A cell-based model of Intrinsic p8thW8y

factor XII
coagulation and the role of HMWK

factor Vlla factor Xl-

4 factor XIa Extrinsic pathway

Maureane Hoffman i
factor IX- factor tXi3 factor Vita
factor VllIa Tissue facbr
Department of Pathology, Duke University Medical Center, Durham. North
Carolina. USA 1 = 4
factor X factor xa - facmr x
factor Va
Abstract Our cell-based model of haemostasis replaces the
traditional ‘cascade’ hypothesis, and proposes that coagulation
takes place on different cell surfaces in three overlapping steps:
initiation, amplification, and propagation. In highlighting the im-
portance of cellular control during coagulation, the cell-based fibrinogen --+ fibrin
model allows a more thorough understanding of how haemosta-
Fig. I. The ‘cascade’ hypothesis: intrinsic and extrinsic pathways.
sis works in vivo, and sheds light on the pathophysiological mech-
anisms behind certain coagulation disorders. For instance, this
model proposes that haemophilia involves a failure of platelet-
surface FXa generation, leading to a lack of platelet-surface : spite a prolonged partial thromboplastin time (FIT), which
thrombin production. Our data suggest that high-dose FVlla is : indicates a disturbance in the functional activity of the in-
able to bind weakly to activated platelets, independently of tissue : trinsic pathway.* In contrast, an increased predisposition to
factor, in otder to generate sufficient amounts of FXa to support : haemorrhagic risk may be present in patients deficient in
a burst bf thrombin generation in the absence of FIXa/FVllla. _ FXI. The degree of prolongation of the FIT in this disor-
The considerable success of high-dose recombinant FVlla (rFVlla; i der, however, does not necessarily predict the extent of the
NovoSeven@, Novo Nordisk, Copenhagen, Denmark) as a ther- ; bleeding tendency, which is typically less severe than that
apy for patients with haemophilia and inhibitors has led to its use : observed in haemophilia.
in a growing number of alternative indications. We believe that The cascade hypothesis cannot account for the varying
even in the presence of the FIXa/FVllla complex, rFVlla may be : degrees of haemorrhagic tendency and diverse clinical obser-
able to enhance both FXa and FlXa levels on the surface of acti- : vations that result from deficiencies of different components
vated platelets, thus increasing the production of thrombin. : of the two pathways. In an attempt to explore the process
0 2003 Elsevier Science Ltd. All rights reserved. : of haemostasis from new angles, we developed experimental
! and conceptual models that would allow us to test hy-
KEY WORDS: coagulation; cell-based model; haemostasis; re- ! potheses in a biochemical or Ed r&o system. This, in turn,
combinant factor Vlla; haemophilia f would increase understanding of how the normal haemo-
: static system actually works in &JO. In addition, we wished
: to explore the mechanism of haemostasis in haemophiliac
INTRODUCTION : patients. Such patients have a normal prothrombin time
*
: (F’T), which measures activity of the extrinsic pathway, de-
* $‘ he classical model of coagulation describes a ‘cas- : spite a prolonged F”IT and a pronounced bleeding tendency.
Q cade’ of reactions involving activation of various clot- : Why, then, does the extrinsic pathway fail to compensate for
a ting factors along either an extrinsic or an intrinsic I the dysfunctional intrinsic pathway? In other words, why do
pathway. According to this model, stimulation of either of j haemophiliacs bleed?
these two pathways can result in the production of a large We have developed a cell-based model of haemostasis that
amount of thrombin and subsequent formation of a fibrin : will replace the classical model of the coagulation cascade.3
clot’ (Fig. 1). : This cell-based model emphasises the interaction of clotting
However, although this cascade paradigm supports labo- : factors with specific cell surfaces4 and appears to be able to
ratory evaluation of coagulation disorders and demonstrates : shed light on many of the unresolved issues highlighted by
the interactions between coagulation factors, it does not : the traditional cascade theory.
adequately explain the mechanisms leading to haemostasis
in viva. Furthermore, it does not provide a great deal of
information regarding the pathophysiology of the haemo- ; THE CELL-BASED MODEL OF HAEMOSTASIS
static system. In particular, the model does not explain why
certain categories of patients demonstrate a haemorrhagic The first step in our investigation was to establish an in vitro
tendency; nor does it facilitate accurare prediction of which experimental system incorporating platelets and plasma con-
patients will actually bleed. For instance, patients with a defi- centrations of various clotting factors and coagulation in-
ciency of factor XII @XII), high-molecular-weight kininogen, hibitors. A cellular source of tissue factor (TF) was consid-
or prekallikrein do not present with a bleeding tendency de- ered to be essential, and inclusion of TF-bearing monocytes

0 2003 Elsevler Sdence Ltd. All rights reserved. Blood Reviews (2003) 17, 51-55
Table I Cell-based model system

Component Concentration (nM)

Cells
Monocytes (cultured -

with agents to induce TF)

Platelets (unactivated) -

Proteins
Prothrombin I400
Factor V 25
Factor VIII 0.4
Factor IX 70
Factor X 135 ’ IX
Factor XI 30
Fig. 2. Haemostasis occurs on two cell surfaces: TF-bearing cells and
Factor Vlla 0.2
platelets. (Reproduced from Hoffman M., Monroe D.M. 3rd. Roberts
Inhibitors
HR. Activated factor VII activates factors IX and X on the surface of
TFPI 3
activated platelets: thoughts on the mechanism of action of high-dose
ATIII 2500
activated factor VII. Blood Coagul Fibrinolysis 1998; 9 (Suppl. I): S6l-
565, with permission.)
TF, tissue factor; TFPI, tissue factor pathway inhibitor; ATIII. antithrombin
Ill.

However, reliable evidence exists suggesting that the reac-

ensured that the system contained all necessary components tions responsible for initiating coagulation occur all the time
of the haemostatic process (Table 1).5 The concepts and outside the vasculature in healthy individuals. Coagulation
theories leading to the cell-based model of haemostasis all factors, including FVII, FX, and prothrombin are able to per-
evolved from this kind of experimental system. colate through tissue spaces, and can leave the vasculature
We now know that haemostasis occurs on cell surfaces. in amounts dependent on their molecular size. These factors
Earlier theories derived from the cascade hypothesis sug- can be detected in the lymph and assayed along with their
gested that the coagulation factors themselves were respon- activated forms and activation peptides. Based on this obser-
sible for controlling haemostasis in a system where cells vation, an ‘idling’ theory has been proposed, in which the TF
merely provided a phosphatidylserine-containing surface on pathway remains constantly active, generating low levels of
which the procoagulant complexes could be assembled. activated factors in the basal state.6 Therefore, continual pro-
However, the cell-based model proposes that cells play very duction of small amounts of thrombin takes place outside
active roles in controlling coagulation,* with certain features the vasculature in healthy individuals, even under normal
of the cell surfaces directing the haemostatic process. In circumstances when vascular integrity remains intact.
this system, cells with similar phosphatidylserine contents In effect, the initiation step of coagulation is proceeding
are able to play very different roles depending on their at all times, but does not lead to formation of a blood clot as
complement of surface receptors.4 its location is separated from other key components of the
The cell-based model also emphasises that coagulation oc- coagulation system by an intact vessel wall.
curs in a series of three overlapping steps that take place on
different cell surfaces, rather than as a cascade that produces Amplification
an abundance of activated factors and inevitably leads to clot As a result of vessel damage, components of the haemostatic
formation. The first phase, or initiation, occurs on a TF-bear- system that are normally unable to leave the vasculature
ing cell. In the amplification phase, platelets and cofactors due to their large size are now able to do so. The most
are activated in order to prepare for large-scale thrombin important of these elements are platelets, PVIII, and von
generation. Finally, propagation occurs on the surface of Wfflebrand factor (vWP). As they leave the vascular system,
platelets, and results in the production of large amounts of they come into contact with the limited amount of thrombin
thrombin (Fig. 2).4 that is being generated on the surface of the TF-bearing
cell. Platelets stick to the site of injury, forming a plug at
Initiation the damaged vessel wall, and become fully activated by the
Coagulation is initiated on a TF-bearing cell, which produces thrombin.
the first activated factors (Fig. 3a). This TF pathway may still This same thrombin is also critically important in activat-
be referred to as an ‘extrinsic’ pathway, as the TF-bearing ing coagulation factors. It completes the activation of w,
cell is, under normal circumstances, outside of the vascular which is released from activated platelets, and is responsi-
system and therefore extrinsic to the blood. A large number ble for the cleavage and subsequent activation of FVIII frc%fn
of cells ekpress TP including stromal fibroblasts, mononu- vWF (Fig. 3b). In addition, studies have shown that thrombin
clear cells: macrophages and endothelial cells, but TF is not can also activate FXI, which binds to high-affinity sites on
usually in contact with the blood until injury or inllammation the surface of activated platelets.7,8 This may explain why
occurs. FXII and other contact factors are not always necessary for

Blood Reviews (2003) 17, 5 l-55 0 2003 Ekevier Science Ltd. All rights reserved.
 Ila

(4 IXa

+ FreevWF

TF

I
I
I
I
(b) (--_______
I
A

TFPI = tissue factor pathway inhibitor.

Fig. 3. The ceil-based model of haemostasis: (a) initiation, (b) amplification, (c) propagation.

coagulation, as initially postulated by the original cascade Propagation

hypothesis.
Although insufficient to result in clot formation by itself,
the small amount of thrombin generated at the surface of
TF-bearing cells during the initiation phase is essential in
amplifying the procoagulant signal. At ;he end of the ampli-
fication phase, platelets activated by this limited amount of
thrombin are clad in activated cofactors and FXIa, and the
process of haemostasis moves into the propagation phase.
During propagation, FlXa combines with its cofactor, FVIIIa,
on the surface of activated platelets. Some of the required
FIXa is produced on the surface of TF-bearing cells by
TF/FVIIa, and can diffuse to the activated platelets as it is
not inhibited by tissue factor pathway inhibitor (TPPI), and
is only slowly inhibited by antithrombln III (ATIII). Factor
IXa can also be produced on the platelet surface by FXIa.
Once formed, the FIXa/FVIIIa complex activates FX to

0 2003 Elsevier Science ltd. All rights reserved. Blood Reviews (2003) I7, S/-S5
 FXa, which immediately combines with its cofactor (Fig. 3~).
The FXa/FVa complex then converts large amounts of pro-
thrombin to thrombin, resulting in the cleavage of fibrinogen
to fibrin monomers, which polymerise to consolidate the ini-
tial platelet plug into a stable fibrin clot.
The cell-based model therefore places an emphasis on
the cellular control of coagulation, and is subsequently able
to explain some clinical aspects of haemostasis that the
classical cascade hypothesis cannot.4 It allows a more thor-
ough understanding of how the coagulation process works
in viva, and provides a greater degree of consistency with
clinical observations of various coagulation disorders,
WHY DO HAEMOPHILIACS BLEED?
When compared to the traditional cascade theory, the cell-
based model facilitates a greater understanding of the patho-
physiological mechanisms leading to haemophilia.
stance, the cascade model does not explain why the extrin-
sic pathway appears unable to produce sufficient amounts of
FX to at least partially compensate for a deficiency of FVIII
or FIX. In other words, why does activation of FX by the
TP/FVIIa complex fail to substitute for the FXa that would
normally be generated by FIXa/PVIIIa?’
The cell-based model does not suggest that FXa genera-
tion by the TF/FVIIa complex is insufficient in haemophilia,
but that it occurs on the wrong cell surface. The FIXa/FVIIIa
complex activates FX on the surface of platelets during the
propagation phase, whereas TF/PVIIa can only produce FXa
on the surface of the TF-bearing cell. The FXa produced
on the TF-bearing cell is unable to move to the activated
platelet surface, as there exist two very efficient inhibitors of
FXa in the plasma: TFPI and ATIII. At normal plasma levels,
both TFPI and ATIII inhibit FXa so rapidly and effectively
that the half-life of FXa is 1 minute or less in the fluid
phase.2 Therefore, FXa that remains at the TF-bearing cell
is relatively protected from inhibition, whereas any FXa that
diffuses from the surface is rapidly inhibited.
Accordingly, the cell-based model proposes
haemophilia is specifically a failure of platelet-surface
generation, which results in a lack of platelet-surface throm-
bin production.’ Haemophiliac patients demonstrate rela-
tively normal initiation and amplification phases of coagula-
tion, and so are able to form an initial platelet plug at the
bleeding site, but they cannot generate the burst of throm-
bin at the platelet surface that is necessary to stabilise the
initial plug into a fibrin clot.
HOW DOES HIGH-DOSE FVBa ENHANCE
HAEMOSTASIS IN HAEMOPHJLIA?
As discussed above, the cell-based model of coagulation
suggests that the total amount of FXa produced is less
important than the location in which it is generated.* We
believe that FXa must be formed on the platelet surface
by FIXa/PVIIIa, in close proximity to w, in order to be
incorporated into prothrombinase complexes. This means
that the TF/FVIIa complex cannot compensate for a lack of
m Blood Reviews (2003) 17, S I-55 0 2003 Elsevier Science Ltd. All rights reserved.
FIX/FVIII, as it makes FXa in the wrong place. If this is the
case, then efficient haemophilia treatment must involve the
restoration of FXa generation on the platelet surface.
Our data imply that high-dose FVIIa is able to do just that
- it can enhance haemostasis in haemophiliacs by activating
sufficient PX on the surface of activated platelets to support
a burst of thrombin generation.’
OriginaIly, our group favoured a TFdependent mechanism
in which high doses of FVIIa could ‘drive’ the TF pathway
in haemophiliacs, enhancing the performance of the extrin-
sic pathway and therefore producing haemostasis. It is well
recognised that PVIIa exhibits very little proteolytic activity
in the absence of TE However, the doses of FVIIa required to
achieve coagulation in haemophiliacs produced plasma lev-
els that were several orders of magnitude greater than the &
for binding of PVIIa to TF, leading some researchers to sug-
gest that FVIIa is unlikely to work through a TFdependent
mechanism.’
For in- We used our experimental model to determine how
high-dose FVIIa supports haemostasis in patients with
haemophilia. It was found that FVIIa binds weakly to ac-
tivated platelets, even though platelets do not carry TE
Once bound to the platelet, PVIIa generates a small amount
of PXa, leading to the production of a limited amount of
thrombin on the platelet surface. These findings are also
consistent with our conceptual model of coagulation, which
postulates that platelet-surface FXa generation is required
for the assembly of the prothrombinase complex and subse-
quent thrombin generation. Furthermore, the concentration
of FVIIa required to produce detectable thrombin generation
correlates with the lowest concentration of PVIIa necessary
for clinical efficacy in haemophilia patients. lo
When compared to the amount of FXa that would usu-
ally be produced by the FIXa/FVIIIa complex, the quantity
generated by platelet-bound FVIIa is low. However, it is sig-
nificantly higher than the level of FXa normally produced
on platelets of haemophiliacs, and is certainly sufficient to
enhance thrombin generation in experimental models of FIX
and FVIII deficiency.
that We believe that haemophilia is characterised primarily
FXa by a faihrre of platelet-surface thrombin generation. If this
is the case, then results from our studies in experimental
models suggest that high levels of FVIIa may partially restore
FXa generation on the platelet surface, leading to enhanced
thrombin production in the absence of FIX or FVIII (Pig. 4).
We have tentatively made two extrapolations of our in
vitro data to the in vivo effects of high-dose FVIIa therapy
in haemophilia. First, OUT data suggest that a high dose is
needed because PVIIa binds to platelets with a low affinity
o(d of 50-100 nM, rather than 5 1 nM or less for FVIIa
binding to TF).2 As a result, a high concentration of FVIIa
is required to achieve even a modest degree of platelet
binding. At the concentrations of PVIIa attained in z&o,
binding to platelets is not saturated. This observation led us
to predict that an escalation of FVIIa dose should therefore
increase platelet-surface thrombin generation. Several groUps
have confirmed this theory by demonstrating that clinical
efficacy may be attained by increasing the dose of FVIIa in
those haemophikac patients who fail to respond to initial
dose recommendations.
 haemostatic process in vivo, and facilitates a greater un-
derstanding of the pathophysiological mechanisms behind
coagulation disorders such as haemophilia.
Haemophilia may be chamcterised by a failure of platelet-
surface thrombin generation in the final propagation stage
of the haemostatic process. High-dose recombinant FVIIa
(rFVIIa; NovoSeven@, Novo Nordisk, Copenhagen, Denmark)
has shown considerable success as a therapy for haemophil-
iacs and inhibitor patients, and this success may be due to
a mechanism of action involving platelet-surface FXa gener-
ation. This results in enhanced thrombin production, and
may partially compensate for the deficiency of FIX or FVIII.
Fig. 4. High-dose FVlla partially restores platelet-surface thrombin gen- The efficacy of high-dose rFVIIa in haemophilia and Inhibitor
eration in haemophilia. (Reproduced from Hoffman M., Monroe D.M. patients has led to its use in a growing number of alternative
3rd, Roberts H.R. Activated factor VII activates factors IX and X on the indications, and data regarding its mechanism of action in
surface of activated platelets: thoughts on the mechanism of action of such circumstances are scarce. However, we believe that
high-dose activated factor VII. Blood Coagul Fibrinolysis 1998; 9 (Suppl. even in the presence of the FIXa/FVIIIa complex, FVIIa may
I): S6 IS65, with permission.) be able to enhance both FXa and FIXa levels on the platelet
surface, thus augmenting the production of vital thrombin.
The second extrapolation from studies of FVIIa in the
experimental model is that the action of high-dose FVIIa
in vivo is not directly dependent on n, but is instead
plateletdependent. Earlier theories postulating a TFdepen-
dent mode of action for FVIIa explain the localisation of
FVIIa a$ti$ity to the injury site, which may account for the
relative lack of thrombotic complications observed during
highdose FVIIa therapy,’ but do not adequately justify the
requirement for high doses. lo However, a platelet-dependent
mechanism in which FVIIa binds to platelets with low aflin- References
ity allows not only foe the localisation of FVIIa activity, but
also explains why high doses are required to attain clinically I. Davie EW, Ratnoff OD. Waterfall sequence for intrinsic blood clot-
effective levels of thrombin generation. While this theory ting. Science 1964; 145: I 3 IO- I 3 12.
of platelet dependence does not preclude other actions and 2. Hoffman M. Mechanism of action of NovoSeven@using a cell-based
effects of FVIIa, it is consistent with empirically determined model. Bloodline Reviews 2002; I : 5-6.
dosing requirements. However, it is important to be aware 3. Veldman A. Hoffman M, Ehrenforth S. New insights into the coag-
that this mechanism is not truly TF-independent, as TF is still ulation system and implications for new therapeutic options with
required for the initiation of coagulation. The theory of a recombinant factor Vlla. Curr Med Chem 2003; IO: 797-81 I.
plateletdependent mechanism of action simply implies that 4. Hoffman M, Monroe DM 3rd. A cell-based model of hemostasis.
the primary effect of FVIIa occurs on the platelet surface. Thromb Haemost 2001; 85: 958-965.
i 5. Monroe DM. Roberts HR. Hoffman M. Platelet procoagulant com-
plex assembly in a tissue factor-initiated system. Br J Haematol 1994;
CONCLUSIONS 88: 364-37 I.
6. Mann KG. Potential analytes for the diagnosis of thrombosis. An
When compared to the traditional cascade hypothesis, we overview. Ann Epidemiol 1992; 2: 365-370.
believe that the cell-based conceptual model of haemosta- 7. Baglia FA, Badellino KO. Li CQ, Lopez JA, Walsh PN. Factor Xl bind-
sis allows a more fundamental understanding of the clinical ing to the platelet glycoprotein lb-IX-V complex promotes factor
problems observed in some coagulation disorders by focus- Xl activation by thrombin. J Biol Chem 2002; 277: I662- 1668.
ing on the central role of specific cell surfaces in controlling 8. Oliver ]A, Monroe DM, Roberts HR. Hoffman MR. Feedback acti-
and directing the haemostatic process. vation of factor XI on platelets in the absence of factor XII. Arte-
Our cell-based model builds upon the foundations laid by rioscler Thromb Vast Biol 1999; 19: 170-l 77.
the traditional cascade theory, but places greater emphasis 9. Hoffman M, Monroe DM 3rd. Roberts HR. Activated factor VII acti-
on the roles of specific receptors present on the surfaces of vates factors IX and X on the surface of activated platelets: thoughts
the cells involved. Importantly, the cell-based model suggests on the mechanism of action of high-dose activated factor VII. Blood
that understanding the structure and function of coagula- Coagul Fibrinolysis 1998; 9 (Suppl I): S6 I-S65.
tion proteins is necessary, but not sufficient, to understand IO. Monroe DM, Hoffman M. Oliver ]A, Roberts HR. Platelet activity of
haemostasis in vivo. Accordingly, this more recent model high-dose factor Vlla is independent of tissue factor. Br J Haematol
provides a potentially more accurate representation of the 1997; 99: 542-547.

0 2003 Elsevier Science Ltd. All rights mserved. Blood Reviews (2003) 17, 5 l-55 m