also integrated using computer simulation models to evaluate thrombin generation. ... FVIII, FIX, and FX were performed by Fletcher Allen Hema- tology Clinic ...
Journal of Thrombosis and Haemostasis, 2: 281±288
ORIGINAL ARTICLE
Thrombin generation: phenotypic quantitation K . E . B R U M M E L - Z I E D I N S , R . L . P O U L I O T and K . G . M A N N Department of Biochemistry, University of Vermont, College of Medicine, Burlington, VT, USA
To cite this article: Brummel-Ziedins KE, Pouliot RL, Mann KG. Thrombin generation: phenotypic quantitation. J Thromb Haemost 2004; 2: 281±8.
Summary. An individual's ability to generate thrombin following tissue factor stimulus was evaluated in 13 healthy male donors in a 6-month study. Thrombin generation in whole blood collected by phlebotomy, contact pathway suppressed by the presence of 100 mg mL 1 corn trypsin inhibitor, was initiated by the addition of 5 pM tissue factor/10 nM phospholipid. Reactions were quenched at 20 min by the addition of an ethylenediaminetetraacetic acid (EDTA), benzamidine, FPRck cocktail. Thrombin generation was determined by an ELISA for thrombin±antithrombin III (TAT) complex formation. Results showed that the levels of TAT observed varied from 245 to 775 nM. Thrombin production was consistent within each individual, CVi 11.6%, but varied signi®cantly within the group, CVg 25.2%, and correlated inversely with an individual's clotting time (r 0.54, P 0.07). No correlations were individually observed between TAT and C-reactive protein, antithrombin III, factors II, V, VII, VIII, IX and X, ®brinogen and prothrombin time. However, computer simulations, which integrated each individual's coagulation factor levels using the Speed Rx method (Hockin et al., J Biol Chem 2002; 277: 18322), predicted maximum active thrombin levels (ranging from calculated values of 220±500 nM) consistent with the empirically determined values. Overall, these data suggest that thrombin generated in whole blood exclusively by tissue factor stimulation can be used as an integrative phenotypic marker to determine an individual's response to a tissue factor challenge. Keywords: assay, phenotype, thrombin.
Introduction The blood coagulation and ®brinolytic systems generate the enzymes thrombin and plasmin to catalyze the formation of the ®brin clot and its subsequent solubilization [1]. The balance Correspondence: Kenneth G. Mann, Department of Biochemistry, 89 Beaumont Avenue, University of Vermont, Given Building, Room C401, Burlington, VT 05405, USA. Tel.: 1 802 656 0335; fax: 1 802 862 8229; e-mail: kmann@zoo. uvm.edu Received 26 August 2003, accepted 15 October 2003 # 2003 International Society on Thrombosis and Haemostasis
between these processes provides a defense against excessive ¯uid loss (hemorrhage) and from inappropriate vascular occlusion (thrombosis). The focal point of the hemostatic process relates to the amount of stimulus that occurs (tissue factor) and to how much thrombin is subsequently generated. This process is regulated by complex zymogen-to-enzyme transformations and the extensive array of stoichiometric and dynamic inhibitor systems [2±4]. The vast majority of patients who suffer from venous or arterial thrombosis have hemostatic systems that fall within the `healthy' range reported by routine screening tools and factor assays, and yet thrombosis is the major cause of death in western countries [5]. In contrast, the propensity to bleed (e.g. hemophilia), while more easily diagnosed by traditional assay methodology (e.g. low factor levels), may not accurately portray an individual's pathology or instruct clinical management [6]. The hemostatic process is initiated when tissue factor either exposed or expressed binds circulating factor (F)VIIa and activates the zymogens factor (F)IX and factor (F)X to their respective serine proteases, FIXa and FXa [7±10]. FXa converts picomolar amounts of prothrombin to thrombin [11±13], which begins to activate platelets [14,15] and the transglutaminase factor (F)XIII [16], initiates release of ®brinopeptides [16,17], and activates the procofactors factor (F)V and factor (F)VIII [18,19]. This initial sequence of events requires only minute concentrations of thrombin and has been referred to as the `initiation phase' of thrombin generation. Only 0.5±2.0 nM thrombin is required during this stage of the procoagulant response [15], which occurs prior to visible clot formation. Clot formation, which occurs at approximately 10 nM thrombin [11,15], coincides with the onset of the major burst of thrombin generation that is referred to as the `propagation phase' of thrombin generation. It is during this phase that the majority of thrombin ( 96% of maximum level) is generated. The onset of the propagation phase coincides with the intrinsic factor tenase complex (FIXa±FVIIIa±membrane±Ca2) being the principle generator of FXa [3] that subsequently activates most prothrombin through the prothrombinase complex (FXa±FVa± membrane±Ca2). The end points of standard clotting assays [prothrombin time, activated partial thromboplastin time (APTT)] occur when 4% of total thrombin is generated. Therefore, these assays, while informative in identifying gross hemostatic defects, do not evaluate the magnitude of the hemostatic, thrombin
282 K. E. Brummel-Ziedins et al
response. A limited repertoire of functional and genetic laboratory assays are currently available for risk assessment. These include FV Leiden [20,21], prothrombin G20210 [22,23], the FXIIIVal34Leu polymorphism [24], FVIII measurements [25], activated protein C resistance assays [26,27] and ®brinogen [28]. These assays, while informative, provide no integrated evaluation of potential stimulus±response coupling. More detailed pro®ling assays include the thromboelastogram [29], the thrombogram or endogenous thrombin potential [30,31], the corn trypsin inhibitor (CTI) whole blood model [11], and integrative computational analysis [3,32]. These assays have potential utility in the realm of phenotype discrimination but require well-trained individuals and complex instrumentation. At present, no assay is clinically instructive with respect to pharmacological intervention in the absence of a clinical thrombotic event. Thrombin generation utilizing the whole blood model developed by this laboratory analyzes the tissue factor pathway of coagulation without the interference of the contact pathway and pro®les thrombin generation in a near physiological milieu [33± 39]. During our comprehensive studies of healthy individuals, we noted that individual volunteers studied multiple times over the course of 60 months showed similar tissue factor stimulus± thrombin production response curves [15]. The current report presents a simpli®ed assay for evaluating an individual's tissue factor stimulus±response coupling which may prove useful in clinical diagnosis. Coagulation pro®les from individuals are also integrated using computer simulation models to evaluate thrombin generation. These computer-assisted model systems have been used in a number of settings which include: comparisons with results from equivalent synthetic mixtures of coagulation proteins [3]; predictive responses to tissue factor challenges in neonatal blood [40]; evaluation of treatment of FVII-de®cient patients [41]; and investigation of a number of thrombin inhibitors [42]. In this report we evaluate an individual's tissue factor±thrombin response utilizing a simpli®ed whole blood model and computer simulations. Materials and methods Materials
HEPES, Tris±HCl, EDTA, tri¯uoroacetic acid (TFA), brain phosphatidyl serine (PS), egg phosphatidyl choline (PC) were purchased from Sigma Chemical Co. (St Louis, MO, USA). Benzamidine±HCl was purchased from Aldrich, Inc. (Milwaukee, WI, USA). Recombinant tissue factor was a generous gift from R. Lundblad and S.-L. Liu (Hyland Division, Baxter Healthcare Corp, Duarte, CA, USA) and was relipidated in PCPS (25% PS, 75% PC) vesicles by a previously described protocol [43,44]. Corn trypsin inhibitor (CTI) was prepared as described [33]. D-phenylalanyl-L-prolyl-L-arginine chloromethyl ketone (FPRck) and biotinylated FPRck were a gift from R. Jenny (Hematologic Technologies, Essex Junction, VT, USA). An ELISA was used to estimate thrombin±antithrombin III (TAT) complex formation (Behring, Westwood, MA, USA).
C-reactive protein (CRP) analyses were performed by E. Macy (Department of Pathology, University of Vermont) using a previously described ELISA [45]. Correlations between TAT and protein factor levels were performed by K. Reinier (Biostatistics Department, University of Vermont). Subjects
Thirteen healthy male subjects, mean age 35 13 years (range 22±60 years) were recruited and advised according to a protocol approved by the University of Vermont Human Studies Committee. All donors had no history of thrombosis/hemorrhage, regular aspirin or drug use. Individuals were studied using the whole blood assay once a month (between 4 and 5 weeks) for 6 months, between 08.30 and 13.00 h. No speci®c limits were provided regarding diet or behavior. Coagulation pro®les including ®brinogen, antithrombin III, prothrombin, FV, FVII, FVIII, FIX, and FX were performed by Fletcher Allen Hematology Clinic (Burlington, VT, USA). Whole blood coagulation
An evaluation of our existing comprehensive model led us to choose 20 min as a discrimination point. This corresponds to a point in healthy individuals well into the plateau of thrombin generation (see [15] for a comprehensive pro®le). Blood collected at the Clinical Research Center (Burlington, VT, USA) by venepuncture was aliquoted (1 mL) into tubes containing CTI (100 mg mL 1) and tissue factor (functionally 5 pM) relipidated in PCPS (1 : 2000 protein/lipid) as previously described [11,17,34]. The same stocks of tissue factor and CTI were used for the duration of the 6-month study. A control tube containing CTI and no tissue factor was used for each experiment. Bloodcontaining tubes were allowed to rock at 37 8C for 20 min (18 cycles min 1) and then quenched 1 : 1 with a cocktail of inhibitors (50 mM EDTA and 20 mM benzamidine±HCl in HBS, pH 7.4 plus 10 mL of 10 mM FPRck in 10 mM HCl), the latter freshly prepared. Clot times were determined visually by two observers. After quenching, samples were centrifuged for 15 min at 1100 g and the clot was separated from the solution phase. Solid and solution phases were stored at 145 8C prior to further analysis. All samples collected during the 6-month study were stored until the completion period of the study and analyzed simultaneously. Thrombin generation
TAT, as an integrator of thrombin formed, was determined using a commercial ELISA from Behring. The limit of detection is 40 pM. A standard curve was generated using four concentrations of standards run in duplicate. Whole blood samples were diluted 1 : 1000 and run in duplicate following the manufacturer's protocol. An optical density reading was obtained on a TMax microtiter plate reader equipped with SOFTMAX version 2.0 software (Molecular Devices, Menlo Park, CA, USA). Concentration results were obtained using a comparison with # 2003 International Society on Thrombosis and Haemostasis
Thrombin generation/phenotypes 283
absorbance at 490 nm and ®t to a four-parameter equation {y (A±D)/[1 (x/C)B] D}. Parameters A±D are determined by the SOFTMAX software from the generated standard curve. Computational analyses of thrombin generation
Computationally generated active thrombin pro®les are obtained utilizing a software package termed Speed Rx [3]. This computational model portrays prothrombin-to-thrombin conversion pro®les, represented as active thrombin, as well as the progress of all reactants, intermediates and products. It is composed of 10 reactants, the procoagulants tissue factor, FVII/ VIIa, FIX, FX, prothrombin, FV and FVIII, and the anticoagulants, tissue factor pathway inhibitor and antithrombin III [3]. The combination of 34 rate equations, 42 rate constants and 27 equilibrium expressions for the 10 reactants describes the processes of reagent consumption and product formation. Speed Rx utilizes an internet-based interface with a generally applicable fourth order Runge±Kutta solver that provides solutions to a family of time-dependent differential equations. The procoagulant and anticoagulant levels for the 10 reactants were obtained for each individual and entered into the computer database and coagulation was initiated with a 5-pM tissue factor stimulus (same as whole blood stimulus) and solved for active thrombin. Active thrombin is a combination of thrombin and meizothrombin weighted according to meizothrombin's activity towards synthetic thrombin substrates. The simulations are run for 1200 s and are evaluated by the maximum level of active thrombin and the time to reach 10 nM thrombin (approximate clot time, based on whole blood data [11]). Statistical analysis
Analyses of variance (ANOVA) were performed on TAT generation (for n 13 subjects) and CRP levels [46,47]. The coef®cient of variation (CV) for TAT and CRP were calculated for: analytical variation (CVa), intra-individual variation (CVi), and inter-individual variation (CVg). The index of individuality,
how individuals vary relative to the population distribution, was calculated as the ratio CVi/CVg. Correlations between TAT and CRP, age, antithrombin III, factor (F)II, FV, FVII, FVIII, FIX, FX, ®brinogen, protime, and clot time were evaluated using Spearman correlation and presented as r and P-values. Comparisons were made using the Wilcoxon rank sum tests between the 20-min TAT level for the cumulative 13-individual dataset (repeated measurements on individuals were averaged) and the historical 20-min TAT levels for healthy individuals, individuals on coumadin therapy, individuals with hemophilia A and B with and without replacement, factor (F)XI-de®cient individuals and the effects of platelet IIbIIIa inhibitors. Differences between the means of the groups are expressed as P-values. Results This study was conducted over a time period of 6 months with 13 male volunteers. Nine out of the 13 individuals had blood collected for all six experiments. Three of the 13 individuals (subjects 6, 8, and 9) missed one blood draw each during the course of the study. One individual (subject 2) had blood collected three out of the six experiments. Factor levels
Coagulation factor analyses for antithrombin III, FII, FV, FVII, FVIII, FIX, FX and ®brinogen for the 13 individuals are shown in Table 1. The normal values reported by the Coagulation Laboratory at Fletcher Allen Health Care are shown at the bottom of Table 1. All subjects fell within normal values for FII, FV, FX, ®brinogen and APTT (28.8 3.9 s). Several individuals had either slightly high or low factor levels as indicated: subject 8 had low antithrombin III levels of 72%, subject 1 had high normal antithrombin III levels with 129% (normal 86±128%), subject 13 was on the low normal side for FVII with 59% (normal 60±140%), subject 1 had low FVIII levels of 55% (normal 64±232%), and subject 10 had low FIX levels of 51% (normal 69±151%).
Table 1 Factor levels of subjects studied over 6 months Subject
AT III (%)
Factor II (%)
Factor V (%)
Factor VII (%)
Factor VIII (%)
Factor IX (%)
Factor X (%)
Fibrinogen (mg/dL)
Protime (s)
1 2 3 4 5 6 7 8 9 10 11 12 13 Normal
129 94 116 112 90 104 115 72 104 114 111 116 95 86±128
110 99 125 96 106 84 114 93 92 87 96 78 78 60±140
96 90 115 131 86 97 92 74 77 85 83 87 77 60±140
91 78 108 98 81 78 99 88 78 81 101 87 59 60±140
55 70 206 99 118 146 67 96 126 77 80 77 170 64±232
144 82 109 88 116 86 101 71 69 51 79 80 67 69±151
114 107 114 97 93 91 132 86 85 85 88 99 72 60±140
411 220 238 246 380 274 250 198 164 174 242 203 290 180±433
13.6 13.8 12.3 12.7 12.9 14.6 13.3 14.1 14.1 12.8 13.0 13.0 14.1 11.7±13.1
Normal levels according to Coagulation Laboratory at Fletcher Allen Health Care, Burlington, VT, USA.
# 2003 International Society on Thrombosis and Haemostasis
284 K. E. Brummel-Ziedins et al Table 2 Clot times Subject
Month 1
Month 2
Month 3
Month 4
Month 5
Month 6
Mean SD
1 2 3 4 5 6 7 8 9 10 11 12 13
4.2 5.7 4.8 4.1 5.7 4.8 8.0 nd 6.8 3.5 6.0 5.8 4.3
5.9 nd 4.8 4.4 6.2 5.1 6.2 5.5 7.0 5.4 6.8 5.2 5.0
5.8 6.4 6.2 6.5 6.3 6.4 7.9 5.0 8.0 6.6 5.4 6.1 5.0
6.1 nd 5.9 6.5 6.5 nd 5.9 4.4 9.1 7.5 4.9 5.8 6.2
6.0 5.3 3.6 5.0 6.5 6.0 7.1 4.6 nd 7.2 5.7 5.2 6.0
6.8 nd 6.2 5.5 5.3 5.4 6.3 5.9 7.2 6.0 5.4 6.0 4.4
5.8 0.9 5.8 0.6 5.2 1.0 5.3 1.0 6.1 0.5 5.6 0.7 6.9 0.9 5.1 0.6 7.6 0.9 6.0 1.4 5.7 0.6 5.7 0.4 5.1 0.8
nd, Not determined.
Whole blood clot time
The visual clot times for the 13 individuals over the course of 6 months are shown in Table 2. Control CTI tubes that did not contain tissue factor clotted at >20 min (studies extending beyond 20 min may have breakthrough contact activation). The mean and standard deviation for the whole population clot time was 5.8 1.0 min. There was little variability between clot times and draw date, with subjects 3, 4 and 10 varying the most with clot times of 5.2 1.0, 5.4 1.0 and 6.0 1.4 min, respectively. Subject 9 consistently had long clot times of 7.6 0.9 min. A negative correlation was identi®ed between an individual's clot time and ultimate thrombin levels (r 0.54, P 0.07). Thrombin generation at 20 min
The thrombin values (as TAT at 20 min) encompass a broad range (245±775 nM) with a cumulative mean of 414.2 111.2 nM. In relation to clot time, subject 9 and subject 10 had relatively low levels of TAT at 20 min as well as longer clot times. Subject 9 had a mean TAT level at 20 min of 401.6 31.1 nM and a mean clot time of 7.6 0.9 min. Subject 10 had a mean thrombin level of 270.6 19.6 nM and a mean clot time of 6.0 1.4 min. The mean values (SD) for each individual are expressed in Fig. 1 while the individual variations over the course of the study in TAT levels are seen in Fig. 2. These results show that there was a general propensity for any given individual to respond with a de®ned level of thrombin for a constant tissue factor stimulus. Levels ranged from 270.6 19.6 nM TAT (subject 10) to 629.6 80.8 nM TAT (subject 3). Subject 6 (*) and subject 3 (~) varied the most, having mean TAT levels of 625.8 131.0 nM and 629.6 80.8 nM, respectively. The two subjects with the lowest levels of TAT varied the least: 270.6 19.6 (subject 10, ^) and 312.4 22.3 (subject 5, ). Statistical analysis ANOVA on the thrombin data shows that our duplicate measurements using the TAT assay kit had an analytical variance
Fig. 1. One-point measurement of thrombin levels. Thrombin as assessed by thrombin±antithrombin III complex formation (TAT) for 13 individuals is shown as a bar graph of the mean SD for 6 months. Subjects 1±13 are labeled below the appropriate bar. Subject 2 is the mean of only three times and subjects 6, 8 and 9 are the mean of ®ve times. X 431.1 nM, analytical variation (CVa) 7.0%, intra-individual variation (CVi) 11.6%, inter-individual variation (CVg) 25.2%, Index of Individuality 0.46, Index of Heterogeneity 1.7, and Critical Difference 37.6%
Fig. 2. Six-month individual thrombin pro®le. Thrombin generation as determined by thrombin±antithrombin III (TAT) at 20 min is illustrated for the 13 individuals over the course of a 6-month time period. Subject 1 (^), subject 2 (&), subject 3 (~), subject 4 (&), subject 5 ( ), subject 6 (*), subject 7 (~, dashed line), subject 8 ( , dashed line), subject 9 ( ), subject 10 (^), subject 11 (&, dashed line), subject 12 (~), and subject 13 (^, dashed line). The mean TAT level of the 13 healthy individuals was 414.2 111.2 nM. # 2003 International Society on Thrombosis and Haemostasis
Thrombin generation/phenotypes 285
CVa(%) 7.0. Over a 6-month time period in this group of healthy individuals, there was little individual variation between monthly blood draws CVi(%) 11.6, whereas within the population individuals did vary from one to another CVg(%) 25.2. The index of individuality is mid range (CVi/CVg) 0.46. Therefore, the amount of prothrombin-to-thrombin conversion in healthy individuals appears to be relatively consistent within individuals and extends over a broad range within the population. These results suggest that individual variations in thrombin generation can potentially be detected upon hemostatic challenges. A marker for in¯ammation, CRP, was evaluated in each individual over the course of the study [45,48]. ANOVA results on CRP levels show an analytical variance of CVa(%) 14.3, and variability within an individual CVi(%) 53.6 and between individuals CVg(%) 27.4. The index of individuality 2.0. Two of the subjects presented with a cold when their blood was drawn and their CRP level for that draw was increased. Subject 4, month 4 had a CRP 3.5 mg mL 1 and subject 13, month 6, had a CRP 3.1 mg mL 1 (normal values 0.8±1.0 mg mL 1). No correlative effect was seen between thrombin generation (as TAT) and increased CRP levels. Therefore, individual variations in thrombin levels (i.e. subject 6) could not be attributed to any changes in in¯ammatory response (based upon CRP levels). Correlation analyses between coagulation parameters and TAT levels for 12 participants with subject 6 removed were performed (subject 6 was left out of the correlations because of the variability in his TAT level over the 6-month study). An inverse correlation was observed between clot time and TAT level (r 0.54 and P 0.07). No signi®cant correlations were observed between CRP, antithrombin III, FII, FV, FVII, FIX, FX, FVIII, ®brinogen and protime with TAT levels using a Spearman correlation. Computer simulations of tissue factor-induced thrombin generation
The plasma analyte levels for each of the 13 individuals (values shown in Table 1) were used to provide numerical analyses for each person's hemostatic response using the Speed Rx method [3]. Active thrombin (mM) vs. time showed maximum active thrombin values ranging from 220 to 500 nM. Subject 3 ( ) had the highest whole blood thrombin generation TAT value of 629.6 80.8 nM, and also showed the highest predicted level of active thrombin, 500 nM, using the numerical model (Fig. 3). The lowest thrombin generator, subject 10 (~) with a TAT value of 270.6 19.6 nM, displayed a maximum active thrombin level of 220 nM in the numerical model. Subject 13 (&), an individual with intermediate levels of thrombin generation, had TAT levels of 500.4 38.6 nM and predicted intermediate maximum active thrombin levels of approximately 350 nM. A theoretical clot time, or duration of the initiation phase, can be obtained from these simulations by determining the time at which 10 nM thrombin is generated [11]. The estimated clot times for subjects 3, 13 and 10 are as follows: 320 s (5.3 min), 460 s (7.7 min) and 530 s (8.8 min). These times are similar in # 2003 International Society on Thrombosis and Haemostasis
Fig. 3. Computer simulation of thrombin generation. Three individuals were modeled for active thrombin generation based upon their protein factor levels found in Table 1. Subject 3, 10 and 13 re¯ect the low, mid and high range of thrombin generation in the tissue factor-based whole blood model. Using a computer-simulated model called Speed Rx [3], the same individual pattern of low mid and high thrombin generation was obtained. Subject 3 ( ) had the highest thrombin levels of approximately 550 nM followed by subject 13 (&) with approximately 350 nM and subject 10 (~) with only 250 nM.
order when compared with the observed whole blood clot times for these individuals. Therefore, the individual parameters obtained in a coagulation pro®le can also be applied in a predictive manner using computational simulations. However, unlike the whole blood studies, the present computational models do not take the cellular contributions into account [3]. The 20-min thrombin level as evaluator of tissue factor stimulus±response coupling
In the population studied, a wide range (245±775 nM) of thrombin was observed in the tissue factor-initiated reaction. Low levels of total thrombin (
