Viscoelastometric Testing to Assess Hemostasis of COVID-19: A Systematic Review

Infection by SARS-CoV-2 is associated with a high risk of thrombosis. The laboratory documentation of hypercoagulability and impaired fibrinolysis remains a challenge. Our aim was to assess the potential usefulness of viscoelastometric testing (VET) to predict thrombotic events in COVID-19 patients according to the literature. We also (i) analyzed the impact of anticoagulation and the methods used to neutralize heparin, (ii) analyzed whether maximal clot mechanical strength brings more information than Clauss fibrinogen, and (iii) critically scrutinized the diagnosis of hypofibrinolysis. We performed a systematic search in PubMed and Scopus databases until 31st December 2020. VET methods and parameters, and patients’ features and outcomes were extracted. VET was performed for 1063 patients (893 intensive care unit (ICU) and 170 non-ICU, 44 studies). There was extensive heterogeneity concerning study design, VET device used (ROTEM, TEG, Quantra and ClotPro) and reagents (with non-systematic use of heparin neutralization), timing of assay, and definition of hypercoagulable state. Notably, only 4 out of 25 studies using ROTEM reported data with heparinase (HEPTEM). The common findings were increased clot mechanical strength mainly due to excessive fibrinogen component and impaired to absent fibrinolysis, more conspicuous in the presence of an added plasminogen activator. Only 4 studies out of the 16 that addressed the point found an association of VETs with thrombotic events. So-called functional fibrinogen assessed by VETs showed a variable correlation with Clauss fibrinogen. Abnormal VET pattern, often evidenced despite standard prophylactic anticoagulation, tended to normalize after increased dosing. VET studies reported heterogeneity, and small sample sizes do not support an association between the poorly defined prothrombotic phenotype of COVID-19 and thrombotic events.

Abstract: Infection by SARS-CoV-2 is associated with a high risk of thrombosis. The laboratory documentation of hypercoagulability and impaired fibrinolysis remains a challenge. Our aim was to assess the potential usefulness of viscoelastometric testing (VET) to predict thrombotic events in COVID-19 patients according to the literature. We also (i) analyzed the impact of anticoagulation and the methods used to neutralize heparin, (ii) analyzed whether maximal clot mechanical strength brings more information than Clauss fibrinogen, and (iii) critically scrutinized the diagnosis of hypofibrinolysis. We performed a systematic search in PubMed and Scopus databases until 31 December 2020. VET methods and parameters, and patients' features and outcomes were extracted. VET was performed for 1063 patients (893 intensive care unit (ICU) and 170 non-ICU, 44 studies). There was extensive heterogeneity concerning study design, VET device used (ROTEM, TEG, Quantra and ClotPro) and reagents (with non-systematic use of heparin neutralization), timing of assay, and definition of hypercoagulable state. Notably, only 4 out of 25 studies using ROTEM reported data with heparinase (HEPTEM). The common findings were increased clot mechanical strength mainly due to excessive fibrinogen component and impaired to absent fibrinolysis, more conspicuous in the presence of an added plasminogen activator. Only 4 studies out of the 16 that addressed the point found an association of VETs with thrombotic events. So-called functional fibrinogen assessed by VETs showed a variable correlation with Clauss fibrinogen. Abnormal VET pattern, often evidenced despite standard prophylactic anticoagulation, tended to normalize after increased dosing. VET studies reported heterogeneity, and small sample sizes do not support an association between the poorly defined prothrombotic phenotype of COVID-19 and thrombotic events.

Introduction
In contrast to conventional clotting tests, viscoelastic tests (VETs) monitor changes of viscoelastic properties of a forming and evolving clot from whole blood, before and beyond the clotting point; they are often referred to as a global hemostasis test, although some aspects of hemostasis are not explored [1][2][3]. Coagulation occurs in the presence of platelets and red blood cells, and fibrinolysis can translate into a decrease in clot mechanical strength after its maximum has been reached, but clot retraction seems to play a role here as well [4][5][6]. VETs are based on the mechanical properties of the clot, like mechanical strength, and are influenced by its composition in platelets, fibrin, red blood cells, and factor XIII [7][8][9]. To our knowledge, the assessment of factor XIII by VETs has not been investigated in COVID-19 patients.
VETs have been considered to provide a comprehensive assessment of the dynamic process of blood clot formation and subsequent lysis. As they can be performed bedside as point-of-care testing and can give useable results about clot formation and a potential hyperfibrinolysis within one hour, they are chiefly considered as convenient tools for realtime assessment of coagulation and fibrinolysis in whole blood and have been gaining in popularity in various hemorrhagic situations, such as cardiac surgery, obstetrics, and traumatology over decades, for the management of acutely bleeding patients [3]. By contrast, COVID-19 disturbance of hemostasis is likely a combination of hypercoagulability and impaired fibrinolysis (a prothrombotic laboratory phenotype), at least in part, contributing to the thrombotic risk and the prothrombotic laboratory phenotype, but VETs have been nevertheless suggested to be potentially useful, in line with previous works on sepsis [10] and trauma [11,12], for example.
Of note, VETs share the same limitations as all currently available clinical lab tests, i.e., negligible effect of endogenous anticoagulants, absence of endothelium, and very low shear in a close system. Furthermore, there are good reasons to challenge the interpretation of hypercoagulability and to question the ability to sensitively detect and accurately quantify hypofibrinolysis, especially when a value equal to zero belongs to the manufacturer's reference range.
Our aim was to assess the potential clinical usefulness of VETs to predict clinical outcomes (mainly thrombotic events) in COVID-19 patients through this systematic review. We also (i) analyzed the impact of anticoagulation and the methods used to neutralize heparin (in other words, was heparin duly neutralized?), (ii) disentangled reported alterations in clotting dynamics and analyzed whether maximal clot mechanical strength brings more information than Clauss fibrinogen, and (iii) critically scrutinized the documentation of hypofibrinolysis with VET under various reactive conditions. The term 'hypercoagulable state' will be uniformly used to refer to the investigators' interpretation of VET findings; we will discuss to what extent this is an appropriate interpretation.
The preanalytical aspects, which are crucial in laboratory hemostasis but scarcely mentioned among the retrieved studies, are beyond the scope of this review and will not be addressed.

Search Methodology
We performed a systematic literature search in PubMed and Scopus databases, regardless of publication status, using the following keywords 'viscoelastic test OR thromboelastometry OR thromboelastography OR sonorheometry OR ROTEM OR TEG OR Quantra OR ClotPro' AND 'coronavirus disease 2019 OR COVID-19 OR severe acute respiratory syndrome coronavirus 2 OR SARS-CoV-2'. Search strategy is provided as Data S1. We also searched the reference lists of selected articles for additional relevant works, and we did not restrict our search to articles published in English and found some articles in Russian and Hungarian. In addition, reviewers performed manual searches and crossreferences in the retrieved papers. The last search was conducted on 31 December 2020. Our review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) guidelines [13], and the PRISMA summary table can be found as Data S2. Due to a considerable heterogeneity among the retrieved studies, we did not extend our systematic review to a meta-analysis.

Study Selection
All references retrieved from our search were screened based upon their title and abstract to assess eligibility. If they were considered relevant, the full-text articles were analyzed to check if they met the selection criteria as follows. As COVID-19 pandemic is a recent phenomenon, and due to the relatively small number of published data on VETs, we did not restrict eligibility according to patients' characteristics, disease severity, or treatment modalities. Studies of any design and case reports, including original data from VETs in COVID-19 patients with neither pregnancy nor known history of coagulation disorder, were deemed eligible. All relevant studies regardless of methodological quality were included when the full-text article was available (Table 1). Reviews, position articles, and guidelines were excluded. All kind of VETs were included but were analyzed separately.

Data Extraction
For each study, data regarding author identification, geographic location, study design, number of patients and their characteristics (including comorbidities and thrombotic events), prospective design or not, timing of blood collection and anticoagulation status, type of VET device used and results, and the results of other conventional hemostasis tests (platelet count, fibrinogen and D-dimers plasma levels), and C-reactive protein were extracted with the aid of a systematic chart.

A Concise Overview of the Different VET Devices
Viscoelastometric testing (VET) should be performed either immediately with native whole blood or within four hours after drawing if performed with whole citrated blood, as most often done [1,3].
ROTEM devices and TEG5000 all rely on the movement of a pin and a cup relative to each other; in the former, the cuvette is fixed, and the pin oscillates, and vice versa in the latter. The oscillations are recorded and graphically displayed with the characteristic normal tuning fork shape [3]. The conventional clotting point roughly corresponds to the reaction time R for TEG, and to the clotting time CT for ROTEM, ClotPro, and Quantra; extended fibrin polymerization is monitored with the kinetics time K and α angle for TEG and with CFT and α angle for ROTEM and ClotPro; the eventual result is maximal mechanical strength (maximal amplitude MA for TEG, maximal clot firmness MCF for ROTEM and ClotPro and clot stiffness CS for Quantra) and its subsequent decrease, as a result of 'endogenous' fibrinolysis monitored by lysis of the clot at given time x LY(x) for TEG and by maximal lysis ML or lysis of the clot at a given time x (LI(x)) for ROTEM and ClotPro, at least in part [2,3,14].
Coagulation can be initiated through the contact phase or the tissue factor pathway (often referred to as intrinsic or extrinsic pathways, respectively) and needs recalcification when citrated blood is used [3]. If the nature of the initiating agents is known, their concentrations are not disclosed. Regarding the former pathway, the limitations of aPTT testing apply, although 'clotting times' are longer, suggesting a lower amount of contact phase activator (kaolin, celite, or ellagic acid) and higher calcium concentration. The different well-known behaviors of those reagents in case of defective contact phase, abnormal factor VIII levels, high CRP (C-reactive protein) levels, lupus anticoagulant, or heparin must be borne in mind. Two reagents can be used to neutralize heparin, either polybrene (hexadimethrine bromide) or heparinase; two to inhibit the platelet contribution to mechanical clot properties, namely cytochalasin D and abciximab, sometimes both together; lastly, two to inhibit fibrinolysis, either aprotinin or tranexamic acid [3]. To what extent those inhibitions are fully achieved is not entirely clear.

ROTEM
Three versions of the ROTEM device exist: from the oldest to the most recent, ROTEMgamma, ROTEM-delta, and the brand-new version ROTEM-sigma. The main difference between them is that ROTEM-gamma and -delta need manual pipetting of the blood sample and the reagents into cups, whereas ROTEM-sigma is a completely automated, closed system. For the latter, reagents consist of a consumable ready-to-use cartridge with four parallel channels prefilled with specific lyophilized reagents [15]. All ROTEM versions can perform the same assays, namely INTEM, HEPTEM, EXTEM, FIBTEM, and APTEM, to investigate the intrinsic pathway (with and without heparinase), the extrinsic pathway, the fibrinogen component, and the fibrinolysis with aprotinin, respectively. Of note, EXTEM, FIBTEM, and APTEM reagents contain polybrene and HEPTEM contain heparinase to neutralize heparin (Table A1) [2,16]. They report the same parameters: clotting time (CT), clot formation time (CFT), α angle, "amplitude of the clot" at a given time x (A(x)), maximum clot firmness (MCF), clot lysis index (LI(x)), and maximum lysis (ML) (Table A2).

TEG
Briefly, regarding TEG5000 a blood sample is pipetted into a cup; liquid reagents are added; ultimately, a fixed pin connected to a detector system is then put in the cup. The graphical representation is called TEMogram. TEG6s for its part is a completely closed and automated system. In contrast to its predecessor TEG5000, it relies on sonorheometry.
Reagents consist of a consumable, ready-to-use cartridge with four parallel channels prefilled with specific lyophilized reagents (Table A3) [17,18].
The two versions of the TEG device can perform the same assays, namely Kaolin TEG with (CKH) or without heparinase (CK), RapidTEG (CRT), and TEG Functional Fibrinogen (CFF), and offer the same parameters: reaction time (R), kinetics time (K), α angle, maximum amplitude (MA), and fibrinolytic activity (Ly) [3,19]. Of note, heparin neutralization differs between TEG500, where neutralization can occur in virtually any channel by using heparinase-coated cups, and TEG6s, where neutralization occurs only in the CKH channel thanks to heparinase (Table A4) [3].

Quantra
The Quantra device also uses sonorheometry. Briefly, an acoustic radiation force is applied to the blood sample. As the blood clot forms, it starts to resonate: oscillations are then correlated with the shear modulus of the blood sample. The resistance of the sample to shear forces can be quantified by the time delay between the ultrasound pulse emission and the returning echoes [20][21][22].
Reagents consist of a consumable, ready-to-use cartridge with four parallel channels prefilled with specific lyophilized reagents [22]. There are currently two kinds of cartridges: the QPlus cartridge and the QStat one dedicated to exploring fibrinolysis [23]. Measurements of clot coagulation time with (CTH) or without (CT) heparinase and coagulation initiation with kaolin, clot stiffness (CS) after initiation with thromboplastin, and fibrinogen contribution to the overall clot stiffness (FCS) after platelet inhibition with abciximab are performed simultaneously in four parallel channels. Of note, channel 2 contains heparinase, and channels 3 and 4 contain polybrene to neutralize heparin. Platelet contribution to clot stiffness (PCS) results from the difference between total CS and FCS (Tables A5 and A6).

ClotPro
The ClotPro device uses rotational technology similar to ROTEM®(Werfen, Barcelona, Spain), but some differences exist between the two devices. First, in contrast with ROTEM, the cuvette rotates and the pin is stationary [24,25]. Second, reagents for each assay are present in dry form in a sponge located in the pipette tip; during pipetting of the patient sample, the reagent is automatically added to the blood [25]. This device can perform the same kind of assays as the ROTEM device (EX-test, IN-test, HI-test, FIB-test, AP-test) plus some other specific ones (RVV-test, ECA-test), and offer similar parameters. Of note, EXtest, tPA-test, and FIB-test contain polybrene to neutralize heparin (Tables A7 and A8) [24].

Literature Search
Our literature search and selection flow chart according to PRISMA statement [13] is summarized in Figure 1. We identified 140 references, resulting in 97 unique citations after duplicates removal. Two additional articles were identified through other sources. Each title and abstract were screened, and 36 references were excluded either because they were not related to the subject (n = 16), because they were position articles or guidelines (n = 7) or reviews (n = 5), or because there was no full-text available at this time (n = 5) or no possible translation (n = 5). A total of 63 potentially eligible articles were considered for inclusion, and the full-text articles were retrieved. The most common reasons for exclusion after the full-text evaluation were that papers were reviews (n = 19), not related to the subject (n = 6), or position articles or guidelines (n = 4). Finally, 44 references [24, met the eligibility criteria.

Originality of Our Systematic Review as Compared to the Existing Ones on the Subject
Reviews have already been published recently, two of them only being systematic [69][70][71][72], but none has so far investigated the four major commercially available VET devices (i.e., ROTEM, TEG, ClotPro, and Quantra) or included such a large number of studies (n = 44). Characteristics of each review are summarized in Table 2.
Overall, case reports were excluded (except for one systematic review [71]); few studies were available and presented extensive heterogeneity. We identified 140 references, resulting in 97 unique citations after duplicates removal. Two additional articles were identified through other sources. Each title and abstract were screened, and 36 references were excluded either because they were not related to the subject (n = 16), because they were position articles or guidelines (n = 7) or reviews (n = 5), or because there was no full-text available at this time (n = 5) or no possible translation (n = 5). A total of 63 potentially eligible articles were considered for inclusion, and the full-text articles were retrieved. The most common reasons for exclusion after the full-text evaluation were that papers were reviews (n = 19), not related to the subject (n = 6), or position articles or guidelines (n = 4). Finally, 44 references [24, met the eligibility criteria.

Originality of Our Systematic Review as Compared to the Existing Ones on the Subject
Reviews have already been published recently, two of them only being systematic [69][70][71][72], but none has so far investigated the four major commercially available VET devices (i.e., ROTEM, TEG, ClotPro, and Quantra) or included such a large number of studies (n = 44). Characteristics of each review are summarized in Table 2. Overall, case reports were excluded (except for one systematic review [71]); few studies were available and presented extensive heterogeneity.

Characteristics of the Selected Studies
Quality assessment of the selected study was performed using the Scottish Intercollegiate Guidelines Network (SIGN) grading system [73]. Overall, the retrieved studies were of low (3, "non analytic studies") to moderate quality (2+, "well-conducted case control or cohort studies with a low risk of confounding or bias and a moderate probability that the relationship is causal"), and details can be found as Data S3. Characteristics of the selected studies are summarized in Table 3.         A total of 1538 inpatients were studied, of which 1393 were COVID-19-positive, among whom 1189 were ICU patients. At least one VET was performed during the hospital stay of 1208 patients, of whom 1063 were COVID-19 patients hospitalized either in an ICU (893 patients) or in a medical ward (IMW, 170 patients). The remaining 145 patients were sex-and age-matched non-COVID-19 controls hospitalized either in the ICU (89 patients) or in IMW (56 patients) for ARDS (acute respiratory distress syndrome) or pneumonia non-related to SARS-CoV-2, or for postoperative care. One article [33] reported data about eight hospitalized children either in a pediatric ward or in a pediatric ICU (PICU).

Characteristics of the Included Patients
Characteristics of the included patients are shown in Table 4.      Values in italics and in brackets are the reference values; we have indicated our reference ranges * for information purposes. Comorbidities: 1 Overweight and obesity, associated with high blood pressure, diabetes and cardiovascular risk factors; 2 High blood pressure, diabetes and some additional comorbidities; 3 Overweight and obesity, with some additional comorbidities; 4 Overweight and obesity; 5 Overweight and obesity, associated with high blood pressure, diabetes, pulmonary disease and cardiovascular risk factors; 6 Overweight and obesity, associated with high blood pressure; 7 Overweight and obesity, associated with cardiovascular risk factors, pulmonary disease and kidney disease; 8 Overweight and obesity, associated with diabetes, cardiovascular risk factors, pulmonary disease and kidney disease; 9 Overweight and obesity, associated with high blood pressure, diabetes, kidney disease and cardiovascular risk factors; 10 Overweight and obesity, associated with high blood pressure, diabetes, pulmonary disease, kidney disease and cardiovascular risk factors; 11 Overweight and obesity, associated with high blood pressure and diabetes. Abbreviations: ICU: Intensive care unit (adults); IMW: Internal medicine ward; PICU: Pediatric intensive care unit; PW: Pediatric ward; IMV: Invasive mechanical ventilation; ECMO: Extracorporeal membrane oxygenation; RRT: Renal replacement therapy; M: Male; F: Female; SOFA score: Sequential organ failure assessment score; APACHE score: Acute physiology and chronic health evaluation score; SAPS score: Simplified acute physiology score; DIC score: Disseminated intravascular coagulation score; SIC score: Sepsis-induced coagulopathy score; BMI: Body mass index; CRP: C-reactive protein; NP: Not provided; TEG: Thromboelastography; ROTEM: Rotational thromboelastometry; TEM: Thromboelastometry; tPA: tissue plasminogen activator.
As a general rule, three assays were performed, mostly INTEM (19 studies), EXTEM (23 studies), and FIBTEM (23 studies). The great majority of the articles reported results from EXTEM assay with or without INTEM assay and associated with FIBTEM assay. Only four articles [26,41,44,50] reported data from HEPTEM assay (Table 6), while almost all patients received anticoagulation by UFH or LMWH at least at a prophylactic dose. The APTEM assay results were only reported by one case report [26] and were consistent with the absence of hyperfibrinolysis. Two studies reported data from TEM-tPA (Table 7), an investigator-modified assay derived from EXTEM assay to investigate a potential hypofibrinolysis [39,40].
Among the five studies reporting data from COVID-19 adult patients both in an ICU and an IMW [28,32,39,41,48], one showed a similar hypercoagulable profile (increased maximum clot firmness) for both groups (p > 0.05 [48]), whereas the four others showed a similar hypercoagulable pattern (increased "amplitude of the clot" or maximum clot firmness with or without a decreased CFT) for both groups compared with healthy controls or manufacturer's reference range; however, the hypercoagulable pattern was more pronounced for ICU patients (p < 0.05 between both groups).
Results reported in children [33] showed a hypercoagulable pattern comparable to adults with an increased in MCF in INTEM, EXTEM, and FIBTEM assays and a slightly decreased CFT.
In summary, four points are worthy of consideration. First, the hypercoagulable profile defined just above was observed early in the clinical course of the disease [41,50]. Second, it was observed in both ICU and non-ICU COVID-19 patients [28,32,39,41,48]. Third, it persisted over time from admission up to 10 to 14 days later [26,27,29,45]. Fourth, it was observed even in the absence of heparin neutralization and despite higher therapeutic intensity anticoagulation administration [26,47]. Of note, only six studies [28,[33][34][35]44,49] gathering 195 patients examined the potential association with thrombotic events occurring, and only two [34,44] reported that patients with thromboembolic complications exhibited low or even absent fibrinolysis.

TEG
A total of 403 patients, of whom 402 were COVID-19 ICU patients, had at least one VET performed with TEG. Most of them were intubated and mechanically ventilated. They almost all received anticoagulation by UFH or LMWH, at least at prophylactic dose.
TEG Functional Fibrinogen (CFF) was assessed in four studies [59,61,62,65], showing an increase in maximum clot amplitude with a median CFF-MA ranging from 41 to 56 mm for all patients as compared with manufacturer's reference range (15 to 32 mm), and with a negative skewness coefficient of −0.37 [59].
Hypercoagulable state assessed by VET with an excessive fibrinogen component to clot strength 1 p < 0.001 compared with healthy subjects, no difference between value at admission and at discharge (p > 0.05); 2 p < 0.001 compared with healthy subjects, higher MA at discharge (p < 0.05); 3 No difference between the two groups (p > 0.05); 4 p = 0.001 compared with patients with D-dimers levels ≤ 2000); 5  In summary, three points are worthy of attention. First, the hypercoagulable pattern defined as just above was observed in both ICU and non-ICU COVID-19 patients [65]. Second, it was observed despite UFH or LMWH anticoagulation at prophylactic doses or higher, and an anti-Xa activity within the target range without heparin neutralization (TEG, CK assay) [49]; and third it persisted over time [53,[59][60][61]. However, the association between the observed pattern and the occurrence of thrombotic events remains a matter of debate. One study including 21 patients reported that an increase in MA provides 100% sensitivity and 100% negative predictive value to discriminate between patients with high or low rate of thrombotic events (44), but only a few patients were reported. Another study (44 patients) reported that no evidence of clot lysis at 30 min (LY30) associated with high D-dimers levels (>2600 µg/L) could predict thromboembolic events (p = 0.008) and need for hemodialysis in critically ill patients (p = 0.004) with COVID-19 [51]. However, seven other studies [53,55,[58][59][60]63,64] comprising 243 patients did not find an association between the VET parameters and the occurrence of thrombotic events.

Quantra
The Quantra device was used in two studies [66,67], both prospective, one of them comparing data from ICU non-COVID-19 patients with ICU COVID-19 patients [66]. The two studies included 44 ARDS intubated and mechanically ventilated ICU patients, of whom 33 were COVID-19 positive. All patients received anticoagulation according to local protocols or guidelines [74]. Tests were performed using the QPlus Cartridge, which contains heparinase in the CTH channel and polybrene in the CS and FCS channels to neutralize heparin. Results are summarized in Table 9.
Both studies suggested a hypercoagulable pattern associated with preserved thrombin generation, assessed by prothrombin fragments 1 + 2 and thrombin-antithrombin complex levels [66] and despite UFH or LMWH anticoagulation at a minimum of prophylactic dosing. However, the VET hypercoagulable pattern tended to normalize [67] after a 50% increase in thromboprophylaxis dosing and based on the body weight. However, the association between the documented hypercoagulable pattern and thrombotic event occurrence was not studied.

ClotPro
ClotPro was used in a retrospective study in Austria [24] and in three cases in Hungary [68] of ICU patients who received anticoagulation at prophylactic doses or greater. VET assays were performed using four reagents and channels (Table A7), namely EX-test, IN-test, Fib-test, and tPA-test. Results are summarized in Table 10.
Results from the tPA-test showed a hypercoagulable pattern (increased maximum clot firmness) associated with impaired fibrinolysis; the latter was assessed either by a decreased lysis capacity of the clot in presence of tPA as compared with manufacturer's reference values, followed few days later by a normalization with still marked elevated D-dimers levels [68], or by an increased clot lysis time as compared with healthy controls with a p-value < 0.01 [24]. However, these findings do not appear to be associated with the occurrence of thrombotic events [24].   Retrospective study

ICU
Reference range established in healthy adults

Discussion
Although all of the studies share the common viscoelastometric testing concept in evaluating COVID 19 patients' hemostasis, the differences in the testing systems and reagents, resultant data and implications, and variability of the patients' severity of illness make interpretation difficult. The association with thrombotic events is not very well established, and might largely depend on the actual VET used. We will more specifically discuss whether VETs provide clinically relevant information about fibrinogen in a COVID-19 patient, and we will discuss its use regarding potential anticoagulation with heparins.

Methodological Issues in VET Studies
There are numerous methodological differences among the 44 studies using VETs to assess the hemostasis in COVID-19 patients we have retrieved and analyzed, explaining why results were not consistent through studies, or sometimes even conflicting. This was already raised by previously published reviews [70,71].
Second, there was also heterogeneity among the patients' characteristics concerning age, severity of the disease, gender distribution, and comorbidities. In addition to the lack of power to evidence a statistically significant association between the VET patterns and thrombotic events, this heterogeneity could explain the differences between the studies' results, at least in part.
One important consideration is different monitoring devices were used. Even though they share the same objective of viscoelastic clot properties evaluation, they present substantial differences from technological and methodological viewpoints. First, they rely on different technologies to monitor clot formation, clot strength, and clot lysis (i.e., thromboelastometry, thromboelastography, and sonorheometry). Second, there are some differences in the way the tests are carried out and the sample and reagents are delivered to perform the assay, specifically the activators. While in the most recent versions of the instruments (TEG6s, ROTEM sigma, Quantra, ClotPro) the reagents are already included in reaction cartridges or in tips and require only the addition of the blood sample, the previous versions (ROTEM gamma and delta, TEG5000) required manual or semi-automated pipetting of reagents and samples, resulting in very high inter-and intra-operator coefficients of variation for some parameters [77,78]. Third, the composition of the reagents differs from one manufacturer to another, especially for the assay aiming to assess the fibrinogen component of clot strength, also called functional fibrinogen. Briefly, clot strength is mainly due to the interaction between fibrin network (containing activated factor XIII), platelets [7], neutrophil extracellular traps [79], and red blood cells [80,81]. Platelets are an important contributor to the clot strength, and the MA (TEG), MCF (ROTEM and ClotPro), and CS (Quantra) parameters reflect both platelet count and function [9,82], as well as fibrin contribution. To assess functional fibrinogen, platelet contribution must be inhibited, and two different approaches are used that include abciximab (GpIIb-IIIa inhibitor, TEG and Quantra), cytochalasin D (cytoskeleton inhibitor, ROTEM), or a combination of both (ClotPro). Some studies [83][84][85] compared the fibrinogen contribution to clot mechanical strength measured with VET using either a GpIIb-IIIa inhibitor or cytochalasin D and showed that the latter was more efficient: functional fibrinogen assessment with TEG or Quantra could lead to an overestimation of fibrinogen levels as compared with ROTEM. However, cytochalasin D alone may not completely remove the platelet contribution, especially with a high platelet count, and a combination of a GpIIb-IIIa inhibitor and cytochalasin D seems to provide more accurate results [83][84][85].
Therefore, we can reasonably doubt that the results obtained from the different devices and studies are interchangeable, as previously noted in non-COVID-19 patients [3,86]. There may even be differences in results between different versions of the same device (as between ROTEM-delta and ROTEM-sigma for example [15,87], or between TEG5000 and TEGS6s [17][18][19]), but overall the devices show good correlations for the main parameters evaluated [14,86,88,89]. To our knowledge, there have not been previous comparisons in COVID-19 patients. Studies on VET have always been plagued by those issues, unfortunately still unresolved.

Definition of a Hypercoagulable State by VET and Association with Thrombotic Events
The conventional clotting time corresponds to the 'reaction time' R for TEG, and the 'clotting time' CT for ROTEM, ClotPro, and Quantra. Extended fibrin polymerization is described as the kinetics time K and α angle for TEG and CFT and α angle for ROTEM and ClotPro. The clot strength is defined as maximal mechanical strength (maximal amplitude MA for TEG, maximal clot firmness MCF for ROTEM and ClotPro, and clot stiffness CS for Quantra).
Beyond a purely biological definition, for which there is no consensus or appropriate term, sometimes 'procoagulant' or hypercoagulable, what matters is the association with the patient's thrombotic risk. Outside the COVID-19 setting, two systematic reviews and a subsequent meta-analysis involving 1285 patients with solid tumors or hematopoietic malignancies [90] or 8944 surgical patients [91] showed that the occurrence of thrombotic events was associated with features consistent with hypercoagulability: acceleration of fibrin polymerization (increase in α angle in both ROTEM and TEG, shortened CFT in ROTEM and shortened K time in TEG) and increased clot mechanical strength (increase in MCF for ROTEM and in MA for TEG). However, another meta-analysis of 1081 patients in a variety of clinical settings [92] showed that ROTEM and TEG had a moderate ability to discriminate between patients who developed a thrombotic event and those who did not, with a diagnostic odds-ratio of 3.6, a low sensitivity (56%) but a somewhat better specificity (76%). It is noteworthy that the performance in the prediction of thrombotic events depends both on the type of device (with a better performance for ROTEM with a diagnostic odds-ratio of 6.3 against 3.2 for TEG), and on the type of thrombotic event (with a sensitivity of 67%, a specificity of 72%, and a diagnostic odds-ratio of 6.4 for arterial thrombotic events, contrasting with a sensitivity of 41%, a specificity of 70%, and an oddsratio diagnosis of 3.1 for venous thrombotic events). Why VET findings should be more associated with arterial thrombotic events than with venous ones is obscure, though.

Ability of VETs to Detect Hypofibrinolysis State and Association with Thrombotic Events
Fibrinolysis is monitored at a specific time x minutes after MA was reached for TEG (LY(x) parameter), and by maximal lysis ML (reduction in clot firmness after MCF in relation to MCF) or lysis of the clot at a given time x minutes after CT was reached (LI(x)) parameter for ROTEM and ClotPro. The diminution in clot maximum amplitude was thought to be due to both fibrinolysis and potentially platelet-mediated clot retraction [4][5][6]93]. However, as no change in clot mechanical strength after the maximum was reached was reported in many studies in COVID-19 patients, platelet-mediated clot retraction does not seem to play a significant role here.
Usually VETs are used to detect major hyperfibrinolytic states [94] that occur in the most severe, advanced stages of hemostasis derangements in clinical settings such as trauma and perioperative hemorrhage. However, could they be used to assess hypofibrinolysis? VETs have shown potential usefulness in sepsis-induced coagulopathy [10] and trauma-induced coagulopathy [11] to detect low levels of fibrinolysis and to identify patients for whom the administration of tranexamic acid should be avoided. Endogenous systemic fibrinolysis is usually weak because of low or even no circulating levels of free plasminogen activators, which are fully complexed to PAI-1 and thus inactive. Normal lysis of a whole blood clot is therefore a slow phenomenon [95], and its visualization on a VET trace recorded during one hour or two seems unlikely. Furthermore, as the zero value belongs to the manufacturer's reference range, speaking about a reduced or an absent fibrinolysis seems awkward if there is no control group for comparison. Among the 25 studies reporting a reduced or absent fibrinolysis, only 8 [24,32,39,40,44,48,52,53] made this assessment by comparison with a control group.
Several ROTEM and TEG modifications have been reported adding urokinase plasminogen activator (uPA) or tissue plasminogen activator (tPA) to demonstrate hypofibrinolysis. A brief literature search revealed multiple protocols for modified ROTEM and TEG including addition of a plasminogen activator. Although they show evidence of hypofibrinolysis in different clinical settings, they all share the same methodological issues and limitations. First, there is a lack of standardization concerning tissue factor concentrations, as low levels added to the sample produce non-reproducible results and often a weak clot [96,97]. Second, there is also a lack of standardization in tPA concentrations studied ranging from 50 to 625 ng/mL [96][97][98], and similar results for modified VET with uPA [99,100], with an 'optimal concentration of uPA' differing from subject to subject, and a wide interindividual variation in lysis parameters [99].
Among the 44 studies analyzed in this review, only four investigated the effect of adding tPA to standard VET. Two used the ClotPro device [24,68] and its ready-to-use tPA-test reagents, which are now CE-marked, whereas the two others [39,40] used an in-house ROTEM assay with two different tPA concentrations (named TEM-tPA), making a comparison between them problematic. No study has investigated a defective fibrinolysis using the Quantra device into the COVID-19 context, whereas a new dedicated reagent cartridge is now available [23]. The four articles share the same conclusion that increased clot maximum amplitude and decreased lysis index reflect an increase in clot strength and a decreased fibrinolytic capacity, results that need to be confirmed with a larger cohort. Further, the TEM-tPA assay needs to be standardized and validated [101], although it seemed to show good intra-and inter-assay precision in healthy controls [39].
Association between impaired fibrinolysis assessed with VETs and clinical outcomes is a matter of debate. Some studies failed to find an association [24,53,55,58,59], while others suggested that impaired fibrinolysis was associated with a higher rate [34] and a shorter time to the occurrence of thrombotic events [51], and together with D-dimer levels it could predict thrombotic events [44,51] and the need for hemodialysis in critically ill patients with COVID-19 [51].

Correlation between Clauss Fibrinogen and Functional Fibrinogen Assessed by VETs
Outside of the COVID-19 context, the ROTEM FIBTEM is the most studied point of care fibrinogen level assay with numerous studies in trauma, cardiac surgery, liver transplantation, and obstetrics. The correlation between Clauss fibrinogen and FIBTEM-MCF or fibrinogen-related TEG parameters was reported as variable with R 2 values ranging from 0.44 to 0.94 for ROTEM [102] and from 0 [103] to at least 0.80 [102] for TEG. The clinical experience with the Quantra device is limited, with only few published studies to date, but the correlation between Clauss fibrinogen and FCS ranged from moderate to very good, with R 2 values ranging from 0.55 to 0.88, with a huge variability between studies [14,22,104,105]. To our knowledge, data concerning correlation between Clauss fibrinogen and clot amplitude and maximum clot firmness provided by the ClotPro FIB-test assay is not yet available.
Among the 44 studies dealing with VETs and COVID-19 patients, only a few ones [33,41,53,58] investigated the correlation between Clauss fibrinogen and 'functional fibrinogen' assessed by VETs. For ROTEM®(FIBTEM), one report noted a good correlation (Pearson's correlation coefficient r = 0.84) [41], while another showed no correlation (p = 0.130) in children [33]. Two studies explored TEG in this regard and reported a moderate to good association (Pearson's correlation coefficient r = 0.453 [58] and 0.74 [53]). These limited results due to low COVID-19 patient numbers and different assays suggest more studies are required.
Further, whether VET characteristics are unique to hyperfibrinogenemia alone is an important question, as almost all COVID-19 patients also present with hyperfibrinogenemia. Patients with hyperfibrinogenemia may exhibit a 'functional fibrinogen' (VET parameter) in the reference range [35,37,66], while other reports of fibrinogen levels within reference ranges exhibit increased functional fibrinogen [26,28,48].
The authors of a previously published review [69] highlighted the potential usefulness of VET in accurately assessing plasma fibrinogen levels in COVID-19 patients receiving direct thrombin inhibitors (DTI) through the assessment of the clot amplitude of the functional fibrinogen assay. Indeed, evaluation of fibrinogen levels by the Clauss method could lead to an underestimation due to the inhibition of the thrombin included in the reagent by the DTI [106,107], ranging from 23 to 96% according to the reagent used [107].

Impact of Differences in Anticoagulation Regimens (Type (UFH, LMWH) and Dosage)
Most currently studied COVID-19 patients receive heparin (LMWH or UFH), either with prophylactic or therapeutic regimens according to local protocol or guidelines [74][75][76]108]. Some studies specifically noted when blood samples for VETs were drawn in heparinized patients, but the timing of administration was often missing as well as anti-Xa levels. While for TEG, heparinase reagents were frequently used (14 studies of 15 [51][52][53][54][55][56][57][58][59][60][62][63][64][65]), this was not the case for ROTEM: only a few ones (4 of 25) generated data with HEPTEM assay, together with INTEM assay. Among these four latter studies, only one [26] reported different results from the two assays, whereas the three others showed similar results with both assays [41,44,50]. This raises questions that include (i) the effect of heparin, particularly at low doses, on VET results, (ii) whether heparinase or polybrene added to heparinized blood completely neutralized circulating heparin, and (iii) whether VETs can be used to guide heparin therapy. These questions were not raised by the previously published reviews [69][70][71][72].
The effect of heparin (UFH and LMWH) on VETs (performed without heparin neutralization) seems, according to the literature, to depend mainly on the heparin dose and the VETs used. Two trends have emerged for prophylactic or therapeutic concentrations (anti-Xa up to 1.5 IU/mL). First, with TEG, anti-Xa levels and R and K parameters (clot initiation) seemed to correlate, while an inverse correlation between anti-Xa levels and α angle and MA is observed (fibrin polymerization), sometimes leading to a 'flat line' with the highest anti-Xa levels [109][110][111][112]. Second, fewer data are available for ROTEM, but there seems to be a correlation only between anti-Xa levels and CT parameter from the INTEM assay (clot initiation) [112][113][114].
In the setting of cardiac surgery with cardiopulmonary bypass, VETs have been increasingly used, and manufacturers have adapted testing to neutralize circulating heparin by the addition of heparinase or polybrene in order to differentiate between insufficient heparin neutralization in patients with protamine from underlying post-bypass coagulopathy. Few data are available, however, on whether this neutralization is complete. In an in vitro study performed with TEG and coated cups with heparinase [109], results were similar between native samples and samples spiked with heparin (UFH or LMWH) or danaparoid, but tested concentrations were too low (0.005 to 0.05 IU/mL) to be clinically relevant. Another in vitro study performed with ROTEM and heparinase [113] showed similar results between native samples and samples spiked with increasing heparin (HNF) concentrations ranging from 0.1 to 1 IU/mL.
Among the 44 COVID studies we retrieved, several mentioned a considerably high incidence of thrombotic events despite thromboprophylaxis, in line with most reports, and raising the potential interest of increasing anticoagulant doses. Could VETs then be useful to identify patients who will benefit from a higher dose of thrombosis anticoagulant prophylaxis? This remains a matter of debate. According to the authors, VETs seem able to detect coagulation abnormalities advocating for a hypercoagulable prothrombotic state in a broad sense, including procoagulable state (with a decreased clot formation time), hypercoagulable state (with an increased clot strength), and impaired fibrinolysis (with a reduced or absent clot lysis) early in the course of the disease, and even if conventional coagulation tests remain in the reference ranges. Second, as reduced or absent fibrinolysis was associated with an increased risk of thrombotic events outcome despite anticoagulation [34,[42][43][44]51,54,57,63,64], we could assume that VET results might be used to adapt level of anticoagulation. Five studies [26,47,59,61,67] have reported VET parameters from ICU COVID-19 patients before and after an intensification of thromboprophylaxis, but results were inconsistent. Two of them showed a decrease in clot mechanical strength and in functional fibrinogen level [47,67], and even in the rate of thrombotic events [47], and the three others [26,59,61] did not find any significant difference. However, if VETs still demonstrate a hypercoagulable pattern despite anticoagulation at least with a prophylactic dose and even an anti-Xa level within the target range [60,61], it was not always associated with thrombotic outcomes, although there was no systematic VTE screening either [28,35,58,63]. Third, three studies have shown an exaggerated thrombin generation despite anticoagulation, at least with a prophylactic dose [39,48,66], advocating for a new way to monitor efficiency of thromboprophylaxis.
Rather than VETs, the study of thrombin generation could be more interesting to adjust anticoagulant therapy as heparin inhibits thrombin generation by multiple pathways as reviewed elsewhere [115]. Several methods exist to study thrombin generation, either with biomarkers such as prothrombin fragment 1 + 2 or thrombin-antithrombin complexes [116] (thrombin generation in vivo), or in vitro by assessment of the levels of thrombin over time (through the use of a chromogenic or fluorogenic substrate) in response to initiation of coagulation. Regarding the latter, several commercial devices and assays are available [117,118]. Of note, thrombin generation assays (TGAs) are highly sensitive to preanalytical aspects [119]. So far, some in vitro studies have reported that there was a heparin concentration dependent decrease in thrombin generation [120,121]. Studies showed that COVID-19 patients had a higher endogenous thrombin potential [31,39,48,66,[122][123][124][125][126][127] than manufacturer's reference range, healthy controls or patients with sepsis, sometimes despite UFH or LMWH anticoagulation at a minimum of prophylactic dosing. Few studies found a heparin dose-dependent decrease in thrombin generation [125,127] as described in vitro. Interestingly, one study showed that a persisting thrombin burst despite anticoagulation correlated with non-survival [123], whereas another found no difference between noncriti-cal and critically ill COVID-19 patients [124]. Further studies are needed to evaluate the clinical value in this context of the in vitro study of thrombin generation, and particularly with the new automated ST Genesia device [118], as already mentioned elsewhere [128]. A potential issue could be the non-availability of such device and the high turn-around-time for a result There is also an unmet need regarding the exploration of fibrin polymerization and lysis [95] with good and convenient assays.

Summary of the Conclusions of the Previously Published Reviews
Conclusions of the previously published reviews are summarized in Table 11. Overall, the four reviews reported the same findings as we do: COVID-19 patients displayed an abnormal VET pattern [69][70][71][72], but further studies are needed for various reasons. Moreover, we challenge the idea that such a pattern represents hypercoagulability; one main reason is that inhibitory systems are not at all taken into account, in sharp contrast with TGAs.
Of note, except the systematic review about the potential usefulness of TEG [71], no consistent association between the abnormal VET pattern and clinical outcome could have been demonstrated. Interestingly, one review [69] pointed out the potential usefulness of VETs in accurately assessing plasma fibrinogen levels in COVID-19 patients receiving DTI (see Section 4.4). VETs can detect a hypercoagulable state and fibrinolysis shutdown in COVID-19 patients, and might be used to identify patients with high prothrombotic risk for whom an antithrombotic therapy would be benefic

Conclusions
VETs are now well established in acute settings to assist in bleeding management and transfusion practices, with convenient, fully automated devices and ready-to-use reagents. Since the beginning of the pandemic, they were used to characterize hemostasis abnormalities in critically ill COVID-19 patients. As already reported in previous reviews [69][70][71][72], almost all the studies we analyzed reported increased clot strength, considered to be a hallmark of the 'hypercoagulable state', often associated with impaired fibrinolysis (with the analytical limitations we have emphasized)-globally referred to as 'prothrombotic pattern', but there was no consistent association with clinical outcomes. Indeed, few studies suggested an association with the occurrence of thrombotic events, as well as with the need for hemodialysis [34,44,51,56]. However, lack of power (low number of studied patients), retrospective design, and no standardized study protocol are of concern.
In the COVID-19 setting, the appraisal of (high) fibrinogen levels through VET as opposed to the Clauss method in the laboratory is not an obvious asset. As already pointed out however [69], VET could be of interest for accurately assessing plasma fibrinogen levels in COVID-19 patients receiving DTI through the assessment of the clot amplitude in functional fibrinogen assay. Modified VETs (with addition of a plasminogen activator) to detect, quantify, and monitor hypofibrinolysis in whole blood (with the advantage for instance to integrate the PAI-1 released by platelets) could be of clinical relevance [24,39,40,68].
Three different types of studies would be needed. First, prospective ones comparing the results from the different available devices are needed. Second, as it was already highlighted by previously published reviews [69][70][71][72], further prospective studies are needed, ideally randomized, to highlight the added-value of VET in predicting the clinical course of the disease, addressing patients to the appropriate ward according to their risk stratification, and identifying which patients would benefit from intensified anticoagulant treatment and those who would show clot resistance to fibrinolysis. Third, prospective randomized controlled trials are needed to evaluate the usefulness of VET and TGA in monitoring and adapting thromboprophylaxis.