Correction of human hemophilia A whole blood abnormalities with a novel bypass agent: zymogen-like FXa^I16L

Summary:

Background:

Approximately 30% of hemophilia A (HA) and 5% of hemophilia B patients develop inhibitors to protein replacement therapy which is the major cause of disease-related morbidity in the developed world. We previously developed zymogen-like factor Xa (FXa) molecules with impaired active site maturation enabling a greater half-life than wild type FXa while maintaining full procoagulant function when assembled in the prothrombinase complex. Here we evaluated the ability of zymogen-like FXa^I16L to correct whole blood ex vivo thromboelastometry coagulation abnormalities of severe HA subjects with and without inhibitors.

Methods:

14 severe HA subjects, including 5 with inhibitors, were enrolled at hemostatic baseline (FVIII:C <1%) >5 half lives from factor or bypass therapy. Subjects’ whole blood was evaluated by rotational thromboelastography (ROTEM^®) using INTEM analysis with varying concentrations of FXa^I16L or recombinant factor VIIa (rFVIIa).

Results:

FXa^I16L dose dependently corrected clot time (CT) in subjects with and without inhibitors. Following addition of 0.1 nM FXa^I16L, CT among HA subjects without and with inhibitors (mean=172, 95% CI=155–189 and mean=174, 95% CI=124–224, respectively) were not significantly differ from control (mean=164, 95% CI=157–171). Addition of 20 nM rFVIIa, simulating a 90 μg/kg dose, resulted in prolonged CTs for HA subjects without and with inhibitors (mean=326, 95% CI=272–381 and mean=255, 95% CI=199–310, respectively) that were similar pretreatment CT times for both groups.

Conclusions:

FXa^I16L restored thromboelastometry CT to control values in severe HA subjects with and without inhibitors. Findings corroborate previous animal data and demonstrate the first evidence of zymogen-like FXa^I16L to correct human HA subject whole blood abnormalities and support the use of FXa^I16L as a novel hemostatic agent.

Introduction:

Congenital hemophilia is characterized by deficiency of factor VIII (hemophilia A [HA]) or factor IX (hemophilia B [HB]). Patients incur sequelae of bleeding resulting from inadequate production of activated factor X (FXa) and ultimately thrombin. Specifically, factor VIII (FVIII) and factor IX (FIX) maintain hemostasis through their essential role in the intrinsic pathway converting zymogen FX to the active protease FXa. Thereafter, the assembled prothrombinase complex (membrane-bound FXa and its cofactor factor Va [FVa]) converts prothrombin to thrombin, which then activates platelets and converts fibrinogen to insoluble fibrin, yielding a hemostatic plug [1,2]. Therapeutically, this aberration in hemophilia patients is restored by peripheral administration of plasma derived or recombinant protein products [3,4].

Despite the overwhelming positive impact of protein replacement therapy, up to 30% of patients with HA and 5% of patients with HB develop neutralizing alloantibodies (inhibitors) to protein replacement which now accounts for the major cause of disease related morbidity and mortality in developed nations [5,6]. Currently available bypassing agents (i.e. recombinant activated factor VII [rFVIIa] and activated prothrombin complex concentrates [aPCCs]) are directed at FXa production and thereby thrombin formation [7,8]. These strategies to bypass the defective intrinsic pathway are viable to achieve hemostasis in most patients with inhibitors, but are not universally effective and do not completely normalize thrombin generation [9–11]. Additionally, bypass therapies have associated prothrombotic risk (particularly if management requires tandem bypassing agents), in the case of aPCCs, are plasma-derived with intendant risks of blood borne disease, and need for frequent infusions results in costly treatment [9,12–14].

As an alternative, administration of FXa to increase prothrombinase complex formation would represent a direct approach for thrombin formation. However, wild-type (wt)-FXa is limited by its rapid inactivation by physiologic inhibitors resulting in a short half-life (<1–2 minutes). Further, the ability of the free protease to activate a range of procoagulant clotting factors with possible pathological activation of coagulation could also be problematic [15–17]. Collectively, these realities preclude the use of wt-FXa as a bypassing therapeutic.

Drawing from the known biochemical properties common to all chymotrypsin-like serine proteases, we previously developed FXa variants (e.g. FXa^I16L, chymotrypsinogen numbering system [18]) with impaired conformational transition from zymogen to active protease [19–21]. For FX, zymogen cleavage between Arg¹⁵-Ile¹⁶ results in a new N-terminus consisting of the conserved amino acid sequence Ile¹⁶-Val-Gly-Gly. The insertion of the nascent N-terminus into a binding pocket followed by salt bridge formation with Asp¹⁹⁴ confers a conformational change driving the zymogen to the active protease state, critical for full enzymatic function. Modification of FXa at Ile¹⁶ and Val¹⁷ results in an immature active site, thereby altering the protein to adopt a zymogen-like state. This effectively causes a redistribution of the zymogen-protease equilibrium that normally lies towards the mature protease. As a consequence of this altered conformation, FXa^I16L is less susceptible to plasma protease inhibitors and therefore has a prolonged half-life (>30 minutes). Importantly however, FVa preferentially binds the protease conformation of zymogen-like FXa effectively ‘rescuing’ the active protease through the principle of mass action. Evaluation of FXa^I16L in murine hemophilia models has not demonstrated evidence of systemic activation of coagulation or undesired thrombus formation [20–21]. The net effect is full procoagulant function and normal thrombin generation demonstrated in both ex vivo and in vivo murine HB and HA models [19–21].

Here we examined whether these FXa variants may be effective procoagulants for hemostatic management of hemophilia patients using an ex vivo approach. This work follows demonstrated efficacy in animal studies and is the first human data. We evaluated if zymogen-like FXa^I16L corrects whole blood thromboelastometry hemostatic abnormalities observed in HA subjects with and without inhibitors and compared the results to the most widely used bypassing agent, rFVIIa.

Methods:

The Children’s Hospital of Philadelphia Institutional Review Board approved participant recruitment for this study. Signed informed consent (parent with child assent where appropriate) was obtained prior to participation. Fourteen severe HA subjects without inhibitors (HA) and five with inhibitors (HA-I) were prospectively and consecutively recruited during outpatient visits to the Hemophilia and Thrombosis Center at CHOP. Subjects were enrolled at hemostatic baseline (FVIII:C <1%). Factor VIII activity and inhibitor values (Bethesda assay) were determined from the same blood draw at sample collection. Patients were excluded from this study if they were <1 year, within 5 half lives of factor replacement or bypass therapy, and/or had a known or suspected secondary hemostatic abnormality.

Blood was collected via peripheral venipuncture into a 3.8% sodium citrate vacutainer. Subject whole blood hemostatic abnormalities were assessed by ROTEM^® thromboelastometry using the INTEM^® assay. The INTEM reagent, an intrinsic pathway activator comprised of kaolin, in the presence of calcium chloride, initiated coagulation. Experimental conditions varied only with respect to the addition of supplemental protein, which included either: varying concentrations of FXa^I16L, 20 nM rFVIIa, FVa with or without 2 nM FXa^I16L, or no supplemental protein (buffer). Thromboelastometry assay analysis was uniformly initiated one hour following peripheral venipuncture. Control samples from five hemostatically normal subjects on no medications were collected and analyzed under the same conditions as study subjects. One-way analysis of variance (one-way ANOVA) was used to evaluate INTEM clot time differences among controls, HA subjects and HA-I subjects for each experimental condition. Adjusted p-values were computed using the Tukey-Kramer procedure, p-values <0.05 were considered significant. Equality of HA, HA-I and control group variances were evaluated by Levene’s test.

Results/Discussion:

Fourteen HA subjects and 5 HA-I subjects were enrolled. All subjects were male ages 1 to 11 years with confirmed FVIII:C <1% at the time of blood draw. Among the 5 subjects with inhibitors, ages ranged from 3 to 11 years with inhibitor values at time of enrollment of 26–96 Bethesda Units (B.U.). Five control subjects were enrolled. All control subjects were adults. Control INTEM assay parameters were consistent with published normal adult values. Of note, adult INTEM clot time (CT) normal values, minimally, if at all, differ from published >6 months pediatric normal values [22,23].

As expected, CT was the most abnormal parameter in the HA subjects and thus, the focus of our analysis under varying experimental conditions. Figure 1 presents CT times for the control group at baseline as well as HA and HA-I groups at baseline and in the presence of 0.1nM FXa^I16L and 20nM rFVIIa.Among HA and the HA-I subjects, pre-treatment INTEM CTs were significantly greater than control CT (Figure 1; p=0.006, HA mean=381, 95% CI=[308–454 seconds; and p=0.001, HA-I mean=289, 95% CI=289–669 seconds; control mean=164, 95% CI=157–171 seconds). HA and HA-I baseline CTs were not significantly differ (p=0.278). There was no significant difference in group variances, (Levene’s Test, p=0.254). FXa^I16L dose dependently decreased INTEM CT in both HA and HA-I subjects. Specifically, after the addition of 0.1 nM FXa^I16L to HA and HA-I subject samples, there was no significant difference between HA and HA-I CT from control CT (F-Statistic 1.36, p=0.841; HA mean=172, 95% CI=155–189 seconds]; HA-I mean=174, 95% CI=124–224 seconds]. Further, Levene’s test for equality of variances was not significant, p=0.277. Thus, irrespective of the presence or absence of inhibitors, addition of 0.1 nM FXa^I16L corrected hemophilic subject CT to control findings.

Figure 1.

Comparison of severe HA subject findings in the presence of FXa^I16L and 20 nM rFVIIa relative to control subjects. HA= hemophilia A; HA-I= hemophilia A subject with inhibitor; FXa^I16L= factor Xa^I16L; ns= not significant; rFVIIa= recombinant activated factor VII; CT= clot time; nM= nanomolar. Box plots represent median, 25% and 75% interquartile ranges; whiskers represent ± 2 standard deviations from the mean. Reported p-values are adjusted using the Tukey-Kramer procedures after a one-way analysis of variance to evaluate differences from control clot time findings.

Next the procoagulant effect of rFVIIa was assessed in 9 of the 14 HA subjects and the 5 HA-I subjects. A concentration of 20 nM rFVIIa was chosen for comparison to approximate plasma concentration achieved after administration of a dose of 90 μg/kg, which is commonly employed for bleeding manifestations in hemophilia patients with inhibitors. Although the addition of 20 nM rFVIIa decreased CT in all subjects, CT remained significantly prolonged relative to controls for both HA and HA-I subjects (p=0.002, mean=326,95% CI=272–381 seconds; and p=0.046, mean=254, 95% CI=199–310, respectively; control mean=164, 95% CI [157–171 seconds]). As expected, in the presence of 20nM rFVIIa, the post-treatment CT did not significantly differ between HA and HA-I subjects (p=0.0761) and Levene’s test for equality of variances was not significant (p=0.123).

Figure 2a shows representative tracings from an HA subject with inhibitor, with the addition of 0.5 and 0.1 nM FXa^I16L compared to 20 nM rFVIIa. The delayed and blunted tracing profile of the HA inhibitor patient’s blood was essentially restored with 0.1 nM FXa^I16L and was nearly indistinguishable from that of normal control blood. The incomplete response with 20 nM rFVIIa highlights the superior effectiveness of rFXa^I16L on a molar basis; however, the data must be interpreted with caution, as conditions in the assay do not simulate in vivo conditions such as drug volume of distribution, complete phospholipid membrane binding surfaces or tissue factor availability. Nevertheless, the data are generally consistent with our prior in vivo mouse studies demonstrating the enhanced effectiveness of murine rFXa^I16L relative to rFVIIa [21].

Figure 2.

Panel A: Sample ROTEM tracings demonstrate FXa^I16L dose response relative to rFVIIa. HA= hemophilia A; B.U.= Bethesda Units; FXa^I16L= factor Xa^I16L; rFVIIa= recombinant activated factor VII; CT= clot time; nM= nanomolar. Panel B: HA subject with an inhibitor: ROTEM tracings in the presence of FXa^I16L and FVa. HA= hemophilia A; B.U.= Bethesda Units; FXa^I16L= factor Xa^I16L; FVa= Factor Va; CT= clot time; nM= nanomolar.

Traditional clotting assays employed in medical practice may predict some measure of clinical outcome but neither adequately captures the hemostatic effect of bypassing agents nor allow for comparison of the hemostatic effect of various bypassing agents. Within the hemophilia community, there is growing interest in the potential use of global viscoelastic assays (e.g. thromboelastography/thromboelastometry) for monitoring therapeutic interventions in hemophilia, particularly those on bypassing therapy [24]. As such, we felt that the ROTEM^® INTEM assay would optimize ex vivo capacity to both observe the hemostatic effect of FXa^I16L and allow for comparison to other bypass strategies, i.e. rFVIIa. Additionally the use of a kaolin based coagulation initiator (INTEM reagent) appears better than tissue factor at discriminating the effects of rFVIIa in hemophilia thromboelastography analysis; however, notably these findings are generally limited to single center studies [25,26]. Thromboelastography clot time (CT), like thromboelastogram R time, is thought to be a measure specific to alterations in coagulation factor protein function and quantity and least influenced (relative to other parameters) by fibrinogen, platelet quantity or function [27].

Although limited by ex vivo analysis, our findings provide initial evidence in human whole blood and corroborate animal data demonstrating FXa^I16L is able to correct hemostatic abnormalities observed in murine models of hemophilia [20,21]. Additionally ex vivo findings of this work are consistent with our previous in vivo observations supporting that FXa^I16L is able to restore hemostasis at much lower concentrations than rFVIIa [21]. A clear difference between rFVIIa and rFXa^I16L are their half-lives (2–3 hours vs. 30 minutes, respectively). While this is an apparent limitation, at present it is unclear how half-life, when coupled to hemostatic effectiveness of the product, would actually impact clinical outcome.

Lastly, due to the requirement of FVa to ‘rescue’ the protease conformation of FXa^I16L, we speculate the procoagulant function of zymogen-like FXa will be limited by the availability of FVa. Therefore, FVa is thought to impart procoagulant injury site specificity and potentially protect against off target thrombosis. To probe this further, in concurrent but separate experiments, we titrated increasing concentrations of FXa^I16L in HA patient whole blood. At FXa^I16L concentrations above 2 nM, CT no longer shortened (data not shown) suggesting FXa^I16L saturation with the in situ generated FVa. To test this further, additional FVa was added to the system. As shown in Figure 2b, the addition of 2 nM FXa^I16L and 5 nM FVa shortened the CT beyond that observed with 2 nM FXa^I16L alone and controls. The addition of 10 nM FVa with 2 nM FXa^I16L further shortened CT beyond that observed with 2 nM FXa^I16L and 5 nM FVa and controls. The addition of FVa (5 nM or 10 nM) without FXa^I16L had no appreciable effect on CT relative to HA patient whole blood analysis without added FVa. These results were recapitulated in three separate patient samples; however, small sample size precluded statistical analysis and further validation of findings. Nonetheless, these observations suggest that, through the requirement of FVa to bind and thereby rescue the FXa^I16L protease, available FVa may limit FXa^I16L procoagulant potential and thereby prevent undesired thrombosis. Given the inherent limitations of ex vivo modeling and artificial circumstances in which FVa quantity may be limited in this assay system, caution must be used in interpreting results. Nonetheless our findings support what is known about the underlying mechanism of FXa^I16L protease conversion. Specifically, zymogen-like FXa molecules may have limited off target prothrombotic potential since available FVa dictates their activity. If true, zymogen-like FXa molecules may have less thrombotic risk than current bypass therapies.

Conclusion:

At much lower concentrations than rFVIIa, FXa^I16L normalized INTEM CT in HA subjects both with and without inhibitors. These findings are congruent with prior pre-clinical hemophilic animal studies and further support the use of FXa^I16L as an alternative bypassing therapy for hemophilia patients with inhibitors. Further, the requirement of FVa to rescue FXa^I16L protease function suggests procoagulant injury site specificity and limited off target thrombotic potential. A Phase I study is currently evaluating the safety of FXa^I16L in healthy human volunteers (ClinicalTrials.gov; NCT01897142).