Maureane Hoffman, Pathology & Laboratory Medicine Service, Durham VA Medical Center, 508 Fulton Street, Durham, NC 27705, USA. Tel.: +1 919 286 0411 ext. 6494; fax: +1 919 286 6818. E-mail: email@example.com
Summary. Understanding the mechanism of action of normal hemostasis and how the bypassing agents recombinant activated factor VII (rFVIIa; NovoSeven) and plasma-derived activated prothrombin complex concentrate (Factor Eight Inhibitor Bypassing Agent [FEIBA]) control abnormal bleeding is imperative for healthcare professionals who treat patients with hemophilia and other bleeding disorders. A cell-based model has improved our understanding of in vivo mechanisms of hemostasis and the basis of the bleeding tendency in hemophilia. Bypassing agents do not restore the normal pathways of hemostasis in hemophilia, but rather boost thrombin generation in spite of a lack of platelet surface FVIIIa–FIXa (‘tenase’) activity. Thus, the common clinical laboratory coagulation assays do not reflect the clinically relevant hemostatic activity of bypassing agents, and no validated assay is available with which to measure the in vivo efficacy of these agents or predict individual patient responses to treatment. Global hemostasis assays measuring overall coagulation capacity have potential for assessment of the effects of bypassing agents. This review will focus on the mechanisms of clotting and their relationship to understanding the mechanisms of action of the bypassing agents in vivo and the methodologies that are emerging to monitor the clinical efficacy of bypassing agent therapy.
Our understanding of the mechanisms underlying hemostasis has evolved substantially over the last decade, moving away from a protein-centric model to one that reflects the role of cells in the clotting process. A cell-based model provides a physiologically relevant framework for understanding hemostasis [1,2]. In contrast to the prevailing ‘cascade’ model [3,4], the cell-based model recognizes that cells are pivotal in localizing and regulating the activity of the coagulation proteins. By extension, this model clarifies the mechanisms underlying the bleeding tendency in hemophilia and related disorders [5,6]. It has also been useful in clarifying the mechanisms by which recombinant activated factor VII (rFVIIa) exerts its effects [7–9].
The bypassing agents rFVIIa and Factor Eight Inhibitor Bypassing Agent (FEIBA) (a plasma-derived activated prothrombin complex concentrate [APCC]; Baxter Healthcare, Westlake Village, CA, USA) are first-line therapies for the treatment of acute bleeding episodes in patients with hemophilia and high-titer inhibitors [10,11]. However, optimal use of these agents is complicated by individual variability in response [12,13] and the lack of reliable laboratory methods for predicting and monitoring their effects [14–18]. The ultimate goal is to optimize and individualize the treatment of patients with hemophilia to rapidly achieve hemostasis with administration of the minimum dose of drug. The objectives of this review are to: (i) discuss the mechanisms of hemostasis; (ii) review the prohemostatic mechanisms of rFVIIa and FEIBA; and (iii) discuss the limitations of conventional and emerging testing approaches to predict and monitor responses to bypassing agent therapy.
The cellular basis of hemostasis
The protein-centric view of coagulation regulation focuses on the role of the plasma coagulation factors assembled on negatively charged phospholipid surfaces. This ‘coagulation cascade’ model (Fig. 1A) involves a stepwise sequence of proteolytic reactions, organized into a tissue factor (TF)-dependent ‘extrinsic pathway’ and a contact factor-dependent ‘intrinsic pathway’ that converge in a ‘common pathway’ of thrombin generation and fibrin deposition.
The cell-based model of hemostasis (Fig. 1B) recognizes the importance of the sequential proteolytic reactions during coagulation, but places them within a physiologic context in which specific cells control the localization and activity of these proteins . In this scheme, the ‘intrinsic’ and ‘extrinsic’ pathways are not redundant, as they appear to be in the cascade model. Rather, the ‘extrinsic’ pathway acts on TF-bearing extravascular cells to initiate the hemostatic process, while the essential factors of the ‘intrinsic’ pathway (FXI, FIX–FVIII, and FX–FV) act on activated platelets to generate the large burst of thrombin generation needed to form a stable fibrin clot. In the cascade model, it is not clear why a deficiency of FVIII or FIX should produce the bleeding phenotype of hemophilia, as it appears that the ‘extrinsic’ pathway should be able to compensate for a defect in the ‘intrinsic’ pathway. However, the cell-based model makes clear that both pathways are required for normal hemostatic function, because they fulfill two different roles on two different cell surfaces during hemostasis .
TF is strongly expressed by pericytes around the adventitial surface of vessels, as well as by epithelial and neural tissues [19,20]. TF around vessels can bind FVII–FVIIa even in the absence of an injury . TF–FVIIa activates FIX and FX even in the basal, non-bleeding state, thereby producing a small amount of thrombin. However, this does not normally lead to fibrin formation, because of the separation of platelets and FVIII–von Willebrand factor within the vasculature from TF-bearing cells in the extravascular space. The initiation phase of hemostasis begins when a breach in a vessel releases blood into the extravascular space. Platelets adhere to collagen in the extravascular matrix, become partially activated, and begin to release their granule contents. The small ‘priming’ amount of thrombin (Fig. 1B) produced by the activity of the TF pathway enhances platelet activation via proteinase-activated receptors and glycoprotein (GP)Ib–IX receptors. The combined stimulation of platelets by collagen and thrombin leads to the formation of highly procoagulant platelets (sometimes called ‘coated’ platelets ). It is interesting that the relative ability of individuals to form these highly active platelets is correlated with the severity of the bleeding tendency in patients with hemophilia .
In the amplification phase of hemostasis, the initial procoagulant signal is amplified as thrombin mediates activation of FV, FVIII and FXI on platelet surfaces (Fig. 1B) [5,24]. The assembly of FVIIIa–FIXa and FXa–FVa complexes on activated platelets sets the stage for the third phase of hemostasis. During the propagation phase, a burst of thrombin generation occurs on activated platelets. This large amount of thrombin drives the recruitment and adherence of additional platelets and cleavage of fibrinogen to fibrin to consolidate the initial platelet plug in a stable fibrin meshwork . Thrombin additionally contributes to stabilization of the fibrin network by activation of FXIII  and thrombin-activatable fibrinolysis inhibitor (TAFI) .
The nature of the hemostatic defect in hemophilia
Viewed in light of the cell-based model, the defect in hemophilia is specifically in the propagation phase of hemostasis that produces the thrombin burst . Whereas the initiation and amplification phases progress normally, the assembly of FVIIIa–FIXa complexes is compromised on platelet surfaces, owing to very low levels of one or the other of these key coagulation factors in patients with hemophilia A or B that result in impaired platelet surface FXa generation and a failure of the thrombin generation needed to produce a stable fibrin clot structure (Fig. 2A,B). The rate and peak level of thrombin activity produced during clotting are low in hemophilic blood, leading to the formation of a fibrin clot that is not structurally stable. Reduced activation of FXIII and TAFI further impairs the stability of the hemophilic clot. Hemophiliacs often initially stop bleeding after injury, owing to the formation of a hemostatic platelet plug. However, failure to adequately stabilize that plug with fibrin leads to later rebleeding.
The mechanisms of action of bypassing agents
Clinical responses to bypassing agents are highly variable, with different studies reporting clinical efficacy ranging between 80% and 85% with both APCC and rFVIIa [13,18]. Several studies have attributed variability in patient response to factors that are particular to the type of bleed and to patient-specific factors, including the location of bleeding, patient age, and the presence of a target joint [24,27,28]. Another important source of variability can be attributed to differences in the mechanism of action of the bypassing agents. Although the mechanism of action of both agents targets the burst of thrombin generation as the final common mediator for clot formation, the precise details of the mechanism(s) by which they promote hemostasis are different (Fig. 2C,D) [7,29].
The original rationale for the use of rFVIIa as a bypassing agent was that FVIIa would drive activation of FX via TF–FVIIa in patients who were unable to form the ‘intrinsic’ FX-activating complex (FIXa–FVIIIa), bypassing the need for FVIII or FIX. In light of the ‘cascade’ model (Fig. 1A), this seemed to be a reasonable hypothesis. However, it soon became clear that the mechanism was not as straightforward as originally hypothesized. Very high levels of rFVIIa are required for hemostatic efficacy, higher than those needed to saturate TF binding . Thus, a debate ensued over whether the mechanism of rFVIIa activity was ‘TF-dependent’ or ‘TF-independent.’
Bom and Bertina  demonstrated that rFVIIa can directly activate FX on a negatively charged phospholipid surface in the absence of TF, and Rao and Rapaport  initially suggested that this mechanism might be responsible for the hemostatic activity of rFVIIa. However, several years later, Rao and Rapaport  found that rFVIIa can compete with the zymogen FVII for binding to TF, and increase procoagulant activity by forming TF–FVIIa rather than TF–FVII complexes. This competition could explain why such high levels of rFVIIa might be needed, even if it acted by a TF-dependent mechanism.
The cell-based model localizes the defect in hemophilia specifically to platelet surface FXa generation. Thus, it seemed plausible that high-dose rFVIIa might act to remedy this defect. Monroe et al.  demonstrated that rFVIIa binds to activated platelets independently of TF, and partially restores thrombin generation, in an in vitro model of hemophilia. Analogously to the mechanism proposed by Bom and Bertina , Monroe et al. proposed a platelet surface mechanism in which rFVIIa binds to the surface of activated platelets, which could account for the localization of rFVIIa activity to a site of injury.
van’t Veer et al. , in turn, provided evidence in support of the Rao and Rapaport hypothesis. They demonstrated that plasma levels of non-activated FVII delayed thrombin generation in a model of hemophilia, and that rFVIIa could overcome the inhibition. They concluded that the prohemostatic effect of rFVIIa resulted from enhanced activation of FX by TF–FVIIa in the presence of high levels of rFVIIa. However, three lines of evidence argue against this mechanism. First, the zymogen FVII delayed the onset of ‘coagulation’ in an experimental model with TF incorporated into phospholipid vesicles, but not when a cellular source of TF was used . Second, the competitive effect of rFVIIa is saturated at ∼ 10 nm, whereas the platelet surface effect of rFVIIa is not saturated at 250 nm. The finding that the efficacy of rFVIIa in patients with hemophilia was increased by dose escalation to levels of 100 nm or more supports a platelet surface mechanism. Third, rFVIIa variants have been engineered with increased TF-independent activity but TF-dependent activity equivalent to that of wild-type FVIIa. Variants with increased TF-independent activity have increased hemostatic efficacy in animal models of hemophilia . Thus, although not ruling out a TF-dependent effect, the available evidence suggests that a platelet surface mechanism (Fig. 2C) is a major contributor to the prohemostatic effects of rFVIIa in hemophilia.
There remain uncertainties about the mechanism of the platelet surface activity of rFVIIa. It was originally proposed that rFVIIa bound to anionic phospholipid on activated platelets; however, Weeterings et al.  found that rFVIIa binds to the platelet GPIb–IX–V complex, in addition to negatively charged membrane phospholipid. This interaction appears to slightly enhance rFVIIa-induced platelet surface thrombin generation. In addition, an rFVIIa variant with increased TF-independent activity showed greater binding to activated platelets than wild-type FVIIa, even though both showed identical binding to phospholipid vesicles . Thus, platelet characteristics in addition to phospholipid composition play a role in rFVIIa binding and activity. Variation in platelet characteristics could possibly underlie the variability observed in the response of patients to treatment with rFVIIa.
A variety of plasma-derived prothrombin complex concentrates (PCCs) have been available over the years and used to manage patients with hemophilia A and inhibitors . APCCs appeared to have increased efficacy in patients with severe hemorrhage . The mechanism of their effects in inhibitor patients was thought to be related to the presence of activated components of the prothrombinase complex. On this basis, the partially activated PCC named FEIBA was developed, and has been available for use in inhibitor patients for over 30 years .
FEIBA is a mixture of vitamin K-dependent coagulation factors separated from a pooled human plasma fraction after removal of the cryoprecipitate [29,41,42]. After concentration by ultrafiltration, the protein mixture is subjected to an activation step followed by a vapor heat treatment procedure for inactivation of blood-borne viruses. The coagulation factors in FEIBA include FII (prothrombin), FVII, FIX, and FX, small quantities of FIXa, FXa, and thrombin, and larger amounts of FVIIa [29,43].
A number of groups have investigated the mechanisms of action of FEIBA and other PCCs. In 1981, Barrowcliffe et al.  suggested that FVIII was protected from inactivation by antibodies by phospholipids present in PCCs. On the basis of animal experiments, procoagulant phospholipid has also been suggested to be the component of PCCs responsible for their thrombogenicity . Later, the knowledge that PCCs contained high levels of FVIIa suggested that this component might be responsible for the prohemostatic effects of these agents. This hypothesis contributed to the development of rFVIIa as a bypassing agent.
Owing to the complexity of its composition, the mechanism of action of FEIBA has been difficult to unravel, and remains incompletely defined. However, the prevailing view has returned to the idea that FEIBA acts by boosting activity of the prothrombinase complex. Turecek et al. [29,46] showed that a complex of FXa and prothrombin could shorten the clotting time of plasma containing a high-titer FVIII inhibitor, provided that FV was also present. It appears that FXa is protected from inhibition by antithrombin when bound to prothrombin. The prothrombin–FXa complex reproduces the effects of FEIBA in a number of models, and thus has been suggested to be the major mechanism of action of FEIBA .
In addition, Gallistl et al.  found that the high level of prothrombin in FEIBA plays a direct role in its hemostatic effect. Thrombin production in vitro has been shown to be dependent on the level of prothrombin [41,42]. High levels of prothrombin may not only increase the velocity of thrombin generation but also facilitate the assembly of FXa and FVa into prothrombinase complexes.
Although there is a significant amount of FVIIa in FEIBA, it does not appear to be a major contributor to its hemostatic action [46,47]. Other coagulation factor zymogens present in FEIBA may also make minor contributions to hemostatic effects [29,46]. Thus, the mechanism of action of FEIBA appears to involve the activity of multiple procoagulants, but is primarily related to enhancing platelet surface prothrombinase activity (Fig. 2D).
Challenges associated with monitoring bypassing agent therapy in patients with inhibitors
Monitoring and predicting the response of patients to bypassing agents remains a significant challenge, as no standardized and validated assay is currently available for that purpose [15,48–50]. The effectiveness of bypassing therapy is primarily monitored by clinical observation [13,42].
Conventional assays, including the prothrombin time and activated partial thromboplastin time (APTT) (Table 1), do not reflect the physiologically relevant hemostatic activity of bypassing therapies, and are not useful for monitoring responses to bypassing agent treatment [15,16,49,51]. This is mainly explained by the endpoint of these assays, which is the formation of detectable fibrin clot formation. However, the first traces of fibrin appear by the end of the initiation phase of coagulation at very low levels of thrombin, < 5% of the total amount of thrombin produced. The majority of thrombin is generated after clot formation during the propagation phase, and this is not captured by a fibrin clotting endpoint. The fact that these assays do not account for the contributions of cells to thrombin-generating capacity is another limitation.
Table 1. Current and emerging laboratory assays for monitoring response to bypassing agent (BPA) therapies*
Type of assay
Monitoring BPA response
APCC, plasma-derived activated prothrombin complex concentrate; APTT, activated partial thromboplastin time; FVIII:C, factor VIII activity; FIX:C, factor IX activity; PPP, platelet-poor plasma; PRP, platelet-rich plasma; PT, prothrombin time; rFVIIa, recombinant activated factor VII; TGT, thrombin generation test. *Based on information contained in Refs [51,14].
PT APTT FVIII:C FIX:C
PT and APTT assays are functional assays using platelet-free plasma FVIII:C and FIX:C assays are functional assays
Coagulation defect in extrinsic or intrinsic pathways Factor level deficiencies can be measured
No assessment of clot structure or stability PT and APTT do not correlate with BPA efficacy FVIII:C and FIX:C levels do not correlate with BPA efficacy
Conventional coagulation and factor assays cannot be used to monitor or predict clinical BPA response in a given patient
Fluorometric measurement of real-time changes in thrombin generation in PPP or PRP
Continuous measurement of thrombin generation in PPP or PRP
Use of PPP may underestimate the hemostatic effect of rFVIIa by ∼ 30% Not yet standardized or validated for clinical use
Has been used to measure thrombin generated in response to APCC and rFVIIa Has been prospectively evaluated to direct BPA therapy for surgery in hemophiliacs with inhibitors; TGT results correlated with the clinical outcome of patients
Measurement of the viscoelastic changes of whole blood during clot formation
Requires whole blood (< 0.5 mL)
Not yet standardized or validated for clinical use
Has been used to measure clot formation and properties in whole blood treated with APCC and rFVIIa
APTT waveform analysis
Changes in light transmittance measured over time during APTT assay
High sensitivity to detect low levels of factor activity
Requires automated analyzer for waveform analysis Not yet standardized or validated for clinical use
Not well characterized; has been used in one study to measure the effects of rFVIIa on clot formation
Assays for determining the plasma level of FVIII activity (FVIII:C) or FIX activity (FIX:C) (Table 1) are also not useful for monitoring the response to bypassing agents, because these agents increase thrombin generation without normalizing the levels of FVIII:C or FIX:C.
The use of ‘global’ hemostasis assays (Table 1) has gained traction recently in the quest for a reliable monitoring assay for bypassing therapy [15,16]. Three types of global hemostasis assay are currently being investigated as assays of overall hemostatic function and to monitor response to hemostatic therapy: (i) thromboelastography (TEG System [Haemonetics, Braintree, MA, USA] or ROTEM [Tem Systems, Durham, NC, USA]); (ii) APTT clot waveform analysis; and (iii) thrombin generation tests (TGTs).
Several studies have examined thromboelastogram profiles in whole blood (Table 1) of inhibitor patients with hemophilia A in response to FEIBA and rFVIIa [48,52–54]. Addition of rFVIIa to hemophilia A blood samples resulted in a change in the thromboelastogram, suggesting that this approach might be useful in monitoring bypassing agent therapy [48,53,55]. An additive effect was observed on thromboelastogram profiles in patients who were switched to the alternative agent after failing to respond to initial treatment with either rFVIIa or FEIBA . These initially promising results have not been confirmed by larger clinical trials. Thromboelastogram parameters did not show a clear dose response with increasing amounts of rFVIIa in most patients tested in a multicenter study . High intrapatient and interpatient variability in thromboelastogram assays has been observed , as well as a high coefficient of variation on assay of external quality assurance samples with the TEG and ROTEM systems . Some of this poor assay performance may be attributable to the lack of standardization of the reagents used to initiate coagulation in thromboelastogram assays. Thus, improved standardization may yet improve the performance of such assays in monitoring bypassing therapy. However, at this point, none has been validated for this purpose.
Another approach to monitoring bypassing agent therapy involves the APTT clot waveform analysis in platelet-poor plasma to evaluate clot formation (Table 1) [58,59]. This automated method measures the velocity and extent of fibrin polymerization in plasma during the APTT assay. The APTT waveform analysis method has been suggested to correlate the severity of bleeding tendency with levels of plasma FVIII:C detected to very low levels (< 1 IU dL−1). Analysis of APTT waveform profiles has also been shown to detect variability in coagulation activity in individual patients [50,59]. However, only one study has tested the usefulness of APTT waveform analysis to monitor bypassing agent efficacy; this in vitro study reported dose-dependent improvements in fibrin polymerization in response to rFVIIa in patients with severe hemophilia A . Currently, no data are available demonstrating a correlation between clinical outcomes of patients and results of APTT clot waveform analysis.
Several groups have reported the ability of FEIBA and rFVIIa to promote thrombin generation in vitro and ex vivo in both platelet-poor and platelet-rich plasma samples with TGTs. Consistent with the platelet surface mechanism of action of rFVIIa, evaluation of rFVIIa effects requires measurement of thrombin generation in platelet-rich plasma. Interestingly, the ability of FEIBA to promote thrombin generation can be assessed by using phospholipid vesicles in platelet-poor plasma. The use of supraphysiologic concentrations of rFVIIa in some studies called into question whether the results of TGTs reflected clinically relevant effects [61,62]. However, a small prospective clinical study in inhibitor patients undergoing surgery showed a good correlation between TGT results and the clinical outcome of patients treated with bypassing therapies . In this study, a three-step protocol was established to tailor therapy for each patient. Prior to surgery, the effect of each bypassing agent on the TGT was determined by addition to a sample of each patient’s blood. The dose of rFVIIa or FEIBA needed to normalize thrombin generation was determined and subsequently confirmed by TGT after administration of the chosen dose of product to the patient. Bypassing therapy for surgery was administered to each patient on the basis of the in vitro and ex vivo TGT results. The effect of treatment was then monitored by TGT in parallel with standard clinical evaluation perioperatively and postoperatively. There were no bleeding complications in patients whose thrombin generation capacity was normalized by the prescribed therapy. However, a patient who developed subnormal thrombin-generating capacity postoperatively also developed bleeding. This study suggests that TGT results correlate with bleeding risk, and supports the potential of TGTs to predict the in vivo response to bypassing agents [63,64].
At present, no standardized protocol has been established for routine measurement of thrombin generation in clinical laboratories. Two international standardization studies on TGTs demonstrated that preanalytic and analytic variables can influence the reliability of TGT results, and underscored the need for a standardized protocol to ensure accuracy and precision of the results [65,66]. The many issues involved in standardization of TGTs have been highlighted in a recent report of the Anticoagulation Subcommittee of the Scientific Sub-Committee of the ISTH . In spite of the many difficulties involved in adapting TGTs to routine clinical use, this approach seems, at present, to have the most potential for predicting and monitoring bypassing therapy.
Validated, standardized laboratory testing strategies for monitoring the clinical response to bypassing agent therapy in patients with hemophilia and inhibitors are currently non-existent. Therefore, a clear need exists for the development and widespread implementation of improved laboratory tests to monitor the response to rFVIIa and FEIBA. Substantial progress has been made in the effort to standardize global hemostasis assays prior to organizing large multicenter clinical trials to evaluate and validate these assays for monitoring bypassing agent therapy. Because global hemostasis assays more closely emulate conditions under which hemostasis occurs in vivo, they show the most promise for assessing the response to bypassing agents, predicting an individual’s bleeding risk, and tailoring therapy for each patient. Barriers to the widespread adoption of global hemostasis assays include cost and availability of technology, lack of automation, and lack of reagent standardization. Although significant challenges remain, new testing approaches hold promise for optimizing the use of expensive pharmaceuticals and improving patient outcomes.
The authors acknowledge the writing and editorial assistance provided by J. M. Palmer (ETHOS Health Communications, Newtown, PA, USA), with financial assistance from Novo Nordisk Inc., in compliance with international Guidelines on Good Publication Practice.
Disclosure of Conflict of Interests
M. Hoffman has received research grant support from Novo Nordisk, Inc. and consulting fees from CSL Behring and Baxter Healthcare. Y. Dargaud has received research grant support and speaker fees from Novo Nordisk and Baxter Healthcare.