Recombinant human Factor VIIa (rFVIIa) cleared principally by antithrombin following IV administration in Haemophilia Patients

Abstract

The objective of the present study was to evaluate the pharmacokinetics and the clearance pathways of rFVIIa after intravenous administration to haemophilia patients. Methods: Ten severe haemophilia patients were included in the study; all patients were intravenously administered a clinical relevant dose of 90 μg/kg (1.8 nmol/kg) rFVIIa. Blood samples were collected consecutively to describe the pharmacokinetics of rFVIIa. All samples were analysed using three different assays: a clot assay to measure the activity (FVIIa:C), an enzyme immunoassay (EIA) to measure the antigen levels (FVII:Ag), and a EIA (FVIIa-AT) to measure the FVIIa antithrombin III (AT) complex. Pharmacokinetic parameters were evaluated both by use of standard non-compartmental methods and by use of mixed effects methods. A population pharmacokinetic model was used to simultaneously model all three datasets. The total body clearance of rFVIIa:C was estimated to be 38 mL/h/kg. The rFVII:AT complex formation was responsible for 65% of the total rFVIIa:C clearance. The initial and the terminal half-life of rFVIIa:C was estimated to be 0.6 and 2.6 hours, respectively. The formation of rFVII-AT complex were able to explain the difference observed between the rFVIIa:C and the rFVII:Ag concentration. The non-compartmental analysis resulted in almost identical parameters.

2 Introduction

Recombinant human factor VIIa (rFVIIa; NovoSeven^®; Novo Nordisk A/S, Bagsværd, Denmark) has been available since 1996 for the management of bleeding episodes in haemophilia patients unresponsive to treatment with FVIII or FIX due to presence of neutralizing antibodies (inhibitors).

Both tissue factor (TF) mediated[1] and platelet surface-mediated[2;3] mechanism of action have been proposed to explain the haemostatic effect of rFVIIa. Evidence is however accumulating that the dominant mechanism of action for rFVIIa in haemophilia is on the platelet surface[4;5]. At clinical doses, rFVIIa binds to the surface of activated platelets at the site of injury, and is able to directly activate factor X and thereby enhance local thrombin generation and the formation of a localized stable haemostatic plug.

Even though the pharmacokinetics (PK) of rFVIIa have been well characterised in healthy volunteers[6;7], in haemophilia patients with and without inhibitors[8], and in trauma patients with severe bleeding episodes[9], the clearance mechanism in humans is poorly understood. A few studies in animals have addressed the in vivo clearance mechanism of rFVIIa. For example, in rats, distribution studies suggest that rFVIIa is widely distributed in the body, with the highest rFVIIa concentrations observed in highly perfused organs, such as the liver[10]. Also in rats, histopathological examinations of the liver suggest that rFVIIa is cleared by hepatocytes and kuppfer cells via clathrin mediated endocytosis[11]. Recent data suggest that rFVIIa associates with the vascular endothelium, specifically with the endothelial cell protein C receptor (EPCR)[12], and perhaps enters into the extravascular space[13], however the clinical significance of these findings are not clear as the amount and the activity status is unknown.

An important limitation of the distribution studies is the fact that they do not distinguish between free rFVIIa and inactivated rFVIIa. It is well known that several different inhibitors found in the circulation may bind to rFVIIa in the circulation (e.g. antithrombin (AT) and tissue factor pathway inhibitor (TFPI)). Studies in mice and dogs demonstrated that rFVIIa-AT complexes accumulated in-vivo after intravenous rFVIIa administration. In the mice, after 3 hours the circulating rFVIIa-AT complexes accounted for significant part of the total antigen left in the circulation[14]. These results suggest that AT plays a significant role in the clearance of rFVIIa:C from the circulation.

The objectives of the present study were to characterize the PK of rFVIIa in haemophilia patients, and investigate the importance of rFVIIa-AT complex formation as a clearance mechanism for rFVIIa.

3 Materials and Methods

3.1 Study design

The Virginia Commonwealth University (VCU) Institutional Review Board approved this study prior to subject enrolment, and the study was conducted in compliance with the Declaration of Helsinki. Following the provision of written informed consent, 10 subjects > 18 years of age with severe FVIII or FIX deficiency with or without inhibitor antibodies were enrolled into this study. All participants were otherwise healthy and in a non-bleeding state. Participants were excluded if they had used any factor replacement therapy within the previous 72 hours before study entry; had active bleeding; or had a medical- or family history of thrombosis. Demographics, medical history, factor deficiency information, and concomitant medications were recorded for each participant. Anonymous identification numbers were used in order to maintain confidentiality.

One week following an initial screening visit, participants meeting eligibility criteria were admitted to the VCU General Clinical Research Center for a rFVIIa 6-hour pharmacokinetic (PK) – pharmacodynamic (PD) dosing study. Each participant had a peripheral line inserted and received rFVIIa (Novo Nordisk A/S, Bagsværd, Denmark) 1.8 nmol/kg (90 mcg kg-1) body weight intravenously. Blood specimens were collected at baseline and at 0.5, 1.0, 2.0, 4.0 and 6.0 h post-dose in 3.2% sodium citrate evacuated containers to document rFVIIa coagulant activity, rFVII antigen concentration, and complex concentration (rFVIIa-AT) at each time point.

3.2 Analytical procedure

FVII antigen (FVII:Ag) concentration was measured by FVII-EIA (DakoCytomatic, Dako, Ejby, Denmark). A FVIIa-AT EIA (Diagnostica Stago, Asnieres-Sur-Seine, France) was used to measure rFVIIa-AT antigen concentration. The assay uses a monoclonal anti FVIIa antibody that does not block AT binding for capture of the complex. A polyclonal anti-human AT antibody peroxidase – conjugate was used for detection. A pre-formed purified complex of human rFVIIa-AT was used as a standard for the EIA calibration. Further more, it was demonstrated, that the rFVIIa-AT complex gave the same signal in the FVII:Ag assay as rFVIIa. The anti AT antiserum is obtained by immunising rabbit with purified human AT (one single band of 65 KD and antigenic cofactor II < 0.02 PEU / PEU of AT). The antibodies produced give a single precipitating line for plasma and no precipitating line for AT deficient plasma. The IgGs are purified by sulphate precipitation and ion exchange (personal communication by the manufacturer).

FVII:C was analyzed using the Staclot® VIIa-rTF assay (Diagnostica Stago, Asnières-Sur-Seine, France) applied with minor modifications. A buffer containing 0.1% bovine serum albumin was used to dilute plasma samples, controls and calibrator. rFVIIa was diluted to a final concentration of 300 mIU /mL and calibrated against the international FVIIa standard (89/688, International Institute of Biological Standards and Controls, United Kingdom). Because the FVII:C assay applied truncated recombinant soluble thromboplastin that did not form complexes with the FVII zymogen, it ensured assessment of FVIIa exclusively. The limit of quantification was estimated to be 100, 8.4 and 5.0 pM for FVII:Ag, FVIIa:C and rFVIIa-AT complex, respectively.

3.3 Pharmacokinetic methods

The pharmacokinetic parameters were assessed both by standard non-compartmental analysis (NCA) and by non-linear mixed effects methods. All data conversions between IU, grams and molar units were performed assuming: 1 IU/mL = 20 ng/mL and 1 nm/L = 50 ng/mL.

3.3.1 Non-compartmental analysis (NCA)

The NCA pharmacokinetic parameters for the individual subjects were determined using WinNonlin (Version 5.2, Pharsight, Mountain View, CA, USA). All calculations were based on baseline corrected values. The areas under the plasma rFVIIa concentration versus time curves were calculated according to the trapezoidal rule. The infinite part of the curve was determined as C_last/λ_z with C_last defined as the last concentration and λ defined as the terminal elimination rate constant. The peak concentration (C_max) and time to reach the peak concentration (t_max) were read from the individual plasma concentrations versus time curves.

3.3.2 Non-linear mixed effects analysis

The pharmacokinetic parameters were assessed by non-linear mixed effect modeling through the NONMEM program, version VI (GloboMax/ICON, Ellicott City, MD, USA), with the Compaq Visual Fortran version 6.5. The conditional first-order method was used for all models tested. Several different models with a different number of compartments and input functions were tested. The FVIIa baseline was included in the modelling, for the individual subjects it was fixed to the concentration observed pre-dose.

The clearance was modelled by dividing it into two fractions: AT-bound and non-AT-bound. The AT-bound clearance controlled the flux of rFVIIa into a rFVIIa-AT complex compartment. The difference observed between the activity and the antigen curve was assumed to be caused by AT-complex formation, and it was assumed the rFVIIa-AT complex formation was irreversible. Plasma concentrations of activity, antigen and AT-complex data were fitted simultaneously. The EIA assay is known to measure both the endogenous FVII:Ag, rFVII:Ag and AT-bound fraction; thus the antigen concentration was assumed to be the sum of activity and the AT-complex concentration. If this approach did not provide an adequate description of the available data, additional clearance routes were tested. The inter-individual variability was modelled by an exponential error model and co-variances between parameters were considered. Combined additive and proportional intra-individual error models were tested. The goodness of fit was evaluated by graphic analysis of, e.g., predicted versus observed concentrations (distribution of the points around the unity line) and by weighted residuals versus predicted concentrations using the R software package[15]. Individual parameter values were obtained as empirical Bayes estimates from the final population model.

4 Results

Nine severe FVIII- and one FIX deficient patient were included in the study, the age ranged from 23-60 years; all patients had normal platelet count and fibrinogen concentration at screening and throughout the duration of the study. No adverse events were observed during the study. After intravenous administration of a clinical dose of rFVIIa, there was a difference between the antigen and the activity profile, and rFVII-AT complexes were formed over time (Figure 1). The pre-dose levels of FVII:Ag, FVIIa:C and FVIIa-AT were (Mean±SD) 8.9±1.6, 0.03±0.01 and 0.10±0.05 nM, respectively, corresponding to about 0.4% of the circulating FVII is in its active form. Thus, the measured FVIIa activity mainly represents the activity of the injected rFVIIa, while the measured antigen levels consist of both endogenous FVII and rFVIIa. Pre-dose levels of FVII-AT complex were 0.1 nM. It has been reported that about 99% of FVII in the plasma of normal individuals exist in the zymogen form, while approx 1 % circulates in the activated form[16;17]; however the normal range has been reported to be quite broad (0.5-8.4 ng/ml)[17]. Lower levels have been reported in patients with haemophilia, 60% and 10% of normal in haemophilia A and B respectively[18], this is in agreement with the present study. The baseline level of FVIIa-AT complexes has previously been reported to be 0.16 nM in healthy subjects[19], in this study we report a mean value of 0.10 nM, the minor difference observed may be caused by either differences in the assay calibration or by differences between the populations tested.

Figure 1.

Mean plasma concentration of rFVIIa:C (blue), baseline corrected rFVII:Ag (black) and rFVII-AT (red) after intravenous administration of 1.8 nmol rFVIIa/kg (∼90 μg/kg) to severe haemophilia patients (n=10). Lines represent mean concentration, open circles the individual observed plasma concentration levels.

After IV rFVIIa administration the plasma concentration versus time curves were best described by a two-compartmental model (Figure 2). The parameters of the model include: the inter-compartment flow rate (Q₂), the volumes of the central and the peripheral compartments (V₁, and V₂), and the volume of the complex compartment (V₃). Since both activity and complex plasma concentrations were assumed to take place in the central compartment, V1 was assumed to be the same as V3. The rFVIIa clearance was split into to fractions: the AT mediated clearance (CL_AT) and the non-AT clearance (CL_NON-AT), the total clearance (CL) equals the sum of CL_AT and CL_NON-AT. Furthermore the clearance of the rFVIIa-AT complex was estimated (CL_FVIIa-AT). Based on this combined model it was possible to fit the activity, antigen and complex data simultaneously. Between subject variability for the pharmacokinetic parameters were evaluated on all clearance and volume terms, however, due to the low variability of the observed data, including an between subject variability was considered not to add benefit to the model output. The within subject variability was estimated to be in the range of 31-47% for the three datasets.

Figure 2.

The model used to fit the data. The model consisted of three compartments (V1, V2 and V3). V1 and V3 represented the central plasma compartment for rFVIIa clot activity and AT complex antigen concentration, respectively. The clearance was split into two pathways, one representing the AT inactivation (CL_AT) and an unspecific fraction (CL).

The applied model fit the data well (Figure 3). The rFVIIa-AT complex formation was able to explain the difference observed between the rFVIIa activity and antigen levels. Over the 6 hour time period, there appeared to be a slight drift in baseline. For example, at time zero adding the FVIIa:C and the FVIIa-AT equals the FVII:Ag parameter, but 6 hours after dosing there is a slight deviation suggesting a minor drift in the baseline. This change was implemented in the model as simple linear drift over the 6 hours of observation, and the model suggested that the drift in baseline equals 0.3 nM/h (Table 1). The CL of rFVIIa:C was estimated at 38 mL/h/kg; consisting of a fraction being cleared by AT inactivation of 25 mL/h/kg, and an unspecific fraction of 13 mL/h/kg. The rFVIIa-AT pathway accounted for 65% of rFVIIa CL. The inter-compartmental clearance (Q) was estimated to be 23 mL/h/kg and the volume of the peripheral compartment (V2) to be 35 mL/kg. The initial and the terminal half-life parameters were estimated to be 0.6 and 2.6 hours, respectively (Table 1).

Figure 3.

Goodness of fit plots after intravenous administration of 1.8 nmol/kg (∼90 μg/kg) to severe haemophilia patients (n=10). All data were modelled simultaneously by use on NONMEM. Blue open dots represent observed plasma concentration, red closed dot mean of observed plasma concentration, and the blue curve the model fit. rFVIIa:C = clot activity; rFVIIa-AT = rFVIIa anti-thrombin III complex concentration; rFVII:Ag = rFVII antigen concentration (includes endogenous FVII, rFVIIa and rFVIIa-AT).

Table 1.

Pharmacokinetic parameters of rFVIIa based on non-linear mixed effects modelling after intravenous administration of 1.8 nmol rFVIIa/kg (∼90 μg/kg) to severe haemophilia patients (n=10).

Variable	Unit	Estimate	%RSE
CL	ml/h/kg	13	19
V1 & V3	ml/kg	78	8.1
Q	ml/h/kg	23	24
V2	ml/kg	35	14
CLAT	ml/h/kg	25	17
CLCOMP	ml/kg	24	14
Slope	nM/h	-0.3	9.7
CLTOT	mL/h/kg	38	-
CLAT	%	65	-
t½ alfa clot	h	0.6	-
t½ beta clot	h	2.6	-
k13 (AT)	h-1	0.3	-
AT-FVIIa inhibition	M-1·s-1	34	-

RSE%: Relative standard error

CL_NON-AT: Non AT rFVIIa Clearance

V1: Volume of central compartment

V2: Volume of peripheral compartment

V3: Volume of the rFVIIa-AT compartment

Q: Inter compartmental flow between V1 and V2

CL_AT: Clearance of rFVIIa by the AT inactivation

CL: Total clearance (CL_NON-AT + CL_AT)

CL_FVIIa-AT: Clearance of the rFVIIa-AT complex

Base: Slope for drift in FVII:Ag baseline over time

%CL_AT: Percentage of total cleared by AT pathway (100 · CL_AT/CL)

t_½α: Initial half-life

t_½β: Terminal half-life

K₁₃ (AT): Rate constant for complex formation

AT-FVIIa inhibition: Second order rate constant for inhibition of FVIIa by ATIII, calculated using a baseline ATIII concentration of 2.6 μM[28]

The PK parameters assessed by NCA methods were similar to the parameters based on the two-compartment model. Due to the limited number of samples, it was not possible to estimate the extrapolated part for the rFVIIa-AT complex, therefore only a limited number of parameters are available for the rFVIIa-AT complex (Table 2). The CL for rFVIIa:C was estimated to be 39 mL/h/kg and the terminal half-life was estimated at 2.1 hours.

Table 2.

Mean (±SD) pharmacokinetic parameters assessed by non-compartmental analysis after intravenous administration of 1.8 nmol rFVIIa/kg (∼90 μg/kg) to severe haemophilia patients (n=10). Pharmacokinetic parameters were based both on a rFVIIa antithrombin complex assay (rFVIIa-AT), baseline corrected rFVII antigen assay (rFVII:Ag) and rFVIIa activity assay (rFVIIa:C).

Variable	Unit	rFVIIa-AT	rFVII:Ag	rFVIIa:C
AUCt	h·nmol/L	28 ± 5	63 ± 10	43 ± 11
CL	mL/h/kg	-	23 ± 5	39 ± 9
Vz	mL/kg	-	88 ± 12	116 ± 26
Cmax	nM	5.8 ± 1.1	19 ± 3.2	16 ± 4.7
MRT	h	-	3.7 ± 0.4	2.6 ± 0.2
t½	h	-	2.7 ± 0.4	2.1 ± 0.2

AUC_t: Area under the curve from time 0 to last sampling point

CL: Clearance

V_z: Volume of distribution

C_max: Maximum plasma concentration

Recovery: Expected Cmax /observed Cmax

MRT: Mean residence time

t_½: terminal half-life

5 Discussion

Previous studies in mice have suggested that inactivation of rFVIIa by anti-thrombin III may play a significant role as a clearance mechanism[14]. In order to further investigate these findings in the clinical setting, we dosed 10 haemophilia patients with rFVIIa 1.8 nmol/kg (90 μg/kg) and collected blood samples over 6 hours. Serial blood samples were analyzed for rFVIIa activity, rFVII antigen and a rFVIIa AT complex (Figure 1) and the PK were assessed using both NCA and a 2 compartmental model. Overall the pharmacokinetic parameters reported in this study (Table 1, Table 2) agree well with previously published data in patients with severe bleeds [9], in haemophilic patients[8] and in healthy subjects[20]. The minor differences observed in the PK parameters may be attributed to differences in the late sampling schedule.

Based on a simultaneous modelling approach, it was possible to divide the total clearance of rFVIIa:C into two fractions, one being inactivation in plasma by AT complex formation and one being an unspecific clearance. We demonstrated that the differences in clearance observed between rFVIIa:C and rFVII:Ag is explained by rFVIIa-AT complex formation. Prior studies suggest the liver is a major clearance organ for rFVIIa[10;11;21]. However, whether rFVIIa is cleared by the liver directly or in a complex is not known. The present study suggests that approximately 65% of the total body clearance of rFVIIa:C is accounted for by AT complex formation. The circulation complex may then be cleared by the liver and further processed and excreted. It should be noted that the clearance of the complex (rFVIIa-AT) and the clearance of rFVII:Ag is almost identical (Table 1, Table 2), which corroborates earlier findings in mice[14]. The remaining 35% of the clearance could eventually be facilitated by other inhibitors such as TFPI, α2-macroglobulin, or other yet unidentified inhibitors, this still remains to be elucidated.

Several in vitro studies have shown that the inhibition of FVIIa by AT is a slow process and requires the stimulatory effect of heparin and tissue factor[22;23], with a second-order association constant around 1.2 M^-1·s^-1[23]. In the present clinical study we report a second-order association constant of 34 M^-1·s^-1, which is substantial higher than that reported from in-vitro experiments, but is in agreement with previous findings in dogs[24]. The reason for the difference observed between in-vitro and in-vivo is not clear, it may be speculated that in the circulation the interaction between FVIIa and the glycans lining the vascular bed could influence the regulation and inactivation of FVIIa by AT.

The knowledge regarding the regulatory role of AT in the clearance mechanism of FVIIa may also be useful in designing new variants with specific profiles. In NN1731, a rFVIIa analogue, where three amino acid changes affecting the active conformation have lead to a molecule which demonstrates a high efficacy and more rapid onset of action in haemophilia mice models[25]. NN1731 also shows an increased susceptibility to inhibition by AT leading to a faster clearance[24]. The interaction with AT can also be reduced leading to rFVIIa with longer plasma half-lives as seen with glycoPegylated rFVIIa derivatives[26;27], where it has been shown that the introduction of a bulky PEG group interfers with the binding to AT and results in a substantial prolongation of the terminal half-life.

In conclusion, the results from this study suggest that complex formation between rFVIIa and AT is the main clearance pathway for rFVIIa in accounting for 65% of an intravenous dose of 90 μg/kg rFVIIa to haemophilia patients. The difference between the activity and antigen profile for rFVIIa in the haemophilia patients can be explained by AT complex formation. Further studies are necessary to explain how AT in the circulation is capable of acting as an efficient inhibitor of rFVIIa or how inhibition of FVIIa by AT is regulated.