Introduction
Thromboelastometry is a point-of-care assay measuring the viscoelastic properties of clot formation and clot lysis of whole blood [
1]. The results of such measurements offer quick information about the global coagulation profile [
2].
This may be useful in certain clinical situations, i.e., during surgery, in bleeding patients, in patients with hemorrhagic disorders, to monitor certain drugs or when massive transfusions are required [
3‐
9]. Furthermore, the impact of hemostatic interventions can be rapidly assessed and decisions may be based on results of such viscoelastic tests [
10]. The advantages of thromboelastometry-guided hemostatic therapy were demonstrated in patients undergoing cardiac surgery [
11]. Blood loss and the need for fresh frozen plasma units could be reduced. Moreover, the perioperative need of blood products could be reduced in patients with severe burn injuries [
12]. The diagnosis of hyperfibrinolysis is often difficult and testing the viscoelastic properties of blood may offer important information to support the diagnosis [
13,
14] or to diagnose hyperfibrinolytic disseminated intravascular coagulation [
15]. Thromboelastometry is also sensitive to anticoagulant drugs such as enoxaparin [
16], apixaban, rivaroxaban, argatroban and dabigatran [
17‐
19].
Thromboelastometry is sensitive to the coagulation activation in the human endotoxemia model [
20] and may be used to identify prothrombotic states in patients [
21‐
23].
However, like every other testing system some limitations need to be considered. First of all, thromboelastometry produced accurate and reproducible results within 30 min up to 4 h, but inconsistent references are published on the stability and reproducibility of measurements after sample storage at room temperature [
24‐
26]. The intra-individual coefficient of variation of thromboelastometry ranged between 11 and 23% in healthy volunteers (own unpublished data). Secondly, trained personnel and available devices are necessary. In the clinical setting, usually rapid results are required which is an important advantage of this method. On the contrary, in the scientific setting, storage of samples for later batch analysis offers obvious advantages.
The aim of this study was to investigate whether the results of thromboelastometry performed on frozen plasma samples are comparable to the results obtained from freshly obtained whole blood. This may help to overcome at least part of these limitations.
Methods
This study was performed at the Department of Clinical Pharmacology at the Medical University of Vienna. The independent Ethics Committee of the Medical University of Vienna approved the study. Twenty healthy volunteers were included in the study, and blood sampling was performed between February and April 2016 (Table
1).
Table 1
Demographics and baseline data of participants
Gender m (f) | 11 (9) |
Age (years) | 26 ± 5 |
Height (cm) | 174 ± 10 |
Weight (kg) | 72 ± 17 |
Hemoglobin (g/dl) | 14.1 ± 1.3 |
Platelets (*10^9) | 265 ± 59 |
Leukocytes (*10^9) | 5.9 ± 1.3 |
Preparation of whole blood and plasma samples
Whole blood samples were obtained from fresh venipunctures in healthy volunteers and collected into 3.8% sodium citrate tubes.
To generate platelet-free plasma, whole blood was centrifuged twice at 2500g for 15 min at room temperature and the supernatant was transferred into polypropylene tubes. Plasma samples were either used for immediate analysis (fresh plasma) or frozen and stored as explained below (frozen plasma).
To investigate the effects of freezing on hyperfibrinolysis, we spiked alteplase (actilyse, 333 ng/ml) into whole blood samples to induce artificial hyperfibrinolysis. ROTEM analysis was performed for fresh blood, and again plasma was generated for immediate (fresh plasma) or later (frozen plasma) analysis as explained above.
To investigate the effects of oral anticoagulants, we spiked rivaroxaban (100 and 200 ng/ml) and dabigatran (500 and 1000 ng/ml) into whole blood and performed ROTEM analysis. Again plasma for immediate and later analysis was generated from the whole blood samples as explained above.
Plasma samples of all 20 healthy volunteers were frozen and stored at −80 °C until later analysis. These samples included native plasma and plasma containing rivaroxaban, dabigatran and alteplase, as explained above.
Additionally, we investigated whether different freezing techniques and storage temperatures affected the results of ROTEM analysis. Therefore, plasma of eight healthy volunteers was frozen using dry ice (−78 °C) and then transferred to −80 or −18 °C storage. Plasma samples of another eight subjects were frozen and stored at −18 °C, and plasma samples of another eight subjects were frozen using liquid nitrogen (−196 °C) and then transferred to −80 or −18 °C storage.
All frozen plasma samples were thawed using a water bath heated to 37 °C (C) for 15 min and analyzed thereafter. Whole blood and fresh plasma samples were tested at the same time, whereas frozen plasma samples were analyzed at one to four weeks later.
Thromboelastometry
The viscoelastic properties of whole blood or plasma were investigated with the ROTEM coagulation analyzer (Pentapharm, Munich, Germany) as previously described [
20]. In short, ROTEM measures shear elastic modulus during clot formation and subsequent fibrinolysis. The ROTEM uses a ball-bearing system for power transduction, which makes it less susceptible to mechanical stress, movement and vibration.
ROTEM measurements produced accurate and reproducible results within 30 min up to 4 h. In fresh blood samples, we performed ROTEM measurements between 1 and 2 h after blood storage at room temperature.
Just before running the assay, citrated blood samples were recalcified with 20 µl of CaCl2 0.2 M (Start-TEM; Nobis, Endingen, Germany) and the test was started. We performed thromboelastometry without adding additional activators (NATEM), as well as EXTEM analysis (recombinant tissue factor- and phospholipid-activated ROTEM), and in eight subjects INTEM analysis (partial thromboplastin–phospholipid-activated ROTEM). The following ROTEM parameters were analyzed: the clotting time (CT), the clot formation time (CFT), the maximum clot firmness (MCF), the alpha angle (alpha), the maximum lysis (ML), the time of maximum lysis (LT) and the area under the curve (AUC).
In samples with alteplase, dabigatran or rivaroxaban, only EXTEM analysis was performed.
ROTEM analysis in frozen plasma samples was generally technically feasible. However, one problem regularly occurred in the testing system and an error code was reported that the stability of the clot rapidly increased and the sample dried. All tests were running for 2 h. This problem may have affected the detection of ML and the AUC, because these parameters are measured continuously until the end of the chosen running time.
Of note, the clot signal amplitude did not reach 20 mm in all plasma samples resulting in no measurable CFT.
Statistical analysis
A formal sample size calculation was not performed, because no data on this topic were available. We performed nonparametric testing using the Friedman ANOVA for overall comparisons and the Mann–Whitney-U test for group-wise comparisons. Descriptive statistics are presented as means and standard deviations unless otherwise stated.
Discussion
We investigated whether ROTEM analysis performed in frozen, stored and thawed plasma samples provides similar results compared to whole blood samples. We only found few reports performing thromboelastometry in human plasma samples [
27‐
30]. Based on these, we expected a smaller MCF, a prolonged CFT and a reduced AUC due to a lack of blood cells. Clotting time was comparable in EXTEM and, to a lesser degree, in INTEM analysis.
In EXTEM analysis, the CT tended to be shorter in plasma samples. This difference, although statistically significant, was minor, and the clinical relevance is at least questionable. Of note, all results were within the reference values suggested by the manufacturer (43–82 s) [
31]. Moreover, another study measured a relatively short median CT of 47 s in EXTEM analysis using platelet free plasma [
27].
In contrast to EXTEM, INTEM and NATEM analysis showed a prolonged CT in fresh plasma samples compared to whole blood. INTEM analysis uses a less potent activator compared to EXTEM analysis, while NATEM analysis does not involve any activator. This suggests that in the absence of strong activators pro-coagulant surfaces seem to be more important for the initiation of the coagulation cascade. Indeed, in some samples coagulation activation did not occur at all or was prolonged to more than 3000 s. The CT of NATEM analysis in whole blood was longer in our population (856 ± 233 s) compared to the manufacturer’s reference values (254–837 s) [
31] and compared to other studies [
20,
32]. However, the variability of results in the non-activated thromboelastometry analysis was similar to these studies.
We furthermore investigated various freezing methods in a small number of samples (
n = 8), and except for samples frozen and stored at −18 °C and the samples frozen in dry ice and stored at −80 °C, the CT in NATEM analysis was 40–60% longer compared to whole blood samples. In EXTEM analysis, no differences were detectable except for the samples frozen and stored at −18 °C, in which the CT was ~10% shorter. It is conceivable that freezing, storage and/or thawing procedures have an impact on the results. One possible explanation for the varying results may be cold or contact activation of coagulation factors, which was shown for factor V, VIII, XII and consecutive thrombin generation [
33,
34]. To minimize the influence of such factors, we applied a strict freezing and thawing schedule. We used citrate-anticoagulated blood samples as suggested by the manufacturer. However, in contrast to citrate, corn trypsin inhibitor sufficiently suppressed contact activation in calibrated automated thrombin generation [
35‐
37]. Thus, it is possible that the use of corn trypsin inhibitor could improve the reproducibility of results. Numerically, the shortest CT of all plasma samples was measured in samples frozen and stored at −18 °C. Cold activation may have played the biggest part in these samples, because freezing itself takes longer than with snap-freezing. We therefore cannot recommend freezing and storing samples at −18 °C. Considering our results and given the limited availability of fluid N2 and dry ice, we recommend freezing and storing of plasma samples at −80 °C.
Residual platelet counts affect global coagulation tests or coagulation factors only by a small margin [
38]. However, we hypothesized that in the non-activated setting, the presence of pro-coagulatory surfaces may be of greater importance. To obtain platelet-free plasma, we performed two centrifugation steps at 2500
g. We measured platelet counts in a smaller number of plasma samples and detected a median platelet count <1 G/L. However, we cannot exclude entirely that very low platelet counts occurred in some samples. Thus, disintegration of any residual platelets and consecutive exposure to phospholipids may have led to the shorter clotting time in frozen plasma samples [
39].
The influence of dabigatran and rivaroxaban on viscoelastic properties of whole blood was demonstrated previously [
17,
40]. In plasma samples, the CT was further increased compared to whole blood. As we used the same whole blood samples for plasma generation as for ROTEM analysis, the concentration itself remained equal. It seems likely that in the presence of coagulation inhibitors such as rivaroxaban or dabigatran, the activator used in EXTEM analysis may not be potent enough to initiate coagulation and the lack of pro-coagulant surfaces becomes relevant. The CT is therefore further prolonged. Due to the magnitude of the observed effects, the use of frozen plasma samples in EXTEM analysis to investigate the presence of inhibitors of coagulation such as rivaroxaban or dabigatran seems especially suitable.
Thromboelastometry may be used as a supportive tool in the diagnosis of hyperfibrinolysis [
14]. To artificially induce hyperfibrinolysis, we spiked alteplase into blood samples as described earlier [
13,
41]. Hyperfibrinolysis was reproducible in frozen and stored plasma samples with a shorter time of maximum lysis in plasma samples. A likely explanation may be the reduced MCF in plasma samples with quicker complete lysis of the smaller clot. Interestingly, no difference was detected in the CT of plasma samples compared to whole blood indicating the reproducibility of the results. However, due to the mentioned technical limitations caused by the regular clotting of the entire sample, we assume that only hyperfibrinolysis of a certain extent is detectable in plasma samples. As we included only healthy volunteers with minimal clot lysis and a usually late onset of clot lysis in our study, we were not able to detect any form of clot lysis. In hyperfibrinolytic patients, lysis occurs within 30 min with a complete clot lysis within 60 min [
14]. This faster onset and the strong fibrinolysis should be initiated and therefore be detectable before the sample has clotted entirely. However, this needs to be confirmed in studies investigating hyperfibrinolysis in patients.
This study has several limitations. First of all, the sample size of 20, and in some instances 8, may be too small to detect differences or may be affected by outliers. Furthermore, we did not perform global or more specific coagulation tests before inclusion in the study or in thawed plasma samples. When ROTEM was performed in plasma samples, the error “dried plasma sample” occurred regularly, which probably affected measurement of low level ML. We only included healthy volunteers in our study. Possibly in thrombophilic conditions, NATEM analysis is more reliable due to the presence of activators of coagulation such as tissue factor.
In conclusion, thromboelastometry may be performed in frozen, stored and thawed plasma samples taking the mentioned limitations into account. Activated ROTEM tests such as INTEM or EXTEM are reproducible, and results are comparable between whole blood and corresponding fresh and frozen plasma, whereas non-activated NATEM analysis is subject to greater variability. Our test setup was suitable to detect potent hyperfibrinolysis with a rapid onset and the presence of anticoagulants.