Rinske Loeffen, Laboratory for Clinical Thrombosis and Hemostasis, Department of Internal Medicine, Cardiovascular Research Institute Maastricht, Maastricht University Medical Center, PO Box 616, UNS50: Box 8, 6200 MD Maastricht, the Netherlands. Tel.: +31 433884262; fax: +31 433884159. E-mail: firstname.lastname@example.org
Summary. Background: Thrombin generation assays are sensitive methods for assessment of the overall clotting potential of plasma, but, despite their common use in thrombosis research, standardization of preanalytic conditions is lacking. In order to set up a standardized protocol, we analyzed different preanalytic variables and validated the calibrated automated thrombogram method.
Methods and Results: Thrombin generation was assessed with 0, 1 and 5 pm tissue factor (TF). Variations in thrombin generation were mostly attributable to the type of collection tube, mainly because of variations in contact activation. The collection tube also determined the influence of other preanalytic variables on thrombin generation, e.g. the need for a discard tube, the storage of whole blood, and the centrifugation method. Regarding the collection system, blood drawn through intravenous catheters or butterfly needles showed significantly more hemolysis than blood obtained with venipuncture using conventional needles. The results showed that a discard tube is still needed for thrombin generation measurements. After blood collection, whole blood is best centrifuged immediately, to prevent activation or degradation of coagulation proteins, and a second centrifugation step at 10 000 × g is recommended. After thawing, plasma is best analyzed immediately, as storage resulted in thrombin generation results outside the 10% range of the reference sample. On the basis of these results, we set up an in-house standardized protocol, which was used for validation, resulting in coefficients of variations of < 15% for all derived parameters with both the 1 and 5 pm TF triggers.
Conclusion: Thrombin generation was greatly influenced by preanalytic conditions, demonstrating the need for an international standardized protocol.
The preanalytic phase includes every step from specimen collection up to the point of actual testing. According to the literature, it is the phase in which most laboratory errors occur [1–4], with reported error rates of > 60% . Because of activation of the hemostatic system in the preanalytic phase, coagulation assays are probably most susceptible to outcome variations related to inadequate sample quality [4,5]. An important specialized coagulation assay is the measurement of thrombin generation by means of the calibrated automated thrombogram (CAT) method (Thrombinoscope BV, Maastricht, the Netherlands). Thrombin generation assays provide an overall assessment of an individual’s coagulation potential, and have the ability to be used as a clinical diagnostic tool [6–8].
Up to now, however, there has been sunstantial heterogeneity concerning the preanalytic variables and the experimental conditions for the assessment of thrombin generation [7–11]. A review of previous clinical studies using the CAT method demonstrated large variability in results, depending on differences in reagents and methodology . Although thrombin generation has the potential to be used as a clinical diagnostic test in the near future, standardized preanalytic conditions and a validated method are required before large multicenter studies can be conducted and routine clinical practice can be applied. Therefore, in order to allow more extensive application of the CAT method and increase test accuracy and reliability, we analyzed the different preanalytic variables of thrombin generation and validated the CAT method, developing a standardized protocol.
Materials and methods
Blood collection system
Venous blood was collected from 12 healthy volunteers with three different collection devices: a conventional Vacutainer 21-gauge needle, 0.8 × 38 mm (Becton Dickinson, Plymouth, UK), a Butterfly winged 21-gauge needle infusion set, 0.8 × 20 mm, with 30-cm grade PVC tubing (Vygon, Ecouen, France), and a Vasofix intravenous polyurethane catheter, 20-gauge, 1.1 × 33 mm, with an injection port (Braun, Melsungen, Germany). The needle diameter was standardized to 21-gauge for the conventional and butterfly needle and to 20-gauge for the intravenous catheter (Braun 21-gauge intravenous catheter not available). These three devices will be subsequently referred to as conventional straight needle, butterfly needle, and intravenous catheter. Consecutive venipunctures were performed on three different veins of the arms, alternating one arm with the other. Blood was collected in 3.2% (w/v) citrated Vacutainer glass tubes (Becton Dickinson), and platelet-poor plasma (PPP) was prepared according to the standard protocol at our laboratory, consisting of an initial centrifugation step at 2000 × g for 5 min and a second centrifugation step at 10 000 × g for 10 min. For all preanalytic experiments, aliquots were snap-frozen in liquid nitrogen and stored at − 80 °C until analysis.
Blood collection tube
Venous blood was collected from 16 healthy volunteers with seven different collection tubes: Vacutainer glass 4.5 mL, 3.2% citrate (Becton Dickinson), Vacutainer plastic 2.7 mL, 3.2% citrate (Becton Dickinson), Vacuette plastic 9 mL, 3.2% citrate (Greiner Labortechnik, Kremsmunster, Austria), Venoject plastic 3.6 mL, 3.2% citrate (Terumo Europe, Leuven, Belgium), Monovette plastic 4.3 mL, 3.2% citrate (Sarstedt, Nümbrecht, Germany), SCAT plastic 4.5 mL, 3.2% citrate (Haematologic Technologies, Essex Junction, VT, USA), and SCAT plastic 4.5 mL, 50 μg mL−1 corn trypsin inhibitor (CTI)/3.2% citrate (Haematologic Technologies). Venous blood collection was performed by antecubital venipuncture with 21-gauge needles, followed by two-step centrifugation.
Venous blood was collected from 16 healthy volunteers, in two consecutive 3.2% (w/v) citrated Vacutainer glass tubes, with 21-gauge needles, without discarding the first tube. The phlebotomy procedure was repeated with two consecutive 3.2% (w/v) citrated Monovette plastic tubes. All tubes were processed and measured independently to determine the influence of the first tube on thrombin generation. Centrifugation was performed according to our standard protocol.
Whole blood stability
Venous blood was collected from 12 healthy volunteers by antecubital venipuncture with 21-gauge needles and 3.2% (w/v) citrated Vacutainer glass tubes. Per individual, four tubes were collected, three of which were stored as whole blood for 6 h at varying temperatures (4 °C, room temperature, and 37 °C). One tube was used as a reference sample, and centrifuged immediately. The phlebotomy procedure and whole blood storage was repeated with four consecutive 3.2% (w/v) citrated Monovette plastic tubes. Two-step centrifugation was performed according to our standard protocol.
Venous blood was collected from 12 healthy volunteers by antecubital venipuncture with 21-gauge needles and two consecutive 3.2% (w/v) citrated Vacutainer glass tubes. The tubes were centrifuged separately by means of two different centrifugation methods. The first method was the standard PPP centrifugation protocol used at our laboratory (2000 × g for 5 min, 10 000 × g for 10 min), and the second method consisted of a single centrifugation step of 2000 × g for 15 min. Venous blood collection and processing according to the two centrifugation methods were repeated with two consecutive 3.2% (w/v) citrated Monovette plastic tubes.
Six consecutive series of three different normal pooled plasma (NPP) tubes (pooled plasma of 80 healthy volunteers) collected by antecubital venipuncture with 21-gauge needles and 3.2% (w/v) citrated Vacutainer glass tubes were thawed for 10 min at 37 °C. Every 10 min up to 60 min, three tubes were thawed and stored at three different temperatures (4 °C, room temperature, and 37 °C). At the same time as the start of incubation of the last set, a fresh NPP tube was thawed, serving as a reference sample. After incubation of the last series, all samples were measured immediately for thrombin generation.
Thrombin generation in human PPP was measured with the CAT method (Thrombinoscope BV), which employs a low-affinity fluorogenic substrate for thrombin (Z-Gly-Gly-Arg-AMC) to continuously monitor thrombin activity in clotting plasma. Measurements were conducted in 80 μL of human PPP in a total volume of 120 μL (20 μL of fluorogenic substrate, calcium chloride [FluCa], and 20 μL of trigger reagent). The following experimental conditions were used: 0 pm tissue factor (TF) with 4 μm phospholipids (PLs) at 20 : 20 : 60 mol% phosphatidylserine/phosphatidylethanolamine/phosphatidylcholine, 1 pm TF with 4 μm PL, and 5 pm TF with 4 μm PL (all CAT reagents, including TF, were obtained from Thrombinoscope BV). Measurements were performed after 10 min of preheating at 37 °C in the fluorometer. In order to correct for inner-filter effects and substrate consumption, each thrombin generation measurement was calibrated against the fluorescence curve obtained in a sample from the same plasma (80 μL), supplemented with a fixed amount of thrombin–α2-macroglobulin complex (20 μL of Thrombin Calibrator; Thrombinoscope BV) and 20 μL of FluCa. Fluorescence was read in a Fluoroskan Ascent reader (Thermo Labsystems OY, Helsinki, Finland) equipped with a 390/460 filter set, and thrombin generation curves were calculated with the Thrombinoscope software (Thrombinoscope BV).
Validation of the CAT method was established by means of the Clinical and Laboratory Standards Institute (CLSI) EP5 protocol . To determine the within-run and between-run variability of the method, the Complex Precision Module was used. For this, thrombin generation measurements were conducted for 10 days, with two runs a day and two replicates per run. We applied the following preanalytic conditions: venous blood collection was performed by antecubital venipuncture with 21-gauge needles and 3.2% (w/v) citrated Vacutainer glass tubes, with the first tube being discarded. We used Vacutainer tubes because, up to now, these have been the standard collection tubes used for patient care at our hospital. Samples were centrifuged within 30 min by double centrifugation, and aliquots were snap-frozen in liquid nitrogen and stored at − 80 °C until analysis. After being thawed for 10 min at 37 °C, samples were analyzed immediately. Measurements were conducted with the following experimental conditions: 1 and 5 pm TF, with 4 μm PL. To avoid differences between batches of reagents, all validation experiments were conducted with the same batch of CAT reagents from Thrombinoscope.
Thrombin generation results were collected for statistical analyses, which were performed with prism, version 5.00 (GraphPad Software, San Diego, CA, USA). According to the distribution of the variables (D’Agostino & Pearson omnibus normality test), data are expressed as mean ± standard deviation (SD) or median ± interquartile range (IQR). Differences between thrombin generation measurements were analyzed with a paired Student’s t-test or a Wilcoxon signed rank test for comparison of two groups, and by repeated measures anova or a Friedman test for comparison of more than two groups. A P-value of < 0.05 was considered to be statistically significant.
Blood collection system
On comparison of the three collection devices, an important finding was that, after centrifugation of the samples, blood drawn through intravenous catheters and butterfly needles showed significantly more hemolysis (optical analysis) than blood drawn with a conventional straight needle. For five of the 12 healthy volunteers, the phlebotomy procedure had to be repeated because of hemolytic samples. In three cases, this was because of hemolyzed samples drawn with intravenous catheters, and in two cases butterfly needles caused the hemolysis. Regarding plasma thrombin generation, no significant difference could be established between the three collection devices in endogenous thrombin potential (ETP) and peak height for the 0, 1 (Fig. 1) and 5 pm TF triggers (data not shown). On comparison of the 0 pm TF thrombin generation results for the butterfly needle and intravenous catheter with those for the conventional straight needle, small average reductions of only 2.6% and 2.9% for ETP, and 4.0% and 2.9% for peak height, respectively, were obtained. For the 1 pm TF trigger, comparison of the butterfly needle and intravenous catheter with the conventional straight needle showed similar reductions of 3.9% and 1.1% for ETP, and 3.3% and 0.5% for peak height, respectively.
Blood collection tube
We compared seven different collection tubes, and, as shown in Fig. 2, thrombin generation results differed importantly between the tubes, for both the 0 and 1 pm TF triggers. For the 0 pm TF trigger, the following median ETP levels were established for the different tube types: Vacutainer glass, 618 nm min (IQR 324–798); Vacutainer plastic, 465 nm min (IQR 419–828); Venoject, 406 nm min (IQR 235–693); Vacuette, 490 nm min (IQR 357–885); Monovette, 270 nm min (IQR 0–386); SCAT, 508 nm min (IQR 466–1091); and SCAT (CTI), 0 nm min (IQR 0–49). These results demonstrate that the addition of CTI to the SCAT tube caused total inhibition of in vitro contact activation. Furthermore, the tubes without addition of CTI showed large variability in ETP and peak height, with the Venoject and Monovette tube activating the contact system least. For the 1 pm TF trigger, the variability between the collection tubes was still detectable, although less pronounced, because of the higher TF concentration. However, when 5 pm TF was used, the thrombin generation results were no longer significantly different between the tubes (data not shown). Blood drawn into Vacutainer glass, Vacutainer plastic, SCAT and Vacuette tubes showed significantly increased thrombin generation as compared with Monovette and SCAT (CTI) tubes, for both the 0 pm TF ETP and peak height. For the 0 pm TF trigger, a significant difference in thrombin generation was also observed between the Venoject tube and both the Vacutainer glass (peak height) and SCAT (ETP and peak height) tubes. As compared with the SCAT tube, thrombin generation remained significantly different when triggering with 1 pm TF for plasmas obtained from the Venoject (ETP), SCAT (CTI) and Monovette (ETP and peak height) tubes. Furthermore, for the 1 pm TF trigger, thrombin generation remained significantly different between the SCAT (CTI) tube and both the Vacutainer glass (peak height) and plastic (ETP and peak height) tubes, and between the Monovette tube and the Vacutainer plastic tube (ETP).
Need for a discard tube
We found that when Vacutainer glass tubes were used (Fig. 3A) a clear difference was seen for both the 0 and 1 pm TF trigger, with a significantly increased (P < 0.0001) ETP and peak height for plasmas obtained from the first tube as compared with the second tube. However, on comparison of plasmas obtained with Monovette plastic tubes (Fig. 3B), no significant difference in thrombin generation could be established between the first and second collected tubes. When triggering was performed with 0 pm TF, considerable average reductions in ETP of 11.2% and peak height of 21.1% were obtained on comparison of the first collected Vacutainer tube with the second one. For the 1 pm TF trigger, the average reductions in ETP and peak height were 8.3% and 18.3%, respectively. For the Monovette tubes, these differences in thrombin generation between the two consecutive tubes were much smaller, with peak heights in the first tube being, on average, 1.6% higher for the 0 pm TF trigger and 3.9% lower for the 1 pm TF trigger than in the second tube. Regarding the ETP, comparison of the first tube with the second tube demonstrated average increases of 6.3% for the 0 pm TF trigger and 1.9% for the 1 pm TF trigger. Triggering with the higher TF concentration of 5 pm led to no significant differences in thrombin generation results for both tube types (data not shown).
Whole blood stability
Our results demonstrate that incubation of whole blood collected with Monovette plastic tubes (Fig. 4B) for 6 h at 4 °C, room temperature or 37 °C had strong effects on thrombin generation as compared with directly centrifuged specimens. However, when Vacutainer glass tubes (Fig. 4A) were used for blood collection, no significant difference in thrombin generation could be established. For the Monovette tubes, significant differences in ETP were obtained between the control and all three incubation temperatures (P < 0.008), for both the 0 and 1 pm TF triggers. Both tubes showed no significant differences in thrombin generation results when triggering was performed with 5 pm TF (data not shown). The peak height was significantly increased for the 0 pm TF trigger when whole blood was stored at 4 °C (P = 0.0092) or room temperature (P = 0.0005) as compared with direct plasma preparation. On average, for the 0 pm TF trigger, there were increases in ETP and peak height of 41.6% (576 vs. 407 nm min) and 30.5% (58 nm vs. 44 nm) when whole blood was incubated at 4 °C, and of 36.1% (553 nm min vs. 407 nm min) and 35.4% (60 nm vs. 44 nm) when it was incubated at room temperature. For the 1 pm TF trigger, these differences in thrombin generation between direct plasma preparation and incubation at 4 °C or room temperature were smaller; however, they were still substantial, ranging between 10.1% and 21.5%. The least variation was found when whole blood was incubated at 37 °C, with both TF triggers generating differences in peak height of < 8.1%, as compared with the control. When Vacutainer glass tubes were used, variations in thrombin generation (0 and 1 pM TF trigger) between direct plasma preparation and whole blood incubation at 4 °C, room temperature and 37 °C were not statistically significant, with average differences in ETP and peak height being 1.5–8.3% and 4.7–13.8%, respectively.
When Monovette plastic tubes (Fig. 5B) were used, a clear difference in thrombin generation was found for both the 0 and 1 pm TF triggers, with a significantly increased ETP (P = 0.0005 and P = 0.0256, respectively) for single-centrifuged samples as compared with double-centrifuged ones. However, on comparison of centrifugation methods for plasma collected with Vacutainer glass tubes (Fig. 5A), triggering with 0 pm TF showed no significant differences in thrombin generation results. Surprisingly, when thrombin generation was triggered with 1 pm TF, a significant increase in ETP (P = 0.0466) was found for single-centrifuged as compared with double-centrifuged samples. Triggering with the higher TF concentration of 5 pm led to no significant differences in thrombin generation results for both tube types (data not shown). On average, for the 0 pm TF trigger, there were increases in ETP and peak height of 31.2% (814 nm min vs. 560 nm min) and 33.3% (83 nm vs. 56 nm), respectively, on comparison of single-centrifuged with double-centrifuged plasma collected with Monovette tubes. For the Vacutainer tubes, these differences in thrombin generation (0 pm TF) between the two centrifugation methods were considerably smaller, more specifically 2.5% (1158 nm min vs. 1129 nm min) in ETP and 7.5% (271 nm vs. 251 nm) in peak height. For the 1 pm TF trigger, average reductions in ETP of 7.1% (1057 nm min vs. 982 nm min) and peak height of 6.3% (136.4 nm vs. 127.8 nm) were found for the Monovette tubes, and of 4.2% (1177 nm min vs. 1128 nm min) in ETP and 10.5% (224 nm vs. 200 nm) in peak height for the Vacutainer tubes.
Figure 6C,D shows the established ETP and peak height values of the incubated plasma samples. As shown in the graphs, for both TF triggers, as compared with the reference sample, incubation of plasma at 4 °C, room temperature and 37 °C for varying incubation periods resulted in > 10% variation (dotted lines) in thrombin generation results. As compared with immediate plasma analyses, incubation of plasma for 60 min at 37 °C resulted in important decreases in ETP and peak height values of 37.9% (900 nm min vs. 559 nm min) and 47.4% (137 nm vs. 72 nm) for thrombin generation triggered with 0 pm TF. Incubation at 4 °C for 10 min up to 60 min caused strong variations in ETP and peak height, best seen for the 1 pm TF trigger, with maximum increases as compared with the reference sample of 28.0% (970 nm min vs. 1242 nm min) in ETP and 37.2% (120 nm vs. 165 nm) in peak height. As compared with immediate plasma analyses, the least variation in thrombin generation was obtained when plasma was incubated at room temperature, with the 0 pm TF ETP results being inside the 10% range. However, the 1 pm TF thrombin generation results for plasma incubated at room temperature showed maximum increases in ETP and peak height, as compared with the reference sample, of 18.6% (970 nm min vs. 1151 nm min) and 18.0% (120 nm vs. 142 nm), respectively. No significant differences in thrombin generation results were found when triggering was performed with the higher TF concentration of 5 pm (data not shown).
Validation of the CAT method
After analysis, the coefficient of variation (CV) and SD of the within-run and between run variation were determined for the ETP, peak height and lag time of both experimental conditions, as shown in Table 1.
Table 1. Results of thrombin generation complex precision analysis
CV, coefficient of variation; ETP, endogenous thrombin potential; SD, standard deviation; TF, tissue factor.
1 pm TF
Lag time (min)
Peak height (nm)
ETP (nm min)
5 pm TF
Lag time (min)
ETP (nm min)
Peak height (nm)
Our results demonstrate that, in the absence of TF and even at low TF concentrations, thrombin generation was strongly influenced by blood collection tube-dependent activation of the intrinsic pathway, which could be completely inhibited by addition of CTI to the collection tube. On the basis of our results, we propose the protocol as indicated in Table 2 for routine clinical thrombin generation analysis. A standardized protocol that reduces the preanalytic variability is needed in order to reduce the analytic variability. Application of a standardized procedure for thrombin generation resulted in acceptable validation criteria, with CVs for most thrombin generation parameters of < 10%.
Table 2. Preanalytic recommendations for the calibrated automated thrombogram assay
Recommendations for thrombin generation
Blood collection system
The conventional straight needle is recommended as the device of first choice; butterfly needles or intravenous catheters are only accepted as reliable alternatives when a proper phlebotomy technique is used and samples are carefully checked for hemolysis
Blood collection tube
Addition of CTI to the collection tube is recommended to inhibit contact activation; of the tubes without CTI addition, the Monovette tube is the one lowest in inducing contact activation and therefore the best alternative to the CTI tube
Discard tube necessity
A discard tube is required for thrombin generation measurements
Whole blood incubation
Direct plasma preparation is preferred to storage of whole blood for 6 h
Double centrifugation is preferred to single centrifugation, with the following protocol: 2000 × g for 5 min, 10 000 × g for 10 min
After plasma thawing, immediate analysis is recommended; when direct measurement is not possible, plasma is most stable when kept at room temperature
In routine clinical laboratory practice, different collection devices are used for venipuncture. However, owing to costs and the risk of obtaining unsuitable samples, the use of butterfly needles and intravenous catheters has generally been discouraged . Different studies demonstrated that blood drawn through plastic catheters caused significantly more hemolysis and test cancellation than blood drawn with a conventional straight needle [14,15]. However, Lippi et al., comparing the butterfly device with the conventional straight needle, found no statistical difference in the rate of hemolysis or coagulation test results between the two devices [13,16]. Our study on the other hand, demonstrated an important difference in hemolysis rate between the investigated collection systems, with the butterfly needle and intravenous catheter resulting in more hemolytic samples than blood collection with a conventional straight needle.
Besides hemolysis, activation of the contact system, caused by activation of factor XII, can occur when blood comes into contact with the surface of a biomaterial [17,18]. It is suggested that the passage of whole blood through butterfly tubing and intravenous catheters might cause increased hemostatic alterations in comparison with blood collection with a conventional straight needle, directly into the tube [13,18]. Regarding thrombin generation testing, the role of the contact system is considered to be one of the potential problems in routine clinical practice. A study on the influence of the collection device on the CAT showed that the use of a butterfly needle for blood collection resulted in increased thrombin generation, most likely because of induced contact activation by the long tubing of the butterfly device . However, in contrast to this previous study, our present data showed no difference in the degree of hemostatic activation generated by the three collection devices. Unfortunately, no conclusive explanation can be provided for these contradictory results, other than the difference in sample number. Despite the fact that no significant difference could be established between the three systems, on consideration of the observed sample hemolysis, we recommend that the conventional straight needle should be the device of first choice for blood drawing.
Regarding the collection tube, preanalytic variability may be related to the citrate concentration, filling level, mixing procedure, and type of material used for the tube itself and the rubber stopper . Studies focusing on prothrombin time (PT) analysis demonstrated both longer [19,20] and shorter PT values [21,22] when blood was collected with plastic tubes as compared with siliconized glass tubes. The studies investigating a broader range of coagulation assays found small, although statistically significant, differences for some of the tests; however, most results were not considered to be clinically significant [23,24]. Regarding the effect of the tube material on thrombin generation, this study showed no significant difference between the Vacutainer glass and plastic tube for all experimental conditions; however, large variations were established between the different plastic containers. The type of collection tube strongly influenced thrombin generation, and, as demonstrated by the effect of CTI addition, this difference was primarily attributable to variations in contact activation. As the draw volume, mixing procedure and citrate concentration were similar between the tubes, the differences in thrombin generation were probably related to altered interactions of the blood with the interior container wall.
The CLSI abandoned its recommendation to draw a discard tube when routine coagulation testing, such as PT and activated partial thromboplastin time, is performed. However, for specialized coagulation assays, and when butterfly needles or intravenous catheters are used, a discard tube is still recommended . Several studies, however, have demonstrated that the evidence for drawing a discard tube is lacking, not only in routine practice [26,27] but also in specialized coagulation assays [28,29]. Plasma derived from the first collection tube is presumed to have an increase in coagulation activation. As thrombin generation measurements are conducted with low TF concentrations, this suspected increased coagulation activation may influence thrombin generation. As demonstrated by our results, the differences in thrombin generation between the first and second collected tubes depended on the type of collection tube used for blood drawing. Whereas Vacutainer glass tubes showed a significant difference in thrombin generation between the first and second collected tubes, this effect was not observed with Monovette plastic tubes. The difference between the two tube types is most likely attributable to the amplifying effect on contact activation caused by the Vacutainer tube, resulting in higher thrombin generation levels in the first tube.
Published studies on coagulation testing vary in their recommendations concerning the incubation time and the temperature of whole blood [30–33]. Delays between sample collection and analysis can cause in vitro degradation of coagulation proteins. Regarding incubation temperature, it is recommended that samples are stored or transported at ambient temperature, as exposure to cold may cause activation of FVII and exposure to heat may cause degradation of FV and FVIII [4,5,34]. Considering our present results, when using Monovette tubes, incubation of whole blood for 6 h at varying temperatures significantly influenced thrombin generation. However, given the relatively high amount of contact activation in Vacutainer glass tubes, we presume that the small effects of activation or degradation of coagulation factors in these tubes did not result in detectable differences in thrombin generation. Although we did not investigate the effect of shorter incubation times (< 6 h), it can be concluded from this study that, for thrombin generation, direct plasma preparation is preferred to whole blood storage.
In order to eliminate platelet debris and microparticles from plasma, which may contribute to the variability in thrombin generation results, a second centrifugation step at 10 000 × g is employed for plasma used for thrombin generation. Depending on the type of collection tube, thrombin generation was influenced by the centrifugation method, suggesting that, for thrombin generation, double-centrifuged samples are preferable. Again, results differed between the two collection tubes analyzed, most likely because of differences in contact activation between the Vacutainer and Monovette tubes.
Thrombin generation is most often performed with frozen samples, frequently by batch testing. With regard to plasma storage after freezing–thawing, our data suggest that immediate analysis is preferable to plasma storage at 4 °C, room temperature, and 37 °C, which resulted in thrombin generation results outside the 10% range of the reference sample. However, when immediate analysis is not possible, plasma is most stable when incubated at room temperature, instead of 4 °C or 37 °C, most likely because of more activation and degradation of coagulation proteins at these temperatures. Besides the plasma incubation temperature, de Smedt et al.  demonstrated that preheating conditions also strongly influenced thrombin generation. In our experiments, preheating was standardized to 10 min at 37 °C in the fluorometer, which was set to 38–39 °C to obtain an in-well temperature of 37 °C.
In conclusion, this study represents the first attempt to define the preanalytic determinants of thrombin generation parameters. We have shown that the preanalytic variable with the strongest influence on thrombin generation is the collection tube. This is demonstrated not only by the differences between the seven analyzed collection tubes, but also by the experiments investigating the influence of the discard tube, the storage of whole blood, and the centrifugation method. In order to take thrombin generation towards routine clinical application, standardization of the test conditions is extremely important.
Disclosure of conflict of interests
This research was performed within the framework of the Center for Translational Molecular Medicine (http://www.ctmm.nl), project INCOAG (grant 01C-201), and supported by the Dutch Heart Foundation.