What is all that thrombin for?

Abstract

Summary.  The hemostatic process initiated by the exposure of tissue factor to blood is a threshold limited reaction which occurs in two distinct phases. During an initiationphase, small amounts of factor (F)Xa, FIXa and thrombin are generated. The latter activates the procofactors FV and FVIII to the activated cofactors which together with their companion serine proteases form the intrinsic FX activator (FVIIIa-FIXa) and prothrombinase (FVa-FXa) which generate the bulk of FXa and thrombin during a propagation phase. The clotting process (fibrin formation) occurs at the inception of the propagation phase when only 5-10 nM thrombin has been produced. Consequently, the vast majority (greater than 95%) of thrombin is produced after clotting during the propagation phase of thrombin generation. The blood of individuals with either hemophilia A or hemophilia B has no ability to generate the intrinsic FXase, and hence is unable to support the propagation phase of the reaction. Since clot based assays conclude before the propagation phase they are not sensitive to hemophilia A and B. The inception and magnitude of the propagation phase of thrombin generation is influenced by genetic polymorphisms associated with thrombotic and hemorrhagic disease, by the natural abundance of pro- and anticoagulants in healthy individuals and by pharmacologic interventions which influence thrombotic pathology. Therefore, it is our suspicion that the performance of the entire process of thrombin generation from initiation through propagation and termination phases of the reaction are relevant with respect to both hemorrhagic and thrombotic pathology.

Introduction

Blood loss through lack of hemorrhage control has captured the attention of individuals from all walks of life and throughout history. The significance of hemorrhage is discussed in the earliest known literature [1]. In the mid-19th century the clotting of fibrinogen by thrombin was accurately described by Schmitt [2], and since that time hemostasis, defined by Stedman's Dictionary[3] as the control of hemorrhage, has been synonymous with blood clotting.

The pursuit of knowledge of natural events can be divided into the categories of inventory, connectivity and dynamics. The development of plasma clot-based assays in vitro, in which the observation of fibrin strands is the endpoint has played a central role both in describing the inventory and in establishing the logical connectivity between the reactants and reactions relevant to the hemostatic process. Morowitz [4] formalized the relationship between the introduction of tissue components into plasma and the formation of a clot. This reaction was exploited quantitatively by Quick [5] in the development of the prothrombin time assay. In its current state, this series of reactions (represented in Fig. 1) includes additional components now known to be required to produce the fibrin endpoint effectively. Langdell and colleagues [6] subsequently explored the spontaneous clotting of recalcified citrate plasma and developed the partial thromboplastin time, thus expanding the inventory catalog and further defining the connectivity of the intrinsic pathway of coagulation [Fig. 1]. The connectivity of this expanded system was defined by the cascade/waterfall paradigms of MacFarlane [7], and Davie and Ratnoff [8]. Intersections connecting these pathways were extended by the work of Österud and Rapaport [9], and Galiani and Broze [10], who identified transactions that explicitly linked the two classical pathways leading to thrombin generation and a fibrin clot. The apparent absence of kinetic efficiency in the tissue factor (TF) activation of factor (F) VIIa was solved by Lawson et al. [11], who showed that FXa–membrane could cooperate in the FVIIa–TF activation of FIX.

Figure 1. Inventory and connectivity of the reaction complexes that lead to the generation of thrombin. Each complex catalyst is illustrated with its participating serine protease and cofactor protein. For the vitamin K-dependent catalysts, an anionic phospholipid-like membrane is required for assembly and expression of activity. Various thrombin substrates are illustrated as sidebars to the complexes.

The title selected by the organizers for this presentation is based upon activities championed by our laboratory and Hemker's, which have explored the dynamics of the total generation of thrombin and the significance of this process in maintaining hemorrhagic and antihemorrhagic qualities from the physiologic and pathologic perspectives [12,13]. In vitro, blood and plasma clot when only a tiny fraction of prothrombin is converted to thrombin, and a great deal of hemostatic and thrombotic physiology and pathology is not captured by the fibrin clotting endpoints used commonly to evaluate the hemostatic process. In short, it appears that hemostasis is not synonymous with the endpoint of the fibrin-clotting reaction, and the latter is not a sufficient descriptor of the pathology associated with errors in the hemostatic process or the potential for development of a thrombotic occlusion.

We begin this review with a disclaimer with respect to the title of this manuscript: we do not know ‘what all the thrombin is for’, but we do believe that the hemostatic process cannot be adequately evaluated by a simple clotting endpoint. In this review, we provide: (i) a description of the dynamics of the TF pathway to thrombin from a mechanistic standpoint; (ii) a collection of our experiences dealing in thrombin generation in hemorrhagic and thrombotic situations; and (iii) some potential approaches to the anticipation of hemostatic and thrombotic responses by using an integrated approach to laboratory analysis.

The central importance of thrombin

Thrombin is a multifaceted protein with functions extending from coagulation activator and inhibitor to cellular regulator. Its central importance in biology, physiology, and pathology is underscored by studies conducted in genetically homogenous transgenic mice made deficient in components of TF pathway that are essential to the thrombin generation process and its regulation: TF [14], FVII [15], TF pathway inhibitor (TFPI) [16], FX [17], FV [18], prothrombin [19], and protein C (PC) [20]^−/− mouse constructs have been reported. In the mouse populations studied, deletion of these key components are lethal, indicating that when evaluated in a homogenous genetic background, thrombin formation/regulation is essential to life. FVIII^−/−[21] and FIX ^−/−[22] mice incur hemorrhagic risk. In contrast, in the outbred human population, total deficiencies in coagulation components that are lethal in mice may yield consequences ranging from mild to extreme pathology. We hypothesize this is a consequence of human genetic heterogeneity that provides alternative pathways that can achieve survival in spite of potentially lethal abnormalities. In this hypothesis we assume that the majority of embryos congenitally deficient in the ‘essential’ components of the TF pathway do not survive.

Plasma clot-based assays have been the endpoints for the most commonly used tests of the hemostatic process. The prothrombin time (PT) and activated partial thromboplastin time (APTT) reflect, however, only the contributions of the diluted plasma component of the hemostatic system in adulterated form. Both tests terminate with endpoints that occur with less than 5% of the reaction complete. Fibrin clotting thus occurs when only minimal levels of prothrombin have been activated. It is also somewhat paradoxical that while we use the terms blood coagulation and hemostasis somewhat interchangeably, fibrinogen deficiency in mouse [23] is frequently only mildly symptomatic, in marked contrast to those deficiencies that negate thrombin formation. While clotting tests have been especially useful in identifying congenital abnormalities associated with hemophilia A, B and C, and in the evaluation of oral anticoagulant therapy, they overlook most thrombin formation, and their utility in evaluating many anticoagulant therapies [24] has been somewhat limited. In addition, prolongation of the endpoint in the APTT may reflect genetic deficiencies with little consequence for hemorrhagic pathology in the host.

The tissue factor pathway

Most investigators believe that the generation of thrombin via the TF pathway is the biologically relevant process by which thrombin is elaborated and hemostasis is achieved. The reaction begins with the expression/exposure of TF, which is maintained in an inactive form either by compartmentalization or by some regulatory process that permits expression of this membrane-bound receptor protein [25–29]. The function of TF is expressed by binding pre-existent plasma FVIIa, which is present at approximately 1–2% of the total FVII zymogen concentration [30]. FVII competes with FVIIa for TF binding, thus serving as a negative regulator in the overall reaction [31]. When bound to TF, FVIIa function is efficiently expressed toward its macromolecular substrates, FIX and FX [9,32–34], with the latter being the more efficient and abundant substrate. The resultant initial FXa with a fraction of active membranes converts prothrombin to thrombin, albeit inefficiently. FXa also contributes to factor IX activation by cleaving the zymogen to the intermediate FIX [11]. The small amounts of thrombin produced by FXa–membrane initiates platelet activation [35], and activates minute amounts of plasma FV and FVIII to the cofactor forms FVa and FVIIIa [36]. These two cofactors ultimately form the receptor sites, both locating and activating FXa and FIXa, and forming the intrinsic FXase and prothrombinase complexes on the activated platelet surface [37]. Each membrane-bound, vitamin K-dependent enzyme complex [Fig. 1] is 10⁴−10⁶ more active than the respective proteases towards their macromolecular substrates in solution [38].

In addition to the feedback activation by thrombin to activate FV and FVIII, other feedback steps accentuate the overall process of catalyst generation. FXa–membrane can also activate FV and FVIII to their respective cofactors [39], although the biological relevance of these processes is suspect considering the small amounts of FXa available and competitive substrates. Thrombin also activates FXI to FXIa [10], initiating an accessory pathway that enhances FIX activation [40].

The hemostatic reaction is under the control of both stoichiometric and dynamic inhibitory systems. TFPI is the principal inhibitor of the extrinsic FX activator [41,42]. TFPI is a high-affinity, low-abundance inhibitor present in plasma and secreted by vascular cells contributing to the local anticoagulant environment of the vascular wall. The major stoichiometric inhibitor of thrombin and its generation is antithrombin III [45]. FXa and IXa in complex display decreased reactivity with antithrombin III [43]. In contrast, the FVIIa–TF complex shows increased reactivity compared with FVIIa in plasma [44,45]. In plasma, FVIIa is almost impervious to inhibition by this ubiquitous serpin, and this lack of reactivity permits the existence of FVIIa in the hostile, inhibitor-rich blood environment. Antithrombin III is present at over twice the concentration (3.2 µmol L⁻¹) of the highest potential procoagulant enzyme concentration (thrombin) (1.4 µmol L⁻¹). This serpin effectively neutralizes all the serine proteases associated with the hemostatic process. Other serpins, including heparin cofactor II, may also contribute inhibitory capacity, suppressing the procoagulant proteases [46,47].

In addition to being an effective feedback procoagulant activator, thrombin is also an effective feedback anticoagulant. The enzyme binds to vascular cell-associated thrombomodulin [48]. The complexed thrombin effectively recognizes the vitamin K-dependent zymogen PC and ceases to be an activator for the procofactors and fibrinogen. The activated PC (APC) product interferes with the coagulation system by binding FVIIIa and FVa in competition with FXa and FIXa and proteolyzing the two cofactors, leading to their destruction [49,50]. The inactivated membrane-bound factor FVai and FVIIIai suppress APC function by continuing to effectively bind the enzyme, thus serving as product inhibitors.

The overall expression of thrombin function is a tightly regulated, highly intercalated system, the performance of which cannot be anticipated with the simple ‘guesstimate’ analyses that investigators typically apply to simple reactant systems. It is the composite of qualitative and quantitative features of the overall system that dictates the ultimate process, including the rates, extents of formation, and durabilities of catalysts and thrombin [51]. In addition, many products occur transiently during the course of the reaction, leading to increased complexity. Even alterations of plasma-protein concentrations in the ‘normal range’ observed in the human populations can have an extraordinary effect on the ultimate amount of procoagulant activity generated for a given level of TF stimulation of the system [52]. Even dilution of blood or plasma produces significant and often unexpected changes. ‘Seat of the pants’ speculation of how these reactions will be altered by changes in reactant concentrations, solvent conditions or temperature are inherently dangerous and have contributed to misinterpretations of the biological processes.

Models of hemostasis

In attempting to examine how the TF pathway works in generating thrombin, our laboratory has evaluated four models to try to describe the dynamics of the thrombin generating system. These are: (i) synthetic plasma mixtures prepared using purified proteins and natural or synthetic membranes induced to react by the addition of lipid-reconstituted TF [52,53]; (ii) numerical models of the coagulation system based upon reaction-rate constants, concentrations and mechanisms [51,54,55]; (iii) para vivo studies involving whole blood induced to clot purely by a TF stimulus [11,56,57]; and (iv) in vivo studies in blood exuding from a microvascular wound [58–60].

Each model has its advantages and disadvantages. Models that involve human volunteers are constrained by both ethical and technical considerations. Synthetic and computer models are used to anticipate knowledge of the true biology of the process and aid in design of both para vivo and in vivo experiments. When the models converge with in vivo observations, the appropriateness of the approximations is assured.

Thrombin generation

In all models, the TF-initiated display of thrombin generation is approximately the same. This behavior, illustrated in Fig. 2, may be operationally described as occurring in two phases. At first, tiny (nanomolar) amounts of thrombin are produced during an interval (the initiation phase). The major bolus (>96%) of thrombin is produced secondarily during a propagation phase. During the initiation phase, the FVIIa–TF complex forms, and generates sub-picomolar amounts of FXa and FIXa. FXa, in collaboration with the membrane surface, activates a small amount of prothrombin to thrombin, which serves to generate the platelet membrane and cofactor components required for the major generation of thrombin. These autocatalytic processes lead to increased catalyst formation.

Figure 2. Thrombin generation during para vivo whole-blood experiments. Thrombin–antithrombin III (TAT) complex formation is plotted as a function of time for sequential samples of human blood at 37 °C inhibited with respect to contact pathway activation and activated by the addition of 5 pmol L⁻¹ tissue factor, 10 pmol L⁻¹ phospholipid. The data (± SEM) are presented for 35 individual experiments with an average clot time of 4.7 ± 0.2 min. The operationally defined initiation and propagation phases correspond to the slow and fast kinetic expressions of thrombin formation. The propagation phase is invisible to clot-based assays. From Brummel et al.[58], by permission.

Signal events occurring during the initiation phase are illustrated in Fig. 3, which shows the inception points for the detection of thrombin products generated during the reaction measured in para vivo experiments [57]. These products provide the elements of the catalysts (Fig. 1) that generate the majority of the thrombin produced during the propagation phase of the reaction.

Figure 3. The onset of detection of products from thrombin substrates during the initiation phase for the experiments presented in Fig. 2. Active thrombin and thrombin–antithrombin III (TAT) concentrations are plotted on the vertical axis on an exponential scale vs. time. The onset of product detection corresponding to platelet activation (OSN) and other well-established thrombin substrates are depicted. The inception of propagation phase corresponds to the point at which there is a transition from slow to rapid thrombin generation. From Brummel et al.[58], by permission.

The cleavage of fibrinopeptide (FP) A and subsequent clot formation (Figs. 2 and 3) occur just prior to the propagation phase of the reaction. Under normal conditions, the activation of platelets and FV occurs rapidly to produce a surplus of FVa and platelet-membrane binding sites, leaving the rate-limiting reagent for prothrombinase formation as the concentration of FXa. However, with congenital deficiencies, thrombocytopenia, platelet pathology or pharmacologic interventions, the reaction can become sensitive to FV or platelets [61].

The endpoint utilized in evaluating hemostasis in most bioassays is the generation of a fibrin clot. As illustrated in Figs 2 and 3, in largely unadulterated whole blood at 37 °C, the formation of a visible fibrin clot occurs at 10–30 nmol L⁻¹ thrombin, or ∼3–5% of the total amount of thrombin produced. This thrombin in turn is provided by only ∼7 pmol L⁻¹ prothrombinase [12,57]. Thus, most catalyst and thrombin formation is undetected by current technology for evaluating clinical hemorrhagic risk or thrombosis.

Figure 4 illustrates the time course of removal of fibrinogen and fibrin products from the fluid phase of blood and the formation of products within the insoluble clot. This figure should be compared with the data of Figs 2 and 3 to register the formation of thrombin with the cleavage of fibrinogen and the formation of the fibrin clot. In Fig. 4A, at the point of visual clot formation (CT), virtually all fibrinogen (and some product already crosslinked) disappears from the fluid phase of the reaction [62]. At this point ∼50% of the FPA has been cleaved, thus the ‘clot’ is a mixture composed of a mixture of fibrin 1 and fibrinogen. The insoluble material present in the fibrin clot (panel B) is virtually all cross-linked by FXIIIa, the activation of which is nearly simultaneous with FPA removal (Fig. 3). In purified systems it has been observed that FPB removal precedes the cross-linking reaction, however, as seen in the α-FPB immunoblot in Fig. 4C, the B peptide antigen epitope is still detectable associated with the Bβ chain.

Figure 4. Fibrinogen and fibrin products during the tissue factor-induced clotting of whole blood. Panel A reflects the immunoblot data for sequential aliquots taken from the solution phase of contact pathway inhibited whole blood induced to clot with 5 pmol L⁻¹ tissue factor–phospholipid. The label ‘CT’ corresponds to the clotting time observed visually for the experiment. Panel B illustrates a Coomassie Blue-stained gel of the reduced insoluble clotted material for the same experiment in Panel A. Panel C illustrates an immunoblot of the gel of panel B using an antibody (supplied by B. Kudryk) specific for the B peptide region of the Bβ chain. From Brummel et al.[63], by permission.

The significance of intrinsic factor Xase

During the transition between the initiation and propagation phases, increased concentrations of the FVIIIa–FIXa complex are generated, contributing an increasing concentration of FXa. The TFPI downregulation of FXa formation by the FVIIa–TF complex and the enhanced efficiency (∼50-fold) of the FVIIIa–FIXa complex effectively switch the primary path of FXa production to the latter catalyst. Figure 5 illustrates a numerical analysis of the percentage formation of FXa by the two complexes during the progress of the reaction. Shortly before the propagation phase of thrombin generation is observed, the majority of FXa begins to be produced by the intrinsic FXase. Operationally, the onset of the propagation phase that signals enhanced thrombin generation is coincident with the intrinsic FXase being the principal generator of FXa.

Figure 5. A numerical estimation of the percentages of FXa produced by the intrinsic FXase and by the extrinsic FXase. Initially 100% of FXa is generated by FVIIa–tissue factor. However, as the reaction progresses, the major FXa production is contributed by the more efficient FIXa–FVIIIa catalyst. The arrow indicated the approximate clotting time. Modified from Hockin et al.[52], by permission.

From these observations one would conclude that in congenital hemophilia A and B a significant deficit in thrombin generation during the propagation phase would occur. This is in fact observed in all models, most significantly in para vivo studies of individuals with hemophilia A and B [Fig. 6]. The blood of these individuals displays a slight prolongation of the time to form a clot, but the major impairment is in thrombin generation during the propagation phase of the reaction. The deficit in hemophilia C, or FXI deficiency, is also observable as a defect in the propagation phase generation of thrombin; however, this observation can only be made when the reaction is initiated by extremely low concentrations of TF with clotting times >10 min [56].

Figure 6. Thrombin generation represented as thrombin–antithrombin (TAT) complex during tissue factor-induced whole-blood reactions. Illustrated are data for normal individuals, and data for two individuals with hemophilia A. The clotting times for the composite and two individual experiments are illustrated on the horizontal axis. Modified from Cawthern et al.[57], by permission.

The clinical diagnoses of hemophilia A, B and C involve functional analyses of plasma clotting using APTT, which is most likely a consequence of an in vitro artefact. While it is well established that these congenital deficiency diseases are associated with hemorrhagic disease, these deficiencies are not reflected in the TF-initiated PT assay. Since we conclude that the TF pathway is relevant to the hemostatic mechanism a rationale for the hemophilia insensitivity of PT requires some explanation. A typical PT assay employs ∼ 20 nmol L⁻¹ TF, which will produce a clot in 11–15 s. Since the presentation of a clot depends only on the generation of 10–30 nmol L⁻¹ thrombin, at the high TF concentrations used, robust generation of the required 7 pmol L⁻¹ FXa by FVIIa–TF completely masks the contribution of the FVIIIa nmol L⁻¹FIXa complex in clot endpoint assays. For the illustrations of Figs 2–6, a concentration of 5 pmol L⁻¹ TF was used, producing a clotting time of ∼4–5 min. At these TF concentrations, the clotting time in hemophilia A and B is prolonged, but the major defect is associated with the absence of a propagation phase [54,56].

Reaction thresholds

The principal stoichiometric inhibitors of the process are TFPI and anti-thrombin III (AT-III). TFPI is the principal regulator of the initiation phase of thrombin generation, while AT-III serves to attenuate thrombin activity and its generation. These two agents, when combined, provide a synergistic regulatory effect by inducing kinetic ‘thresholds’ such that the initiating TF stimulus must achieve a significant magnitude to propel thrombin generation [53]. TF concentrations below the threshold concentration are ineffective in promoting robust thrombin generation, because of the cooperative influence of the inhibitors; concentrations in excess of the threshold yield robust and almost equivalent thrombin generation. In a similar fashion, TFPI and the dynamic PC–thrombomodulin–thrombin system cooperate to provide a threshold-limited synergistic inhibition of thrombin production [63]. In this instance, TFPI slows the initiation phase while the APC system interferes with the activation of the cofactors FV and FVIII.

The activations of the cofactors FV and FVIII are multistep processes, and at high thrombomodulin concentrations the PC system is an effective neutralizer of the reaction. FV activation involves cleavages at arginines 709, 1018 and 1545 [64]. The cleavage at Arg709 occurs first and produces the heavy chain (residues 1–709) of the molecule. FVa activity, however, requires cleavage at Arg1545 to produce the light chain (residues 1546–2329) of FVa. APC inactivates FVa (and intermediates in the activation process) principally by cleavage at Arg506 and Arg306. Each of these cleavages is in the heavy chain [50]. Thus, the heavy chain can be inactivated prior to generation of the light chain of FVa, eliminating its procoagulant receptor ‘activity’ prior to its genesis.

The effectiveness of the synergy between TFPI and AT-III can be observed in the empiric experiments described in Fig. 7. In these experiments, performed with a synthetic plasma system, the reaction mixture is initiated with varying concentrations of FVIIa–TF. At the two highest concentrations, using 125 pmol L⁻¹ and 25 pmol L⁻¹, thrombin generation is almost equivalent. When the activator concentration is reduced to 10 pmol L⁻¹ a virtual shutdown of the reaction system occurs. Similar observations have been made with a combination of TFPI and APC.

Figure 7. Procoagulant mixtures are induced to thrombin generation with the addition of FVIIa–tissue factor complexes at the concentrations listed. Reaction mixtures contain procoagulants and the stoichiometric anticoagulants. The total (AREG) of thrombin produced at 125 pmol L⁻¹ activator and 25 pmol L⁻¹ activator are essentially the same. When the concentration of activators is reduced to 10 pmol L⁻¹ the reaction is largely attenuated, thus a reaction threshold exists between 25 pmol L⁻¹ and 10 pmol L⁻¹ activator in this illustration. From van't Veer et al.[54], by permission.

The consequence of the integration of the effects of procoagulant and anticoagulant systems governing thrombin generation is the establishment of an effectively ‘digital’ system that is either off or on. Once a threshold has been reached, the response of the process is only dictated by the pro- and anti-coagulant reagent concentrations present in a reaction volume. This produces a teleologically desirable expectation for the hemostatic system, i.e. it responds effectively and rapidly to a threat dealing with hemorrhage but is not provoked to activation by inconsequential stimuli.

The role of FXI

The significance of FXI as an important procoagulant is established by the bleeding pathology frequently associated with its qualitative or quantitative absence [65]. This zymogen is also a substrate for thrombin and has been invoked in the ‘revised pathway of coagulation’[10]. In studies of natural hemophilia C blood, antibody-acquired hemophilia C, or synthetic plasma FXI deficiency, the significance of FXI deficiency is only prominent at the lowest TF concentrations [56]. At moderate concentrations of TF (5–10 pmol L⁻¹), congenital FXI deficiency has little or no effect on thrombin generation or other procoagulant parameters. However, at low levels of TF (1–2 pmol L⁻¹), which produces clotting in the range of 12–15 min, the generation of thrombin and formation of fibrin are defective in FXI deficiency. The variability of pathology in FXI deficiency is perhaps a reflection of the dimension of the TF stimulus associated with the vascular lesion.

The significance of thrombin generation

The ‘initiation/propagation phase’ description of thrombin generation is an operational concept that permits a segregated discussion of the heterogeneous kinetics of the thrombin generation process. However, the two phases are intrinsically intertwined, and thrombin production during both phases is essential to the overall hemostatic process. Since the blood clotting process coincides with the transition between the initiation and propagation phases of reaction, most of our knowledge of hemorrhagic syndromes is presently confined to the associations of bleeding pathology with thrombin formation during the initiation phase of the reaction. However, it is clear that in well-established hemorrhagic syndromes only modest alterations of thrombin production occur during the initiation phase, while depression of thrombin production during the propagation phase is the hallmark of these congenital bleeding diseases.

We have studied a variety of drug effects, pathologic syndromes and congenital polymorphisms using the model systems described in this presentation. In this section we describe the influences of these events on thrombin generation.

Platelets and thrombin generation

An essential platelet function provides the binding sites for the vitamin K-dependent complexes the components of which are derived from plasma. We have studied the influence of thrombocytopenia, thrombocytosis, and impaired platelet function by pharmacologic agents on the process of thrombin generation [61,66,67].

While some platelet function is essential to all elements of thrombin generation in a TF-induced reaction, the principal impact of reduced platelet concentrations or pharmacologically impaired platelet function is on the propagation phase of the reaction. The initiation phase of the reaction, defined as the time to generate 10 nmol L⁻¹ thrombin or the clotting time of blood, becomes impaired when platelet concentrations fall below 10 000 mm⁻³.

Because of the extreme sensitivity of platelets to thrombin, platelet activation is ordinarily not a rate-limiting step in the TF pathway. In fact, preactivation of platelets with the thrombin receptor activation peptide has no effect on thrombin generation either during initiation or propagation phases. However, the inhibition of platelet function by strong antagonists has profound effects both on the initiation and propagation phases of the reaction. Selective inhibitors, such as the clinically useful glycoprotein IIbIIIa antagonists Integrelin and Abciximab, suppress the propagation phase of the reaction. Perhaps this propagation phase inhibition contributes to the antithrombotic effects of these agents.

Plasma protein quantitative and qualitative alterations

The journal Blood uses the term ‘healthy’ to describe the properties occurring in an apparently normal population. Studies of severe congenital hemorrhagic and thrombotic disease are confined to the extremes of the ‘normal’ or ‘healthy’ population distribution [Fig. 8]. However, significant concentration variations occur within the normal population. For many plasma proteins, variations in concentration extend over a range of 50–150% of the mean plasma value. These values are frequently used in clinical evaluations as independent variables.

Figure 8. Hypothetical frequency distribution of the pro- and anti-coagulant parameters for the human population. At the extremes of the population distribution are those individuals with profound congenital defects leading to either hemophilia or thrombophilia. The majority of individuals fall in to the ‘healthy’ population distribution range, nominally 50–150% of the mean for procoagulants and anticoagulants.

In studies in vitro, altering plasma protein concentrations within the nominal range of 50–150% can produce major influences on thrombin generation [52]. AT-III and prothrombin have major individual impacts, with their effects principally observed on the propagation phase of the reaction. Data for prothrombin over the range of 50–150% are illustrated in Fig. 9. Panel A illustrates a numerical simulation while panel B illustrates the result for synthetic plasma when prothrombin concentrations are varied over the range of 50–150% of the mean plasma value. When FV concentrations are at 50% of the normal mean value, the duration of the initiation phase is extended; however, little effect on the propagation phase of the reaction occurs. Individual variations of most plasma coagulation proteins in the ‘normal’ range produces little or no effect on thrombin generation; however, multiple simultaneous alterations of these same constituents can have quite significant influence.

Figure 9. The extent of thrombin generation for reaction mixtures containing various levels of prothrombin ranging from 50% (^*) to 150% (◆). The left panel is a numerical simulation of thrombin generation under these conditions while the right panel is empiric ‘synthetic plasma’ data corresponding to the same conditions. From Hockin et al.[52], by permission.

Patients treated with oral anticoagulants display significant impairment of both the initiation and propagation phases of thrombin generation [68]. The extent of these alterations is not predicted by the international normalized ration (INR), which is conventionally used to monitor anticoagulant therapy. For the same individual with INR ∼2.5, based upon PT analysis, the initiation phase duration was observed to extend from 10 to 23 min while the maximum thrombin levels varied from 100 to 300 nmol L⁻¹ (normal value >750 nmol L⁻¹).

FV Leiden, when combined with reduced TFPI levels, can produce an unregulated propagation phase of thrombin generation [69]. This is illustrated in Fig. 10 for a synthetic plasma experiment in which low normal (50%) TFPI was combined with homozygous FV Leiden. In the control with normal FV (▪) the propagation phase of thrombin generation is attenuated substantially at the TF concentrations used in this experiment (i.e. it is near threshold). However, with FV Leiden, and low normal TFPI (●), the propagation phase of thrombin generation is equivalent to a situation in which the PC regulatory system was totally dysfunctional (□). This observation would suggest the risk of thrombotic pathology. An in vivo parallel to this in vitro experiment has been reported by Eitzman et al. [70], who showed FV Leiden^+/+ mice with reduced but low levels of TFPI^+/− died of thrombosis.

Figure 10. An illustration of gene–gene influence on regulation: Homozygous FV Leiden with reduced (50%) levels of tissue factor pathway inhibitor. The combination of synthetic plasma alterations leads to a profound deficit in antithrombotic function. Modified from van't Veer et al.[64], by permission.

Conclusion

Many of the illustrations provided show an association of established pathology with the absence of robust thrombin generation during the propagation phase of the TF-initiated reaction system. These observations lead to the hypothesis that impairment of the propagation-phase generation of thrombin is associated with the pathology of hemorrhagic disease and that an unquenched propagation phase of thrombin generation would be associated with thrombotic disease. These observations should also be viewed in light of the observation that impairments of TF reactions essential to thrombin generation are associated with hemorrhagic pathology in human beings, and complete absence of this pathway and its regulation in mice is lethal. In contrast, in both humans and mice, hypofibrinogenemia is not necessarily associated with hemorrhagic pathology and is certainly compatible with life. These observations lead to the speculation that the terms ‘blood coagulation’ and ‘hemostasis’ should not be used equivalently; rather, it appears that the fibrinogen clotting process is only one facet of the antihemorrhagic system and we have a great deal more to learn about ‘what all that thrombin is for’. We conclude that new technologies aimed at evaluating all thrombin formation will be useful in evaluating hemostatic and thrombotic disease and prophylaxis thereof.

Acknowledgement

This study was supported by HL-34575, HL-46703, Training Grant HL-07594 from the National Heart Lung and Blood Institute.