BACKGROUND: The calibrated automated thrombogram (CAT) assay in plasma is a versatile tool to investigate patients with hypo- or hypercoagulable phenotypes. The objective was to make this method applicable for whole blood measurements.
METHODS: Thin-layer technology and the use of a rhodamine 110–based thrombin substrate appear to be essential for a reliable thrombin generation (TG) assay in whole blood. Using this knowledge we developed a whole blood CAT-based assay.
RESULTS: We demonstrated that the whole blood CAT-based assay is a sensitive and rapid screening test to assess function of the hemostatic system under more nearly physiological conditions than the TG assay in plasma. Under conditions of low tissue factor concentration (0.5 pmol/L) and 50% diluted blood, the intraassay CV of the thrombogram parameters, endogenous thrombin potential and thrombin peak height, were 6.7% and 6.5%, respectively. The respective interassay CVs were 12% and 11%. The mean interindividual variation (SD) of 40 healthy volunteers was 633 (146) nmol · min/L for the endogenous thrombin potential and 128 (23) nmol/L for the thrombin peak. Surprisingly, erythrocytes contributed more than platelets to the procoagulant blood cell membranes necessary for optimal TG. Statistically significant (P < 0.001) and potentially clinically significant correlations were observed between circulating factor-VIII concentrations in blood of hemophilia A patients and endogenous thrombin potential (r = 0.62) and thrombin peak height (r = 0.58).
CONCLUSIONS: We have developed a reliable method to measure TG in whole blood. The assay can be performed with a drop of blood and may provide a useful measurement of TG under more physiological conditions than plasma.
The thrombin generation (TG)3 assay is increasingly being recognized as a versatile diagnostic tool in the field of thrombosis and hemostasis. It is well accepted that the transient TG profile in clotting plasma is a better determinant of the overall function of the hemostatic system than are clotting-time–based assays, e.g., prothrombin time (PT) and activated partial thromboplastin time (1,–,5). TG in plasma is most accurately measured with the calibrated automated thrombogram (CAT) assay as developed by Hemker and coworkers (6,–,9). This method has been designed for application with platelet-free and platelet-rich plasma (PRP) as analytes. The method is, however, not applicable for measuring TG in whole blood. Variable quenching of the fluorescent signal by hemoglobin and red blood cells that sediment, cluster, and retract with the clot lead to highly erratic signals (10, 11). The use of fibrin as an indicator product for TG in whole blood, as in thromboelastography (TEG), does not reflect the thrombin-generating capacity of patients, because fibrinogen as a substrate is quickly exhausted and therefore informs about TG only at the earliest phases of the coagulation process (5, 12). The information about the mechanical properties of clots that is obtained in later stages of a TEG experiment is related to TG in a complicated and nonunequivocal manner. Mathematical treatment of the TEG signal so as to make it resemble a TG curve therefore does not contribute except to confusion (13, 14).
A reliable whole blood TG assay is important because it enables us to go one step closer to physiological conditions than is possible in present practice. Apart from this conceptual advantage there is the practical advantage that no centrifugation step is required, which opens the way to point-of-care TG measurements. We found that the problems caused by the presence of red blood cells can be overcome by dispersing the blood in a porous matrix, thereby creating a thin layer of blood, and by using rhodamine-110 as a signaling product (15). By these modifications of the plasma CAT assay, we acceptably reduced quenching and issues related to unequal distribution of red blood cells. We here report the technical validation of a whole blood CAT-based assay (WB-CAT) of acceptable imprecision that measures TG in a thin layer of whole blood. As proof of concept, we compared the whole blood thrombogram parameters with factor-VIII plasma concentrations in hemophilia A patients.
Materials and Methods
BSA5 buffer contains 20 mmol/L HEPES, 140 mmol/L NaCl, and 5 g/L BSA (Sigma) with a pH of 7.35. The rhodamine-based thrombin substrate (P2Rho) was a gift from Diagnostica Stago (Gennevilliers, France). The calibrator, α2-macroglobulin-thrombin (α2M-T) complex, was prepared in-house as previously described (7). Recombinant human tissue factor (TF; Innovin®) was from Siemens Healthcare Diagnostics. The FXIIa inhibitor corn trypsin inhibitor was purchased from Hematologic Technologies. Phospholipid vesicles (PV) contained 20 mol% phosphatidylserine, 60 mol% phosphatidylcholine, and 20 mol% phosphatidyl-ethanolamine (Avanti).
Blood was taken from healthy adult volunteers who gave full informed consent according to the Helsinki declaration and had not taken any medication that interferes with hemostasis. Twenty-seven adult hemophilia A patients with a mean (SD) age of 31 (10) years routinely treated in the Lyon Clinical Hemostasis Unit (France) were included in the study after the patients provided written informed consent. The study met all institutional ethics requirements. Of these 27 hemophilia A patients, 7 were had severe (factor VIII:C <1 IU/dL), 9 had moderate (factor VIII = 1–5 IU/dL), and 11 had mild (factor VIII = 5–30 IU/dL) hemophilia.
BLOOD AND PLASMA PREPARATION
Blood (9 volumes) was aseptically drawn in Vacutainer tubes (BD Vacutainer System) containing 3.2% sodium citrate (1 volume) from the antecubital vein of healthy study participants and hemophilia patients. The blood was kept at room temperature (±22 °C) and used within 4 h from withdrawal. PRP was obtained by centrifugation of blood at 240g for 15 min and platelet poor plasma (PPP) by centrifugation twice at 2630g for 10 min.
PREPARATION OF RECONSTITUTED BLOOD
Red blood cells were isolated by centrifugation of the blood at 240g for 15 min to obtain PRP. The PRP was discarded and the red blood cell suspension was resuspended in an HEPES buffer (10 mmol/L HEPES, 136 mmol/L NaCl, 2.7 mmol/L KCl, 2 mmol/L MgCl2, 0.1 weight% glucose, and 1 mg/mL BSA, pH 7.45). This procedure was repeated 3 times to assure that virtually all platelets and leukocytes were removed from the red blood cell suspension. Samples of this suspension were diluted with HEPES buffer up to different red blood cell concentrations. One volume of diluted red blood cell suspensions was mixed with 1 volume autologous PPP to obtain samples with a different hematocrit level, but with a constant amount of coagulation factors. The blood cell count and hematocrit were determined with a thrombocounter (Beckman Coulter).
TG MEASUREMENT IN WHOLE BLOOD
Paper discs of 5-mm diameter and 180-μm thickness (Whatman 589/1, Whatman GmbH) were placed in flat-bottom wells of a 96-well polystyrene microplate (Thermo Electron Corporation). Separately, 30 μL of citrated whole blood was mixed with 10 μL P2Rho (1.8 mmol/L) and 20 μL TF and CaCl2 (1.5 pmol/L and 50 mmol/L, respectively). For TG in plasma, 30 μL PPP was mixed with 10 μL P2Rho (1.8 mmol/L) and 20 μL trigger solution containing TF (3 pmol/L), PV (12 μmol/L), and CaCl2 (50 mmol/L). Immediately after mixing, we pipetted a 5-μL sample on paper disks and covered them with 40 μL of mineral oil (USB Corporation). The final concentrations for whole blood were: 50% whole blood, 0.5 pmol/L TF, 16.7 mmol/L CaCl2, and 300 μmol/L P2Rho; and for PPP were: 50% PPP, 1 pmol/L TF, 4 μmol/L PV, 16.7 mmol/L CaCl2, and 300 μmol/L P2Rho. In a parallel calibration experiment the TF-containing solution was replaced by 20 μL human thrombin calibrator (α2M-T, 300 nmol/L thrombin activity). Fluorescence signal was recorded with a Fluoroskan Ascent microplate fluorometer with λex = 485 nm and λem = 538 nm (Thermolabsystems). Samples were run at least in triplicate and the calibrated TG curves were calculated as previously described (16). All procedures were performed at 37 °C.
We measured FVIII activity using a deficient plasma kit, FVIII-deficient plasma (Precision BioLogic), on a Destiny Max hemostasis analyzer (T-Coag).
We performed statistical analyses using Graph Pad Instat 3.0 software. The Spearman statistic was applied for correlation analyses. A P value of <0.05 was considered statistically significant. The thrombograms shown are the mean of a set of 3 separate measurements. The TG parameters were calculated from the separate curves and shown as the mean (SD).
Despite the presence of red blood cells, the stability of the fluorescence signal in blood was as good as that of the signal in plasma. The mean variation of the signal during the final slope, i.e., the in situ residual activity of the α2M-T complex formed during the assay in plasma and blood was, respectively, 0.1% and 0.4%. Fluorescence tracings showing the response of the WB-CAT assay to activated whole blood had profiles that were qualitatively the same as the profiles obtained with PPP derived from the same blood sample (Fig. 1A). However, it can be seen that despite nearly identical reaction conditions, the increase of the fluorescence signal was much stronger in plasma compared with whole blood. The higher signals in the plasma experiments can be explained by the absence of the fluorescence-scattering effects of the red blood cells and the quenching effect of hemoglobin. Collectively, these results demonstrate that the TG assay can be performed in a thin porous matrix and that such matrices minimize red blood cell–induced scattering of the fluorescence readings. However, calibration of the whole blood TG assay remains an important issue because of varying hemolysis and/or hematocrit of the blood samples.
A representative fluorescence tracing (mean of 3 replicates and solid lines representing SD values) measured with the WB-CAT assay responding to the reaction between P2Rho (300 μmol/L) and α2M-T (100 nmol/L thrombin activity) in citrated blood is shown in Fig. 1B. To examine whether the fluorescence intensity changes in time deviated from linearity, the first derivative of the mean of 3 replicates was plotted vs time (dotted line). We concluded from these findings that a correction for substrate exhaustion and inner filter effect, as observed in the classical CAT method with the use of plasma as an analyte, is not required for TG measurements in a thin layer of whole blood. The linear slope of the fluorescence intensity vs time curve yields the calibration factor that converts the fluorescent signal of the raw TG curve into a molar thrombin concentration.
CONVERSION OF RAW DATA INTO THROMBIN CONCENTRATION
To obtain TG profiles, a model was fitted (Fig. 2, dotted line) to the raw fluorescence data as previously described (16). The goodness of fit was apparent from the small residuals (<1% of the raw values) that were randomly scattered around zero (data not shown). The increase in fluorescence intensity due to formation of α2M-T activity was subtracted from the total fluorescence intensity (Fig. 2, dashed line). The first derivative of this corrected fluorescence–time curve and conversion of fluorescence intensity with the calibration factor yielded a thrombogram with thrombin activity expressed as nanomoles per liter (Fig. 2, solid line). To describe the thrombogram, the following parameters were calculated: endogenous thrombin potential (ETP, expressed as area under the curve; nmol · min/L), lag time (LT, min), thrombin peak (TP; nmol/L), and time to TP (TTP; min).
OPTIMAL TF CONCENTRATION FOR MEASURING TG IN WHOLE BLOOD
We measured TG at varying TF concentrations in the presence of P2Rho (300 μmol/L) and CaCl2 (16.7 mmol/L). The thrombogram parameters as a function of the TF concentration are depicted in Table 1, in which the table values are given as the mean (SD) of 3 measurements. Maximal TG was observed with TF concentrations higher than 2 pmol/L. We recommend performing the assay with a low (0.5 pmol/L) TF concentration to also include the contribution of the intrinsic pathway factors VIII, IX, and XI. We note that in the absence of exogenous TF, a substantial amount of thrombin was produced. To investigate whether this TG is the result of preanalytical contact activation, the assay was carried out with blood collected on citrate and in the presence of an FXIIa inhibitor (i.e., 50 μg/mL corn trypsin inhibitor). A comparison of the thrombogram parameters revealed that when blood was collected on corn trypsin inhibitor, ETP decreased slightly (8%), the TP decreased by 52%, and the LT was prolonged 4-fold. Because of the artifactual and variable character of contact activation, collection of blood on citrate with an inhibitor of FXIIa is recommended when <1 pmol/L TF is used to trigger blood coagulation.
SOURCE OF WHOLE BLOOD PROCOAGULANT MEMBRANES
Thrombin generation in TF-activated plasma requires the presence of procoagulant phospholipid membranes for the assembly of thrombin-generating enzyme complexes (17). It has been shown that platelets are an essential contributor to TG in PRP that is triggered with a low amount of TF (18). However, it has also been suggested that red blood cell membranes may contribute to TG (19, 20). With the tool described in the present report, this issue can now be studied in detail.
The data shown in Fig. 3A illustrate that the addition of synthetic PV (10 μmol/L) to whole blood has little influence on TG, whereas such addition is essential for TG in PPP. This finding demonstrates that blood cells provide sufficient procoagulant phospholipids for optimal TG. The contribution of red blood cells was demonstrated by a partial restoration of TG when 1 volume of PPP was mixed with 1 volume of washed red blood cells to obtain a hematocrit of 38% (Fig. 3B). A full restoration was not expected because the plasma fraction was diluted as well, resulting in suboptimal coagulation factor levels. To discriminate between platelets and red blood cells, we added convulxin, a potent platelet activator that acts via glycoprotein VI (21), to whole blood and to PRP. Whereas in PRP convulxin had a significant stimulating effect, there was no effect in whole blood (Fig. 3C). Demonstration of the dependency of the TG on hematocrit provides evidence that red blood cells are the major source of procoagulant membranes. When TG was measured in mixtures that contained varying amounts of washed red blood cells, devoid of platelets and leukocytes, and a fixed amount of PPP, the TP values increased with hematocrit and reached a plateau value at about 14% (Fig. 3D). Addition of 10 μmol/L PV to the reconstituted blood samples normalized the TP heights in samples with a hematocrit of <14% (data not shown).
PRECISION OF THE WB-CAT ASSAY
To obtain intraassay precision data, we performed TG measurements under similar conditions (0.5 pmol/L TF, 300 μmol/L P2Rho, and 16.7 mmol/L CaCl2) with the WB-CAT method (15 replicates). The mean values and the CV of the thrombogram parameters are depicted in Table 2. Interassay precision values of the thrombogram parameters were calculated from 14 independent measurements in triplicate of a single whole blood sample. Interassay variation could not be assessed from measurements over a time period longer than 1 day because of donor differences and changes that occurred during storage of the blood sample. The mean values and the interassay CV are depicted in Table 2. The results indicate that the intraassay and interassay precision are sufficient for useful application of the assay in clinical and fundamental research.
To find the variation of thrombogram parameters in healthy individuals, we performed WB-CAT assays with blood from 40 volunteers. The mean (SD) values and CVs (between brackets) for ETP and TP were 633 (146) (23%) and 128 (23) (18%), respectively. The median (range) of the thrombogram parameters ETP and TP were 606 nmol · min/L (405–1237) and 126 nmol/L (91–220), respectively.
TG IN WHOLE BLOOD OF HEMOPHILIA PATIENTS
As a proof of concept, we activated whole blood of 27 patients with hemophilia A with 0.5 pmol/L TF and measured TG. Each blood sample was run in triplicate. The relationship of the thrombogram parameters ETP and TP height with FVIII concentrations as measured with the classical 1-stage clotting assay based on activated partial thromboplastin time are shown in Fig. 4. We found statistically significant correlation between plasma FVIII:C activity and the TP height (r = 0.62 95%; CI = 0.3–0.81; P = 0.0006) and between FVIII concentrations and ETP (r = 0.58 95%; CI = 0.25–0.79; P = 0.0014). No correlation was observed between circulating plasma FVIII concentrations and LT of WB-CAT (P = 0.17).
To establish that the precision of the assay also holds for whole blood with low FVIII concentrations (3 IU/dL), we performed replicate (n = 15) measurements and found a CV value of 10.8% for ETP and 8.0% for TP. To confirm a correlation between FVIII concentration and ETP and TP, we measured TF-driven TG in blood from a patient with severe hemophilia A (FVIII <1 IU/dL) that was spiked with increasing amounts of recombinant FVIII (Hexilate, CLS Behring). In accordance with an earlier study (24), using the plasma CAT method, we found a linear relationship between TP and FVIII concentrations up to 5 IU/dL. Saturation was seen when FVIII concentrations reached 30 IU/dL. For ETP, the saturation level was reached at FVIII concentrations higher than 5 IU/dL.
Recently, Hemker and coworkers (7) developed a method for continuous monitoring of TG in PPP and PRP. This calibrated, automated, and fluorogenic TG assay has multiple potential clinical uses, e.g., estimation of hemorrhagic and thrombotic risk, monitoring of therapy, and detection of circulating procoagulants and anticoagulants (4, 22). However, of major concern is the need for a centrifugation step to separate plasma from blood cells, which may result in uncontrolled variations in cell counts, activation of blood cells, production of platelet- and monocyte-derived microparticles, and removal of heterologous blood cell aggregates. These variables are known to differentially modulate TG (23). Therefore, measurement of TG in whole blood is likely approaching (patho)physiological conditions more closely compared to the use of PPP or PRP and opens the opportunity to measure TG in a point-of-care setting, either at home or in the hospital. Despite reports to the contrary (10), we consistently found erratic fluorescence signals with this method applied in whole blood because of variable quenching by red blood cells that sediment in the sample holder.
We here report a method that allows the real-time fluorescence monitoring of TG in whole blood. This novel method relates to the use of a thin matrix that contains blood in the interstices of its structure and the use of a rhodamine-based fluorogenic substrate that has a high quantum yield and is not consumed significantly during the time of the assay. The excitation and emission wavelengths of the rhodamine substrate have less overlap with the absorption spectrum of hemoglobin compared to the Z-Gly-Gly-Arg-7-amino-4-methylcoumarin substrate used in regular CAT experiments. As a proof of this concept, we demonstrated that the fluorescence data obtained with the WB-CAT method deviated <1% from the expected data points. The presence of red blood cells, however, significantly reduces the fluorescence signal but up to a level that is still acceptable. To convert a fluorescence signal into a time course of thrombin concentration, 3 steps are required. First, there should be correction for inner filter effect and substrate exhaustion. It appears, however, that this correction is not necessary in the present assay, in which a rhodamine 110–based thrombin substrate is used. Second, the contribution of in situ generated α2M-T activity should be subtracted from the raw data. This can be done by a mathematical procedure as described (6). Finally, the fluorescence change (dFI/dt) should be converted to thrombin concentration. This conversion was done by measuring the linear time dependency of the FI (fluorescence intensity) of a reaction with a known amount of calibrator and the thrombin substrate in a whole blood milieu. In addition, calibration is essential because of the varying quenching of fluorescence signal with red blood cell count.
Thrombin generation assays in PPP are often conducted with a rather high amount of TF (5 pmol/L). With this TF concentration less variability in ETP and peak height were reported in comparison with low TF (0.5 pmol/L). However, an important disadvantage of using a high TF concentration is the loss in sensitivity for the intrinsic pathway. That is, high TF concentrations overpower the coagulation pathway that is dependent on factors VIII, IX and XI (24). Although the optimal TF concentration for clinical studies has still to be evaluated, we used a low (0.5 pmol/L) TF concentration in the whole blood TG assay. To avoid interference caused by contact activation in whole blood, we recommend for clinical studies that blood be collected in a tube containing citrate and a FXIIa inhibitor as earlier suggested for measuring TG in plasma (25, 26).
It is of interest to note that red blood cells contribute directly to TG in whole blood, because these cells are a bigger source of procoagulant cell membranes than those generated by (thrombin) activated blood platelets. Obviously, in view of the antithrombotic action of antiplatelet drugs, it might be concluded that the effects of these drugs probably are produced by inhibiting receptor activation rather than inhibiting the formation of procoagulant cell membranes or cell fragments.
The WB-CAT assay showed acceptable intraassay imprecision of thrombin parameters when blood was triggered with 0.5 pmol/L TF. The interindividual variation of ETP and TP height were 23% and 18%, respectively. Interestingly, an earlier report showed a similar interindividual variation in plasma ETP (17.5%) (7). It is apparent that, at least in a group of healthy volunteers, the presence of blood cells does not add up to a larger individual variation of the whole blood TG parameters. We also demonstrated that the ETP and TP height do not vary with hematocrit values higher than 14%.
Our results showed a significant correlation between plasma FVIII concentrations of patients with hemophilia A and the main whole blood–thrombogram parameters (i.e., ETP and TP values). Despite an overall correlation between WB-CAT and factor concentrations, a certain variation in ETP was observed for FVIII concentrations <2 IU/dL, suggesting that the FVIII concentration is not the only determinant of TG capacity in hemophiliac patients. We have previously reported that the clinical heterogeneity of the bleeding phenotype in patients with FVIII <1 IU/dL is not associated with their FVIII concentration, but with the plasma ETP, which suggests the influence of reduced concentrations of physiological anticoagulants that regulate the coagulation system (24). The large variations in blood ETP reported here support our previous findings. These preliminary results suggest the assay is of potential interest for use in patients with hereditary bleeding disorders such as hemophilia. In addition, the whole blood TG assay can be readily developed for application at the point of care to monitor the hemostatic balance in patients with bleeding or thrombotic tendency.
In conclusion, we developed a method that accurately measures whole blood TG by use of a thin layer of blood contained in wells of a multiwell microtiter plate. In research settings, this method mimics certain aspects of the in vivo situation even better than the existing methods do. Its potential utility in clinical settings should be further investigated.
The authors thank Brigitte Chatard for her excellent technical assistance. The rhodamine-based thrombin substrate (P2Rho) was a generous gift of Diagnostica Stago.
Part of this work was presented at the XXIII ISTH Congress, Kyoto (Japan), July 25, 2011.
↵3 Nonstandard abbreviations:
- thrombin generation;
- prothrombin time;
- calibrated automated thrombogram;
- platelet rich plasma;
- whole blood CAT-based assay;
- rhodamine-based thrombin substrate;
- tissue factor;
- phospholipid vesicles;
- platelet poor plasma;
- endogenous thrombin potential;
- lag time;
- thrombin peak;
- time to thrombin peak.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: T. Lindhout, Synapse BV.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: Center for Translational Molecular Medicine, project INCOAG, grant 01C-201, the Netherlands Heart Foundation.
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
- Received for publication February 8, 2012.
- Accepted for publication May 3, 2012.
- © 2012 The American Association for Clinical Chemistry