Association between Thrombin Generation and Clinical Characteristics in COVID-19 Patients

Coronavirus disease 2019 (COVID-19) is associated with thromboinflammation, in vivo activation of coagulation, endothelial dysfunction, hypofibrinolysis, and an increased risk of sepsis-induced coagulopathy (SIC), venous, arterial, and microangiopathic thromboses as well as bleeding complications [1‒9]. The International Society on Thrombosis and Hemostasis (ISTH) guidelines suggest monitoring prothrombin time (PT), D-dimer, platelet count, and fibrinogen as they may have prognostic implications in hospitalized patients with COVID-19 [10]. Additionally, anticoagulant therapy was found to improve prognosis in hospitalized COVID-19 patients under a variety of circumstances [11‒13]; therefore, prophylactic dose low molecular weight heparin (LMWH) is uniformly supported by international guidelines for these patients [10, 13‒15].

Thrombin generation (TG) is a global coagulation assay that provides a quantitative assessment of clot formation, aiming at evaluation of the entire coagulation process. TG is more sensitive than the PT and activated partial thromboplastin time (aPTT) tests and can identify moderate changes in thrombin levels, which are sufficient to have clinical implications, thus indicating an increased risk of thrombosis or hemorrhage. It is valid for comparison of different states within the same patient, such as before and during anticoagulant therapy, and can be predictive for both primary and recurrent venous thromboembolism (VTE) [16‒18].

The use of TG as an ancillary tool enabling personalized tailoring of therapy for patients with severe bleeding disorders as well as prothrombotic conditions has previously been reported, and our group demonstrated that TG could predict thrombotic events in patients with multiple myeloma which may warrant anticoagulant therapy [19], as well as guide treatment with bypassing agents for pediatric hemophilia B patients with high-responding inhibitors [20]. A high TG capacity has been previously reported among COVID-19 patients [21], and despite thromboprophylaxis, TG levels remained similar to those of healthy individuals [22, 23]. The association between TG parameters, disease severity, clinical prediction scores, and outcomes in these patients has not been well described. We aimed to assess the relationship of TG with the sequential organ failure assessment (SOFA) [24] and SIC [25] scores and to investigate the association of TG with clinical course and outcomes in hospitalized COVID-19 patients.

Consecutive adult (age >18 years) COVID-19 patients who were hospitalized at Sheba Medical Center (Ramat Gan, Israel) between March 1st and October 30th, 2021 were included. SARS-CoV-2 infection was confirmed by quantitative real-time reverse-transcriptase polymerase chain reaction according to the diagnostic criteria defined by the World Health Organization (WHO) [26]. Patients transferred to intensive care units (ICUs) or those who failed to provide informed consent were excluded from our study.

The study was approved by the Institutional Ethics Committee at Sheba Medical Center. All patients provided written informed consent to participate. Aside from plasma collection for TG, no investigational procedure, intervention, or treatment was mandated by this study.

Blood samples were obtained during patients’ hospitalization in 2.7 mL BD Vacutainer® vacuum tubes containing 0.109 M buffered sodium citrate. In 16 out of 32 patients, sequential samples were obtained from the time of hospitalization to discharge. Repeated samples for coagulation as well as TG studies were obtained according to physicians’ discretion and patients’ clinical course of disease. All samples were centrifuged and frozen at −70°C for TG analysis. Additional samples from healthy individuals were analyzed for TG to establish reference values.

TG was measured in platelet-poor plasma (PPP) by a calibrated automated thrombogram method as previously described by Hemker et al. [27]. Eighty µL of PPP were added to 20 µL of buffer (PPP-Reagent LOW; Diagnostica Stago, Gennevilliers, France; final concentration ∼1 pM tissue factor; and ∼4 µM phospholipids) and placed in round-bottom 96-well plates. TG was initiated by the adding 20 µL of a fluorogenic substrate/CaCl₂ buffer (FluCa Kit; Stago Inc, Parsippany, NJ, USA). Endogenous thrombin potential (ETP) (nM × min) and thrombin peak height (nM) were calculated according to fluorescence intensity by a dedicated software (version 3.0.0.29; Thrombinoscope BV, Maastricht, Netherlands). All measurements were performed at least in duplicates.

All patients were tested for routine coagulation tests (PT, aPTT, fibrinogen, D-dimers) and complete blood counts upon admission and as needed during hospital stay. Coagulation assays were analyzed by the ACL-TOP 750 autoanalyzer (Instrumental Laboratories, Bedford, MA, USA), and complete blood counts were analyzed by the Beckman-Coulter analyzer (Brea, CA, USA). Demographic and clinical data were retrieved from the computerized medical records. SOFA [24] and SIC [25] scores were calculated for all patients. Disease severity was defined in accordance with the NIH guidelines on the clinical spectrum of SARS-CoV-2 infection [28].

Statistical analysis was performed with IBM SPSS Statistics (version 23.0; IBM Corp.). Continuous variables were presented as median [interquartile range (IQR) or range, as indicated]. Categorical variables were presented as counts, proportions, and/or percentages. The Mann-Whitney U test was used to compare patients versus healthy controls or 2 independent patient subgroups (mild + moderate vs. severe + critical disease, vaccinated vs. non-vaccinated, prophylactic vs. higher than prophylactic anticoagulation, survivors vs. non-survivors). The Friedman test followed by Dunn’s post hoc test was used to compare the patients’ TG parameters on different occasions. The Spearman correlation test was used to evaluate the association of the TG parameter with routine coagulation tests, SOFA, SIC scores, and disease severity. Two-tailed p values of less than 0.05 were considered statistically significant.

Thirty-two patients (68.8% male) were assessed during their hospitalization at our center. The median age was 69 (IQR 49–76, range 29–90) years. Among patients’ comorbidities, hypertension (n = 17), cardiovascular illness (n = 17), dyslipidemia (n = 14), and malignancy (n = 14) prevailed. Eight patients (25%) had no comorbidities, whereas 22 patients (68.8%) had at least two comorbidities. Most patients (n = 20) were vaccinated with at least one dose of the BNT162b2 vaccine at the time of their hospitalization, and 17/20 had received two or three doses.

COVID-19 severity was defined as mild in 21.9% (n = 7), moderate in 9.4% (n = 3), severe in 62.5% (n = 20), and critical in 6.3% (n = 2). Demographic data of our patients are presented in Table 1. All but 3 patients (90.6%) received anticoagulant therapy while hospitalized. Most patients (n = 20) received prophylactic doses of LMWH (enoxaparin 40 mg QD). Six patients received therapeutic doses of anticoagulation (enoxaparin 1 mg/kg BID or equivalent doses of direct oral anticoagulants), including 4 patients due to preexisting atrial fibrillation and 2 patients for the treatment of VTE. Three patients received an intermediate dose of LMWH.

All patients received a supportive care protocol which included supplemental vitamins and zinc. Twenty-six patients (81.3%) were treated with dexamethasone, 21 (65.6%) received remdesivir, and 7 (21.9%) received immunomodulatory agents (tocilizumab, sarilumab, or baricitinib) (Table 2).

Two patients had thrombotic events, of which one was a lower extremity deep vein thrombosis and the other was pulmonary embolism. These patients received a prophylactic and an intermediate dose of LMWH, respectively, which were increased to a full therapeutic dose upon VTE diagnosis.

Three patients experienced bleeding events. The first patient had received prophylactic LMWH and had a gastrointestinal hemorrhage. The second patient had received prophylactic LMWH and bled during endotracheal intubation. The third patient had received a full therapeutic dose of LMWH due to preexisting atrial fibrillation, had thrombocytopenia with values around 70,000/µL, experienced bleeding from various sites (gastrointestinal, central venous catheter, insertion site, and endotracheal tube), and eventually died of sepsis.

Following hospitalization, 12 patients (37.5%) died, of whom 8 mortalities were directly related to COVID-19. When compared between COVID-19 survivors and the patients who died during hospitalization, no difference was observed regarding disease severity per NIH definition (p = 0.703), vaccination status (p > 0.999), or anticoagulant dosing regimens (p = 0.854). Patients’ treatment and outcomes are presented in Table 2.

While PT and aPTT on admission were close to the normal reference values of our laboratory, fibrinogen was already mildly elevated (median 506, IQR 344–646 mg/dL), and D-dimers were uniformly increased (median 1,573, IQR 994–5,563 ng/mL FEU). When coagulation assays were reassessed during hospitalization, a significant impairment of PT values and elevated D-dimers were noted among patients who died as compared with survivors (Fig. 1).

The median SOFA score was 3 (IQR 2–6, range 0–10), and the median SIC score was 3 (IQR 2–4, range 0–6). Compared to survivors, non-survivors had higher median SOFA and SIC scores (p < 0.001 and p = 0.004) and were older, but the difference did not reach statistical significance (p = 0.078).

The initial lag time, ETP, and peak height from our cohort of anticoagulated COVID-19 patients did not significantly differ from the values obtained from non-anticoagulated healthy controls. ETP and peak height were slightly increased among patients with severe and critical COVID-19 as compared with mild and moderate disease per NIH definition (p = 0.136 and p = 0.099, respectively) (Fig. 2).

Fibrinogen levels positively correlated with COVID-19 severity (r_s = 0.47, p = 0.007) as well as with ETP (r_s = 0.48, p = 0.010) and peak height (r_s = 0.57, p = 0.001) obtained at admission (Fig. 3), while PT and aPTT correlated with only lag time (r_s = −0.41, p = 0.028 and r_s = 0.43, p = 0.023, respectively). No correlation between any of the TG parameters and D-dimer levels was detected. Patients who received higher than prophylactic doses of anticoagulant therapy demonstrated lower PT % activity and longer aPTT (p = 0.001 and p = 0.016, respectively) as well as longer lag time, lower ETP, and peak height (p = 0.003, p = 0.037, and p = 0.006, respectively) (Fig. 4). A significant shortening of the lag time was observed in patients with repeated TG testing (p = 0.023 when comparing TG samples obtained on three different occasions, n = 7); however, neither ETP nor peak height showed any significant change with time (data not shown).

No correlation was found between any of the TG parameters and the SOFA score; however, statistically significant negative correlation was detected between ETP and the SIC score (r_s = −0.40, p = 0.038) (Fig. 5). Nonetheless, TG parameters did not correlate with COVID-19 severity according to NIH definitions [28] and did not differ between survivors and non-survivors.

The TG parameters were similar among vaccinated as compared to non-vaccinated patients. As very few patients presented with thrombotic or bleeding complications, no significant associations between TG parameters and these outcomes could be established.

Coagulation activation and coagulopathy are inherent to COVID-19 infection and are associated with inferior outcomes. The coagulopathy of COVID-19 is unique and usually involves increased D-dimer and fibrinogen levels, fewer abnormalities in PT and platelet counts, and higher rates of thrombosis compared with non-COVID-19 disseminated intravascular coagulopathy and SIC [1‒4]. TG may represent an invaluable assay in predicting disease severity and outcomes [29].

In our study, 32 COVID-19 patients were assessed for TG during hospitalization. TG was not associated with mortality or disease severity according to the SOFA score; however, interestingly, a reduced ETP was noted to be associated with an increased SIC score. Notably, impaired TG was previously reported in patients with severe systemic infection and was associated with mortality [29]. In a study which measured prothrombin fragment 1 + 2, as indirect markers of TG, values obtained at admission were elevated, and sequential measurements discriminated survivors, whose values decreased at follow-up, from non-survivors, where values remained stable or increased [30].

Similar to previous reports [2‒4], we found elevated fibrinogen and D-dimers values already on admission, manifesting an acute phase reaction [31]. Higher fibrinogen levels were noted among our patients with severe and critical disease, and they correlated with higher ETP and peak height obtained at admission. Decreased PT % activity and increased D-dimers were associated with mortality (Fig. 1). A significant shortening of the lag time was observed in patients with repeated TG testing, possibly reflecting a worsening of the hypercoagulability; however, neither ETP nor peak height showed any significant change with time.

The rates of thrombosis (2/32 patients) in our study are lower than previously reported in COVID-19 patients [6‒8], as most patients were anticoagulated from the first day of hospital admission. On the other hand, rates of bleeding (3/32) and inhospital mortality (37.5%) are higher in our cohort [9]. The higher bleeding rates may be attributable to the almost uniform use of anticoagulant therapy in our institution; nonetheless, differences due to the relatively small number of patients and events cannot be ruled out.

In line with previous observations, TG parameters from COVID-19 patients were similar to those obtained from non-anticoagulated healthy controls, suggestive of a hypercoagulable condition unmitigated by anticoagulant therapy. A study from Italy found that patients with COVID-19 had an increased TG at diagnosis as reflected by greater peak height and ETP. Thromboprophylaxis reduced TG to levels of healthy controls. Furthermore, intermediate dose anticoagulation as compared to standard prophylactic dose, significantly reduced ETP in correlation with anti-Xa activity (r = 0.49, p = 0.0017) [21]. Likewise, in a French study of COVID-19 patients, peak height and ETP were not decreased despite thromboprophylaxis [22]. In concordance with these reports, our study patients who received higher than prophylactic doses of anticoagulant therapy had significantly increased lag time, lower ETP, and a reduced peak height (Fig. 4). Unfortunately, as only a small fraction of patients was sequentially evaluated, significant shortening of the lag time was observed with time, yet neither ETP nor peak height showed any significant difference.

The routine administration of higher than prophylactic doses of LMWH in patients with COVID-19 may be beneficial in selected non-critically ill patients [32], whereas critically ill patients and ICU patients do not benefit from therapeutic dose LMWH [32, 33]. In a pooled analysis, thrombosis and bleeding rates in patients who received higher than prophylactic doses of anticoagulation were similar to those in patients who received standard-dose thromboprophylaxis [34]. An updated guideline by the American Society of Hematology supports prophylactic intensity over higher-intensity anticoagulation in critically ill COVID-19 patients without a confirmed or suspected VTE [35]. Interestingly, ETP and peak height were slightly increased among our patients with severe and critical COVID-19 (Fig. 2), a finding which provides further support for the association of more severe illness with coagulation activation.

Early in the course of the pandemic, higher SOFA scores and D-dimer above 1,000 ng/mL FEU on admission were found to be associated with reduced survival [36], as was a SIC score ≥4 [37], whereas thromboprophylaxis was reported to be associated with improved survival in patients with a SIC score ≥4 [11]. The sequential organ failure assessment (SOFA) score has poor discriminative ability for death in severely or critically ill patients with COVID-19 requiring ICU admission, as most patients suffer mainly respiratory failure [38], and special modifications of this score have recently been suggested but require validation in large COVID cohorts [39]. Notably, severe COVID-19 with SIC implicates immune response, endothelial cell dysfunction, platelets, and complement activation, which are related with worsening disease and death. These pathways lead to initiation and enhancement of TG via the contact system and the tissue factor pathways [40]. Therefore, interestingly, the correlation we found between ETP and the SIC score suggests that TG may assist in identifying those patients at risk for developing COVID-19-associated coagulopathy.

The major strength of our study is the prospective enrollment of representative COVID-19 patients managed at internal medicine departments, and the availability of TG analysis, extensive clinical data, and disease severity scores for all patients, as well as repeated TG samples in some patients, which allowed for the first time, the assessment of the correlation between TG and disease severity scores in COVID-19 patients. Among study limitations, it is the small size of our COVID-19 cohort since we only enrolled consecutive patients who were able to provide signed informed consent and for whom proper venous sampling was possible. For this reason, we could not enroll COVID-19 patients admitted to ICUs. The small sample size and diversity of the study population (e.g., COVID-19 patients with different disease severity, vaccination status, different comorbidities including malignancy in 43.7% of the patients, and 53.1% of the patients had cardiovascular diseases) may have impacted study results. Indeed, obtaining real-world prospective data and special global assay tests from a variability of hospitalized COVID-19 patients (whose comorbidities reflect the hospitalized population at the time) has been extremely challenging. This study may therefore be underpowered to assess the association between TG parameters and clinical outcomes. Second, we do not have data on TG before the initiation of anticoagulant therapy, which could allow us to determine the net effect of anticoagulant therapy on TG. Despite these limitations, our results are the first to correlate TG with the SIC score relevant for COVID-19 patients.

In conclusion, we confirmed that TG parameters in hospitalized COVID-19 patients resemble values obtained from healthy controls despite anticoagulant therapy, indicating a hypercoagulable condition; however, TG was not associated with disease severity according to the SOFA score. A correlation between ETP and the SIC score was noted and deserves attention. Further research is required to validate these findings that may potentially be incorporated into future clinical practice.

This study protocol, entitled “Potential associations of global coagulation assays with clinical disease among hospitalized COVID-19 patients, a thrombin generation guided study” was reviewed and approved by our Institutional IRB at Sheba Medical Center, Tel Hashomer, Israel, approval number 7334-20-SMC. In this study, plasma samples of hospitalized COVID-19 patients were sequentially collected (along with other routine tests) and analyzed for TG. Demographic and clinical data were retrieved from medical records. Written informed consent form was obtained from all donors.

Sarina Levy-Mendelovich received a grant and research support from Pfizer and Novo Nordisk, and Gili Kenet received a grant and research support from Alnylam, Bayer, BPL, Opko Biologics, Pfizer, Shire, and honoraria for consultancy/lectures from Alnylam, Bayer, CSL, Opko Biologics, Pfizer, Takeda, and ROCHE. Other authors have no relevant conflict of interests.

This study was supported by an IIS grant from Pfizer.

Omri Cohen and Gili Kenet wrote the manuscript. Galia Rahav performed critical review. Nitsan Landau, Uri Manor, Eyal Meltzer, and Gad Segal recruited patients, obtained informed consent, and collected samples. Orly Efros collected and gathered data. Einat Avishai, Tami B. Barazani, Tami Livnat, Keren Asraf, Ram Doolman, and Sarina Levy-Mendelovich performed experimental analysis. Ivan Budnik performed the statistical analysis. All the authors read and approved the final manuscript.

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