consent via telephone within 72 h of admission from patients and from a legally authorized representative for all vulnerable participants was obtained. Verbal consent was confirmed to be approved by the Committee for the Protection of Human Subjects of the University of Texas Health Science Center and Memorial Hermann Health System, who approved our study. A waiver of consent was obtained if a patient was discharged or died within 24 h. Waiver of consent in the event of patient discharge or death within 24 h was confirmed to be approved by the Committee for the Protection of Human Subjects of the University of Texas Health Science Center and Memorial Hermann Health System, who approved our study.

JCC has received speaker honoraria from Grifols. The authors (JCC, CEW, BAC) have a patent on the therapeutic use of AT in the acute care environment (17/146,91212).

This work was supported by a Grifols Investigator-Sponsored Research Award to JCC. The sponsor had no role in study design; collection, analysis, and interpretation of data; or writing of the report. JCC is additionally supported by grants from the Department of Defense and Aniara. CEW is supported by grants from the Farrish Fund, Howell Family Fund, and Red Duke Distinguished Professorship.

All the authors met ICMJE authorship criteria. Amber Chen-Goodspeed designed the research, interpreted data, and wrote the manuscript. Gouthan Dronavalli and Bela Patel designed the research, collected data, and edited the manuscript. Xu Zhang analyzed the data and edited the manuscript. Jeanette M. Podbielski designed the research, collected data, and edited the manuscript. Katalin Modis assisted with data analysis, interpretation, and editing the manuscript. Bryan A. Cotton and Charles E. Wade assisted in study design, data interpretation, and editing the manuscript. Jessica C. Cardenas designed the research and assisted in data analysis and interpretation and wrote the manuscript. All the authors have read and approved this manuscript and agree to be accountable for all aspects of the work.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.