Keywords haemophilia - factor VIII - antibody
Schlüsselwörter Hämophilie - Faktor VIII - Antikörper
Introduction
The activated derivative of factor VIII (FVIIIa) participates in the coagulation cascade
as a nonenzymatic cofactor for activated factor IX (FIXa) in the generation of activated
factor X (FXa).[1 ] Its physiological relevance in this enzymatic reaction is illustrated by the notion
that the functional absence of FVIII predisposes to a severe bleeding diathesis, known
as haemophilia A.
For decades, the treatment of haemophilia A has mainly consisted of replacement therapy
using FVIII concentrates of varying purity. As of today highly purified plasma-derived
or recombinant FVIII concentrates are broadly used for the prophylactic treatment
of haemophilia A, thereby not only markedly reducing spontaneous bleeding events but
also enabling home therapy.[2 ]
[3 ] Although effective, replacement therapy comes with several disadvantages, like the
need for frequent intravenous applications and the development of anti-FVIII antibodies
in 20 to 30% of the patients with severe haemophilia A.[4 ]
[5 ]
[6 ] It is not surprising therefore that a search for alternative strategies to improve
haemophilia A treatment has been ongoing. In this review, we will focus on antibody-based
treatment options, and three examples will be discussed ( [Fig. 1 ]). First, a bispecific antibody (emicizumab) that combines FIXa and FX to enhance
FXa generation; second, antibodies that block the function of tissue factor pathway
inhibitor (TFPI), thereby promoting activity of the tissue factor/factor VIIa (TF/FVIIa)
and prothrombinase complex; and third, a FVIII–nanobody fusion protein that displays
prolonged survival and reduced immunogenicity.[7 ]
[8 ]
[9 ] Importantly, this review will limit itself to the functional and mechanistic aspects
of the antibodies, and will not discuss clinical efficacy.
Bispecific Antibody Emicizumab: Comparison with FVIII
Bispecific Antibody Emicizumab: Comparison with FVIII
Emicizumab (also known as ACE910 or Hemlibra) is a bispecific antibody that interacts
with FIX/FIXa and FX/FXa.[8 ] The rationale behind this antibody is that it brings the enzyme FIXa and its substrate
FX in close proximity, favouring conversion of FX into FXa. To some extent, this antibody
resembles FVIIIa, the cofactor within the tenase complex that is also capable of simultaneously
interacting with FIXa and FX. Nevertheless, emicizumab and FVIII are very different
proteins in terms of structure and regulation, and consequently they will act differently
during coagulation. To understand these differences, it is perhaps convenient to first
provide a reminder of how FVIII actually functions. We have recently described the
mode of action of FVIII in great detail,[10 ] and will therefore only provide a summary of the key points in this regard.
On/off switch : FVIII circulates as a precofactor that requires activation by thrombin and/or FXa
before it becomes an active cofactor. However, because FVIIIa is a very labile protein,
it will lose its function by just falling apart within minutes following activation.[11 ] Also, FVIIIa is an efficient substrate for activated protein C, which is able to
inactivate FVIIIa via site-specific proteolysis.[12 ] FVIIIa activity is thus regulated by an on/off switch.
Specific cofactor functions : FVIIIa's cofactor function consists of different explicit tasks. FVIIIa initiates
formation of the tenase complex by recruiting FIXa to the phospholipid surface, limiting
the movement of the proteins to a two-dimensional space.[13 ] While interacting with FIXa, FVIIIa stabilizes the protease domain of FIXa.[14 ] This interaction has two main functions: to reduce the flexible movements of the
surface loops that surround the active centre and to position the active centre at
the appropriate height from the phospholipid surface. Both functions are crucial to
optimize the catalytic capacity of FIXa. Once in place, the complex is ready to receive
FX. By interacting with FVIIIa, the activation peptide of FX is optimally aligned
to the active site of FIXa, which further promotes FXa generation.[15 ] Thus, multiple aspects of FVIIIa cofactor function define its enhancing potential
within the tenase complex.
Specificity : Although FIXa binds equally well to FVIII or FVIIIa (with this interaction only
being limited by the presence of von Willebrand factor [VWF]), the opposite is untrue.
FVIIIa (or FVIII) is unable to interact with the FIX zymogen.[16 ] This restricts the formation of the complex to one between the enzyme and cofactor,
rather than between the zymogen and cofactor. Furthermore, FVIIIa is able to interact
with FX, while less so with FXa.[15 ] Indeed, the low affinity for FXa compared with FX allows the release of FXa from
the FIXa/FVIIIa complex, making the tenase complex available for a next substrate
molecule. Hence, FVIIIa displays specificity toward both the enzyme and substrate.
Stimulating activity : The magnitude by which FVIIIa enhances the catalytic activity of FIXa is often determined
in assays using purified components. Accordingly, variables like phospholipid concentration
and composition, as well as concentrations of the proteins, are determinant for the
catalytic efficiency that is measured. In general, FVIIIa enhances FIXa activity by
1,000- to 1 million-fold.
Limiting factor within the tenase complex : To establish what is the limiting factor within the tenase complex, it is of relevance
to consider the respective protein concentrations. It is obvious that FVIII concentrations
(0.3–1 nM) are considerably lower than those of FIX (90 nM) or FX (135 nM). The actual
concentrations of FVIIIa and FIXa during in vivo coagulation are difficult to establish,
but even if we consider that 50% of FVIII is activated versus 5% of FIX, this would
still mean 9 to 30-fold less FVIIIa than FIXa (0.15–0.5 vs. 4.5 nM). It is thus fair
to assume that FVIIIa is the limiting factor within the tenase complex.
In view of this complex mechanism of action and regulation of FVIIIa, it is obvious
that the characteristics of the antibody-based protein emicizumab will be rather different.
A first important difference is that emicizumab has similar affinity for FIX versus
FIXa and for FX versus FXa.[17 ] This indicates that the specificity of FVIIIa toward FIXa and FX is no longer present
in the bispecific antibody. It is therefore possible that the following ternary complexes
will be formed in the circulation: emicizumab/FIX/FX, emicizumab/FIXa/FX (which is
the one that is relevant for FXa generation), emicizumab/FIX/FXa, and emicizumab/FIXa/FXa.
How much of each complex is being formed is very much dependent on the respective
concentrations of the individual constituents. Based on concentrations and respective
affinities, it is possible to calculate how much of each complex is formed. In the
normal circulation and at therapeutic emicizumab concentration of 55 μg/mL (370 nM),
the emicizumab/FIX/FX complex will be the most abundant, with ∼0.8 nM ternary complex.[17 ]
Interestingly, in normal individuals (and there is no reason to assume that this will
be different in haemophilia A patients), there is ∼120 pM of FIXa that is constitutively
present in the circulation.[18 ] Since emicizumab does not make any distinction between FIX and FIXa, only a small
portion of these 120 pM of FIXa will bind to emicizumab. Indeed, the ternary complex
relevant for FXa generation (emicizumab/FIXa/FX) present under basal conditions in
these treated patients can be calculated to be ∼2 pM. Of note, since emicizumab is not regulated by an on/off switch, this may indicate
that there is a constitutive background activity of this emicizumab/FIXa/FX complex.
It is not to be excluded that such low background activity may act already during
small subclinical bleeds that do not provoke a full trigger of the coagulation cascade.
An important question is of course how efficiently does emicizumab enhance FIXa activity
in comparison to FVIIIa? The antibody configuration of emicizumab will prevent it
from capturing FIXa and FX to the phospholipid surface (something that does not exclude
the activation reaction to be still phospholipid-dependent). Furthermore, with the
emicizumab binding site being located in the epidermal growth factor 1 (EGF1)-like
domain of FIXa, the bispecific antibody is also unlikely to modify the flexibility
and position of the FIXa protease domain to a significant extent. It is therefore
difficult to expect that emicizumab to be as efficient as FVIIIa in stimulating FX
activation. Experiments using purified components showed that emicizumab was approximately
10-fold less efficient in enhancing the catalytic efficiency by which FIXa activates
FX.[8 ]
[17 ]
Emicizumab: Monitoring Using Activity Assays
Emicizumab: Monitoring Using Activity Assays
To monitor emicizumab-based therapy in the patients, two issues are of relevance.
First, one might want to verify levels of emicizumab during therapy. In the current
clinical trials, a specific enzyme-linked immunosorbent assay (ELISA) system using
anti-idiotype antibodies has been applied.[19 ] Alternatively, one could also employ a chromogenic FVIII activity assay, with the
necessity of such an assay to contain human FIXa and FX (for example Biophen FVIII:C
by Hyphen Biomed), which uses emicizumab as standard. The main difficulty is to find
a so-called FVIII-equivalence using activity assays. Several assays are available
to measure FVIII activity, like the activated partial thromboplastin time (aPTT),
thrombin generation assays (initiated by TF or factor XIa), or clot waveform analysis.
Each of these assays is based on FVIII being the limiting factor, while the amount
of FIXa that is generated in these tests is quite variable from one test to the other,
and even depends on the activating reagent used. Knowing that emicizumab is present
in large excess over all other relevant components (FIXa and FX) and is always in
its “on”-mode, it will thus be the amount of FIXa present that will determine the
activity of emicizumab. With the FIXa concentrations being different from one assay
to the other, it will be complicated to assign a FVIII-like value to a certain emicizumab
concentration. Care should therefore be taken to extrapolate findings concerning emicizumab
in laboratory assays to the in vivo haemostatic potential of this molecule. We would
like to emphasize that this does not mean that emicizumab activity in vivo is per definition lower than what will be measured in laboratory assays. It cannot be excluded
that emicizumab has the intrinsic capacity to modulate the extent of FIXa formation
in a way that goes unnoticed in the regular activity assays. For example, it is known
that FXa is able to generate the FIX activation intermediate FIXα, which is a much
better substrate than FIX for the TF/FVIIa complex. Akin to emicizumab promoting formation
of FXa by FIX, it is theoretically possible that it also promotes FXa-mediated FIXα
formation, which in the presence of TF/FVIIa is then readily converted into FIXa.
This pathway could eventually explain why emicizumab seems to have higher activity
in TF-based thrombin generation assays compared with FXIa-based thrombin generation
assays.
Emicizumab in Combination with Other Procoagulant Agents
Emicizumab in Combination with Other Procoagulant Agents
Under certain conditions, emicizumab has been or will be used in combination with
other procoagulant proteins, such as FVIII concentrates or FVIII-bypassing agents
like factor eight inhibitor bypassing activity (FEIBA) or FVIIa. It is of interest
to consider how the co-use of such products impacts coagulation. In case of FEIBA
use, several issues need to be considered. First, FEIBA consists of many different
proteins, including high concentrations of prothrombin, FX, and FIX.[20 ] By adding FIX and FX to the circulation, one automatically increases levels of the
ternary complexes. A dose of 100 U/kg of FEIBA will more than double the concentration
of the ternary emicizumab/FIX/FX complex in the circulation, assuming a therapeutic
emicizumab level of 370 nM.[17 ] Furthermore, FEIBA contains considerable amounts of FIXa and a dose of 100 U/kg
may ultimately result in circulating FIXa levels that may reach 1 nM.[21 ]
[22 ] Since emicizumab makes no distinction between FIX and FIXa, and does not require
an on-switch to become active, the presence of FIXa may directly be used to trigger
FXa generation: the ternary, FXa-generating emicizumab/FIXa/FX complex may rise to
a concentration of 20 pM.
With regard to the use of FVIIa, it will be rather different. The use of FVIIa will
not change the basal FIX or FX levels, suggesting ternary complex formation remains
unaltered. However, FVIIa will cause the generation of both FIXa and FXa, thereby
increasing the participation of these proteins in the complexes with emicizumab. Primate
studies using a dose of 50 μg/kg FVIIa revealed that such a dose doubles the FIXa
concentration in the circulation from ∼120 to 240 pM.
[23 ] This suggests that that FVIIa may contribute to an increase in active emicizumab/FIXa/FX
complexes, albeit probably to a lesser extent compared with FEIBA.
As for the combination with FVIII, two distinct situations should be considered. First,
in patients having no anti-FVIII antibodies receiving regular doses of FVIII concentrates
(25–50 U/kg), FVIII will be present together with emicizumab and compete for the interaction
with FIXa. Since FVIIIa displays much higher affinity for FIXa than emicizumab (2–15 nM
vs. 1.5 μM), and will not be diluted to nonactivated FIX zymogen, it seems likely
that any FIXa present will first choose to partner with FVIIIa. However, also in these
situations, FIXa will be present in excess over FVIIIa. The remaining FIXa thus remains
available to interact with emicizumab. Depending on the excess of FIXa, there might
therefore be the formation of ternary complexes of emicizumab/FIXa/FX in addition
to the tenase complex. However, these ternary complexes are 11-fold less efficient
than the tenase complex in generating FXa. Will this be different under conditions
where patients receive immune-tolerance therapy? Such patients have inhibitory antibodies
against FVIII, and receive high doses of FVIII on a daily basis. Depending on the
nature of the inhibitory antibodies and their concentrations, they may prevent FVIII
from interacting with FIXa. As such, FIXa remains available for binding to emicizumab.
Given the polyclonal character of the antibodies in most patients, it is probable
that FVIII will not be interacting with FIXa in these inhibitor patients. Emicizumab
will therefore be able to assist FIXa during immune tolerance therapy.
Monoclonal Antibodies Targeting TFPI
Monoclonal Antibodies Targeting TFPI
TFPI is a Kunitz-type inhibitor, known to interfere with the initial phase of coagulation.[24 ]
[25 ] Two isoforms are present in humans: TFPIα, which contains three Kunitz domains and
a basic C-terminal region, and TFPIβ in which the third Kunitz domain and the basic
region are replaced by a glycosylphosphatidylinositol-binding sequence. The mode of
action of TFPI is perhaps less straightforward than many seem to appreciate.[24 ]
[26 ] At low concentrations, its Kunitz-2 domain interacts loosely with FXa, before isomerizing
into a tight complex. This complex is then able to rapidly inhibit the activity of
the TF/FVIIa complex, via interactions between the Kunitz-1 domain of TFPI and the
FVIIa active site. TFPI alone is a less effective inhibitor of TF/FVIIa compared with
FXa/TFPI.
The fact that TFPI plays an important role in the regulation of the initial phase
of coagulation has sparked interest related to its use as a target in the treatment
of haemophilia.[7 ] The rationale would be that by blocking TFPI, more FXa and FIXa will be generated
by the FVIIa/TF complex, which in turn will lead to increased thrombin generation
and subsequent fibrin deposition and platelet activation. Preliminary studies indeed
revealed that anti-TFPI antibodies shorten the coagulation time of haemophilic plasma
and reduce the bleeding tendency in a rabbit model of acquired haemophilia A.[27 ]
[28 ]
As to our knowledge, three monoclonal antibodies targeting TFPI are currently under
clinical development: PF-06741086 by Pfizer, BAY1093884 by Bayer Healthcare, and concizumab
by Novo Nordisk. Results from efficacy studies using these antibodies are awaited
with great anticipation. Meanwhile, it is of interest to reflect on the various aspects
of how such monoclonal antibodies affect TFPI biology. Several issues seem to be of
relevance in this regard. First, TFPI is not only an inhibitor of the FXa/FVIIa/TF
complex, but also recognized as a direct inhibitor of the prothrombinase complex.[29 ] At present, it is unclear which of the three clinical candidates is able to interfere
with TFPI-mediated inhibition of the prothrombinase complex, and how this affects
the prohaemostatic potential of such antibody.
The second issue relates to the biodistribution of TFPI. The majority of the TFPI
molecules are located at the surface and within endothelial cells, while only a small
percentage of TFPI is present in plasma, mostly associated to lipoproteins.[30 ] There is also a portion that is stored in platelets.[31 ] It will thus be difficult to monitor how much of the total TFPI population will
be inhibited by the therapeutic antibodies, while having only access to the patients'
plasma. Moreover, in vivo studies using mice with a combined tissue-specific deficiency
of TFPI and full deficiency of FVIII revealed that particularly platelet-stored TFPIα
is of particular relevance to compensate for the absence of FVIII.[32 ] At this point, it is uncertain which of the three monoclonal antibodies in clinical
development is able to specifically target platelet-stored TFPI. Lastly, we know that
the FVIIa/TF complex may exert functions beyond haemostasis, in part via protease
activated receptor 2 (PAR-2) mediated signalling processes.[33 ] For example, animal studies have revealed that cell-bound TFPI contributes to reduce
tumour metastasis and atherosclerosis.[34 ]
[35 ] Furthermore, TFPI may also inhibit other proteases that are not involved in haemostasis,
and the FVII-activating protease has been reported to be among them.[36 ] It would be important therefore to investigate how the prophylactic use of anti-TFPI
antibodies modulates the extra-haemostatic activities of the TF/FVIIa complex and
other proteases.
A Factor VIII–Nanobody Fusion Protein
A Factor VIII–Nanobody Fusion Protein
For our third example of how antibodies could be used in the development of novel
strategies in the treatment of haemophilia, we would like to focus on a specific kind
of antibodies, namely heavy chain-only antibodies.[37 ] Such antibodies are found in members of the Camelid family, which include camels
and llamas. These heavy chain-only antibodies lack the light chain that is present
in classic immunoglobulin G (IgG) molecules, and consequently all the antigen-binding
information is constrained within the variable heavy-chain region of the heavy chain-only
antibodies. This variable region (also known as VHH or nanobody) can thus be used
similar to the single-chain Fv fragments of classic IgGs. In isolated form, nanobodies
have a molecular weight of ∼15 kDa. Despite their small size, they can still display
high affinity for their target antigen. In the present example of this review, nanobodies
have been used as a fusion partner to FVIII.[9 ] Of course, many other possibilities to use these specific antibody fragments do
exist (see for instance reviews in Beghein and Gettemans 2017, Manglik et al 2017,
and Vincke and Muyldermans 2012).[38 ]
[39 ]
[40 ]
The rationale behind the use of nanobodies as a fusion partner for FVIII lies in the
observation that although FVIII binds with high affinity to VWF, there is still 2
to 5% of FVIII that circulates as free protein.[41 ]
[42 ] Free FVIII differs from VWF-bound FVIII in that it is cleared 6- to 10-fold more
rapidly. Moreover, VWF modulates the anti-FVIII immune response by affecting the pathway
by which FVIII is endocytosed by antigen-presenting cells.[43 ]
[44 ] The underlying hypothesis in the design of the FVIII–nanobody fusion protein was
therefore to create a FVIII molecule that forms a tighter complex with VWF, thereby
reducing the amount of free FVIII. To do so, nanobodies were generated against the
VWF D'-D3 domain, the region that contains the FVIII-binding site. These nanobodies
indeed display a ∼50- to 100-fold slower dissociation rate constant from VWF compared
with FVIII.[9 ]
Two copies of one of these nanobodies, i.e., KB-VWF-013, were cloned into the FVIII
complementary DNA (cDNA), replacing the nonfunctional FVIII B-domain.[9 ] The final protein, designated FVIII-KB013bv, was produced in mammalian cell lines
as a single chain protein, and analysed for function in in vitro assays. In terms
of activity, FVIII-KB013bv was having full cofactor activity in one stage and chromogenic
assays. FVIII-KB013bv further had a 25-fold higher apparent affinity for VWF compared
with wild-type FVIII (13 vs. 330 pM), confirming that the presence of the nanobodies
considerably improves binding to VWF. This increased affinity means that at a concentration
of 1 U/ml, more than 99.8% of FVIII-KB013 is bound to VWF, whereas this is ∼96.7%
for wild-type FVIII.[9 ]
The effect of improved VWF binding was evaluated in FVIII-deficient mice. When determining
circulatory survival, it was observed that FVIII-KB013bv was cleared 2-fold slower
than wild-type FVIII. The increased survival was further established by its capacity
to correct bleeding in FVIII-deficient mice 24 hours after a single intravenous injection,
whereas by using the same dose wild-type FVIII was no longer able to do so.[9 ] Thus, the FVIII–nanobody fusion protein is fully functional in vivo and has a prolonged
half-life. With regard to the development of anti-FVIII antibodies, we observed that
FVIII-KB013bv was less immunogenic than wild-type FVIII (15% of mice developed antibodies
against FVIII-KB013bv vs. 85% developed antibodies against FVIII).[9 ] The reason for this difference will require further studies, but may be related
to the fact that upon binding of the classic FVIII/VWF complex to antigen-presenting
cells, FVIII enters into these cells, whereas VWF remains mainly at the cell surface.[45 ] By having a FVIII molecule that sticks to VWF much more tightly, it is possible
that less FVIII is able to enter the antigen-presenting cells, resulting in the formation
of fewer anti-FVIII antibodies. In comparison to emicizumab and the anti-TFPI antibodies,
FVIII-KB013bv is not yet ready to enter clinical evaluation. Nevertheless, it represents
an interesting example of how antibodies can be used to optimize treatment in general
and that of haemophilia A in particular.
Conclusion
This review has presented three different ways of how antibodies signify novel strategies
to innovate the management of bleeding episodes in haemophilia. Importantly, the use
of antibodies is not limited to a simple inhibition of the function of its target.
Two of the three examples presented here, such as emicizumab that has procoagulant
cofactor activity and the FVIII–nanobody fusion protein that ameliorates the molecular
properties of the complex FVIII molecule, represent these options. The future will
teach us if other types of application will become available to further improve the
clinical management of haemophilia.
Fig. 1 Antibodies in haemophilia treatment
. The mechanisms of action of three different types of antibodies that are or may be
useful in the management of haemophilia are schematically depicted.
