Preclinical Safety and Pharmacokinetics of Recombinant Human Factor XIII

Abstract

Factor XIII (FXIII) is a thrombin-activated protransglutaminase responsible for fibrin clot stabilization and longevity. Deficiency in FXIII is associated with diffuse bleeding and wound-healing disorders in humans. This report summarizes results from several studies conducted in adult cynomolgus monkeys (M. fascicularis) to evaluate the safety and pharmacokinetics of recombinant human factor XIII A₂ dimer (rFXIII). Intravenous slow bolus injection of rFXIII resulted in the expected formation of the heterotetramer rA₂cnB₂, prolonged circulating half-life (5–7 days), and increased plasma transglutaminase activity. Recombinant FXIII was well tolerated as a single dose up to 20 mg/kg rFXIII (2840 U/kg), as repeated daily doses up to 6 mg/kg (852 U/kg) for 14 days, and as 3 repeated doses of 8 mg/kg (1136 U/kg) separated by 14 days. Overt toxicity occurred after a single intravenous injection of ≥ 22.5 mg/kg rFXIII (3150 U/kg), or with 2 doses of =12.5 mg/kg (1775 U/kg) administered within 72 hours. The rFXIII-mediated toxicity was expressed as an acute systemic occlusive coagulopathy. Evaluation of plasma samples from dosed animals demonstrated formation of cross-linked fibrin/fibrinogen oligomers and higher-order protein aggregates, which are hypothesized to be responsible for the observed vessel occlusion and associated embolic sequelae. These results demonstrate that rFXIII-mediated toxicity results from exaggerated pharmacological activity of the molecule at supraphysiological concentrations. The absence of observed toxicological effect with repeated intravenous doses up to 8 mg/kg (1136 U/kg) was used to support an initial clinical dose range of 0.014 to 0.35 mg/kg (2–50 U/kg).

Introduction

Congenital Factor XIII (FXIII) deficiency is a rare inherited autosomal recessive blood disorder estimated to affect 1 in 5 million persons worldwide (Gootenberg, 1998; Di Paola et al., 2001). Congenital FXIII deficiency usually presents as delayed bleeding from the umbilical stump or circumcision site in homozygous newborn infants. The majority of FXIII deficient patients not receiving prophylactic therapy with FXIII-containing plasma products develop spontaneous hemorrhages, subcutaneous hematomas, superficial ecchymoses, spontaneous bleeding, epistaxis bleeding, and delayed bleeding from sites of trauma (Board et al., 1993; Shaikh and Khurshid, 1993; Egbring et al., 1996). Because of the high mortality associated with intracranial hemorrhage, prophylaxis with FXIII-containing blood components or human plasma concentrates is recommended in these patients (Board et al., 1993; Miloszewski and Losowsky, 1991; Gootenberg, 1998). In addition to congenital deficiency, acquired FXIII deficiency can occur subsequent to blood loss, chronic coagulation system activation, or FXIII inhibitor development. Factor XIII-containing plasma products and human plasma-derived concentrates have been shown to restore clot stability and increase plasma FXIII activity as measured by in vitro assays. Multiple long-term observational studies have demonstrated that monthly prophylaxis (every 4–6 weeks) with FXIII-containing plasma, cryoprecipitate, or human plasma-derived concentrates have virtually eliminated bleeding episodes, including intracranial hemorrhage, among those who are congenitally deficient (Miloszewski and Losowsky, 1991; Board et al., 1993; Emrich et al., 1993). Recombinant FXIII (rFXIII) is being developed as a safe and effective alternative to FXIII-containing preparations derived from human plasma.

Factor XIII is present in plasma and in platelets, monocytes, and other cells of monocytic lineage. Plasma FXIII (pFXIII) circulates as a heterotetramer composed of 2 FXIII-A and 2 FXIII-B subunits (A₂B₂). Enzymatic activity resides with the FXIII-A subunit. The FXIII-B subunit circulates in excess in plasma and acts as a carrier, thus conferring a longer circulating half-life to the FXIII-A subunit. The cellular form of FXIII (cFXIII) lacks the FXIII-B subunits and is a homodimer of FXIII-A subunits (A₂). Recombinant FXIII (rFXIII) is produced in Saccharomyces cerevisiae as a nonglycosylated FXIII A₂ homodimer that is virtually identical to human cFXIII. Both human cFXIII and rFXIII readily form heterotetrameric A₂B₂ complexes in the presence of human FXIII-B subunit (Radek et al., 1993). Studies on the binding of rFXIII to cynomolgus FXIII-B subunit conducted to support selection of species for toxicology studies confirm appropriate formation of the heterotetramer (data not shown). The activity of rFXIII on human fibrin is similar to native FXIII (Bishop et al., 1990).

Factor XIII functions as a transglutaminase and forms Nɛ(γ-glutamyl)lysyl cross-links between a glutamine residue of 1 protein and a lysine of another. Factor XIII circulates in plasma in a nonactive form (zymogen or proenzyme). Figure 1 indicates the pathway by which the protransglutaminase is activated by thrombin at the site of an injury and stabilizes the maturing thrombus by catalyzing the formation of cross-links between the γ-chains of fibrin, the α-chains of fibrin, and other proteins such as α ₂-plasmin inhibitor α and fibronectin (Lorand, 1972; McDonagh and Fukue, 1996; Lewis et al., 1997; Brummel et al., 1999). The cross-linking of fibrin and fibrinolytic inhibitors retards subsequent fibrinolysis. Under normal conditions, clot formation occurs through the concurrent thrombin-mediated proteolysis of both fibrinogen and FXIII to form a cross-linked fibrin clot. Both pFXIII and cFXIII can also be activated nonproteolytically in plasma, although the functional role of the activation pathway is not well understood (Muszbek et al., 1999). Nonproteolytic activation of FXIII leads to a fully competent molecule capable of catalyzing all of the known cross-linking reactions common to thrombin-activated FXIII. Thus, nonproteolytically activated FXIII could give rise to inappropriate cross-linking or cross-linking under conditions where the normal coagulation pathways are not active. The activity of rFXIII on human fibrin has been demonstrated to be very similar to native FXIII (Bishop et al., 1990).

Summarized here are results from several studies examining the preclinical safety and pharmacokinetics of single and repeated intravenous rFXIII injection to adult cynomolgus monkeys. Cynomolgus monkeys were used as a study model because their coagulation parameters are similar to those of humans (Seaman and Malinow, 1968) and they demonstrate pharmacologic response to rFXIII including the expected in vitro plasma fibrin cross-linking activity and formation of the Factor XIII heterotetramer complexes between endogenous FXIII-B subunit and rFXIII (i.e., rA₂cnB₂). Administration of rFXIII at supraphysiological levels results in saturation of available uncomplexed FXIII-B subunit and leaves excess, uncomplexed rFXIII in circulation; uncomplexed FXIII-A₂ dimer is not observed under normal physiological conditions (nor is it expected to occur under conditions of therapeutic rFXIII administration). Evaluation of plasma samples after rFXIII treatment revealed evidence of rFXIII activation and the associated formation of high molecular weight protein complexes comprised of expected substrates for FXIII transglutaminase activity. These high molecular weight complexes appeared to have a relatively long persistence and their increased burden presumably underlies the observed occlusive coagulopathy that is characteristic of rFXIII toxicity.

Materials and Methods

Drug Product Composition

The drug product used in the preclinical safety studies was manufactured by ZymoGenetics, Inc. (Seattle, WA) and supplied as 5 mL vials containing an isotonic 5 mg/mL (710 U/mL) rFXIII solution formulated in 3% sucrose, 20 mM histidine, 200 mM glycine, and 0.01% polysorbate 20 at pH 8.0 suitable for intravenous injection. The concentration of recombinant FXIII (mg/kg) can be expressed as units of enzymatic activity (U/kg) by using the following conversion: 1 mg rFXIII = 142 U, where 1 U equals the amount of transglutaminase activity in 1 mL of normal human plasma. The composition and purity of the test article was characterized by high-performance liquid chromatography (content, identity, % aggregate), anion exchange high-performance liquid chromatography (rFXIII-related charge heterogeneity), potency (total activity, % activated rFXIII), and for endotoxin content. The control article was of the same formulation without rFXIII.

The rFXIII protein was extensively characterized by amino acid analysis, N-terminal sequencing, peptide mapping by high-performance liquid chromatography (HPLC) and liquid chromatography-mass spectrometry-mass spectrometry, whole mass measurement by electrospray ionization-MS (ESI-MS), size exclusion chromatography-multiangle light scattering (SEC-MALS), and SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Amino acid analysis and peptide mapping demonstrated that the protein had an amino acid composition predicted from the human wild-type cDNA sequence. An average protein mass of 166 kD measured by SEC-MALS is consistent with the molecular weight of a FXIII [A₂] dimeric protein. N-terminal sequencing showed no evidence of proteolytically activated FXIII, internal cleavages, or contaminating sequences. There was no evidence of N- or C-terminal heterogeneity, or unexpected posttranslational modifications.

In Vivo Study Design

Experimentally naïve male and female cynomolgus monkeys (Macaca fascicularis) aged 2.5 to >5 years of Chinese origin were used in these studies. Animal care and experimental procedures were conducted at the Charles River Laboratories, Sierra Division (Sparks, NV) or Shin Nippon Biomedical Laboratories USA (Everett, WA). All study procedures were performed according to written protocols approved by the institutional animal welfare and use committee of each facility responsible for study conduct. Strict compliance to accepted animal welfare standards, including AALAS Policy, U.S. Public Health Service Policy on the Humane Care and Use of Animals, the ILAR Guide for the Care and Use of Laboratory Animals, and the Animal Welfare Act (9 CFR, Parts 1, 2, and 3) was observed.

The safety of rFXIII administered as a single dose or repeated dose intravenous slow bolus injection (=2 mL/min) was evaluated in cynomolgus monkeys as indicated in Table 1.

Animal Dosing

Within studies, animals were randomized to dosing with the goal of balancing mean body weights across dose groups. Control article dosing volumes were matched against the largest test article volumes used in any given study. Animals were dosed by slow bolus intravenous injection into a saphenous or cephalic vein. Animals were routinely observed at least twice daily for clinical observations beginning at least 7 days prior to test article administration and continuing through the end of study. Continuous clinical monitoring was conducted through 6 hr after test article administration.

Cardiovascular Safety Pharmacology

Cardiovascular safety data consisting of heart rate, blood pressure, body temperature (collected with a Hewlett Packard 78534B physiological monitor), and electrocardiography (collected with a Hewlett Packard Pagewriter XLE or 200) were obtained for all available study animals in each study. Analyses were conducted 2–3 times prior to dosing, after the first dose (for single dose studies: between 2–4 hours, at 24 hours, and at 48 hours postdose; for repeated dose studies: at 24 hours post-dose) and prior to sacrifice (either at the end of the dosing period or after a 4-week dose-free period).

Pathology

All animals deemed moribund by the veterinarian were euthanized and subjected to necropsy examination. In the single-dose studies, a subset of animals observed to have no or minor clinical effects were sacrificed 7 days after dose administration; remaining study animals without clinical effect were returned to the facility colony. Animals assigned to the repeat dosing studies were euthanized and necropsied either 48 hours after the last dose or after a 28-day dose-free period. The animals were euthanized by exsanguination while under deep anesthesia induced with ketamine and Nembutal. A complete gross necropsy was performed by, or under the supervision of, qualified veterinary pathologists and included collection of all macroscopic observations, organ weights, and comprehensive collection of tissue samples for microscopic examination. Tissues were preserved in neutral-buffered 10% formalin (except for the eyes, which were preserved in Davidson’s fixative for optimum fixation), routinely processed and embedded in paraffin, sectioned, stained with hematoxylin and eosin, and examined by light microscopy by a board certified veterinary pathologist.

Clinical Pathology

Comprehensive hematology analysis was performed on an Abbott Cell-Dyn 3500 automated hematology analyzer (Abbott Labs, Pomezia, Italy) using approximately 0.5 mL of blood collected from a femoral or saphenous vein via butterfly catheter into a syringe containing EDTA anticoagulant. Endpoints measured included red blood cell (RBC) counts, mean cell hemoglobin (MCH), white blood cell (WBC) and differential cell counts, mean corpuscular volume (MCV), hemoglobin concentration, mean corpuscular hemoglobin concentration (MCHC), hematocrit, platelet counts, reticulocyte counts, blood cell morphology, red cell distribution width (RDW), and blood cell morphology. For single-dose studies, hematology analyses were conducted on all available animals 1–2 times prior to dosing and 2–4 hours, 24 hours, 72 hours, and 168 hours postdose, with additional collections at 30 minutes, 8 hours, and 312 hours for some animals. For repeated dose studies, numerous hematology analyses were obtained on all animals so as to establish baseline levels prior to dosing, and to evaluate acute responses during the dosing period (generally 4–72 hours postdosing) and chronic effects or recovery from induced changes during an extended (up to 4-week) dose-free period.

Comprehensive serum chemistry analysis was performed on a Beckman Synchrotron CX7 automated chemistry analyzer (Beckman Instruments, Palo Alto, CA) using approximately 0.5 mL serum isolated from whole blood collected from a femoral or saphenous vein via butterfly catheter into a syringe. Endpoints measured included sodium, calcium, potassium, phosphorus, chloride, urea nitrogen (BUN), carbon dioxide, creatinine, total bilirubin, total protein, alkaline phosphatase (AP), albumin, lactate dehydrogenase (LDH), globulin, aspartate aminotransferase (AST), albumin/globulin ratio, alanine aminotransferase (ALT), glucose, gamma-glutamyltransferase (GGT), cholesterol, C-reactive protein (CRP), and triglycerides. For single-dose studies, serum chemistry analyses were conducted on all available animals once prior to dosing and 24 hours, 72 hours, and 168 hours postdose. For repeated dose studies, periodic serum chemistry analyses were obtained on all animals so as to establish baseline levels prior to dosing, and to evaluate responses during the dosing period (at least weekly) and chronic effects or recovery from induced changes during an extended (up to 4-week) dose-free period.

Coagulation analyses of activated partial thromboplastin time (APTT), prothrombin time (PT), and fibrinogen were conducted on an AMAX CS-190 Coagulation Analyzer (Sigma Diagnostics, St. Louis, MO). Plasma was isolated from whole blood generally collected from a femoral or saphenous vein via butterfly catheter into a syringe containing sodium citrate as the anticoagulant. For single-dose studies, coagulation analyses were conducted on all available animals 1–2 times prior to dosing, and 2–4 hours, 24 hours, 72 hours, and 168 hours postdose, with additional collections at 30 minutes, 8 hours, and 312 hours for some animals. For repeated-dose studies, periodic coagulation analyses were obtained on all animals to establish baseline levels prior to dosing, and to evaluate responses during the dosing period (at least weekly) and chronic effects or recovery from induced changes during an extended (up to 4-week) dose-free period.

When possible, blood was collected for clinical pathology analyses from moribund animals. Additional clinical pathology analyses were also collected at the request of the clinical veterinarian.

Pharmacokinetics

The pharmacokinetics of rFXIII was assessed for all study animals, including controls, using plasma isolated from whole blood collected from a femoral or saphenous vein via butterfly catheter into syringe containing EDTA as the anticoagulant. For single-dose studies, blood was collected prior to dosing (0 hours) and 15 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 24 hours, 72 hours, 120 hours, 168 hours, 336 hours and 672 hours post-dose from all available animals. For repeated dose studies, blood samples for pharmacokinetic analysis were collected prior to dosing and 15–30 minutes, 6 hours, and 24 hours after selected doses and at periodic intervals during an extended dose-free period; these analyses were conducted to capture peak and trough plasma concentrations and to identify possible effects on pharmacokinetics of antibody formation over the course of the study.

Validated enzyme linked immunosorbent assays (ELISA) were developed to monitor each of the possible FXIII molecular species in plasma including FXIII-B subunit, endogenous A₂B₂, newly formed rA₂cnB₂, and uncomplexed rA₂. These assays included the Total A₂ assay for the combined detection of all molecular forms containing the FXIII-A sub-unit (cnA₂cnB₂, rA₂cnB₂, and rA₂), the Free B subunit assay for detection of uncomplexed cynomolgus FXIII-B subunit (cnB) and the A₂B₂ Tetramer assay for specific detection of heterotetrameric FXIII species (i.e., endogenous FXIII (cnA₂cnB₂) and the formed heterotetramer (rA₂cnB₂)). The capture-detection configuration, performance characteristics, and molecular species detected by each ELISA assay are presented in Table 2. Plates were read on a Spectramax 190 Microplate Spectrophotometer (Molecular Devices, Sunnyvale, CA) or equivalent.

Protein standards established for these assays by ZGI included rFXIII, and highly purified and well-characterized human FXIII A₂B₂ protein and human FXIII-B subunit protein. The human FXIII-B protein standard was derived from partially purified pFXIII (Enzyme Research Laboratories, South Bend, IN) treated with thrombin to cleave the FXIII-A subunit, and then Ca⁺⁺ to separate the subunits. The mixture was centrifuged to pellet the cleaved FXIII-A, and the supernatant was purified by anion exchange chromatography. Purified human pFXIII A₂B₂ protein was obtained from partially purified pFXIII (Enzyme Research Laboratories) by size-exclusion high performance liquid chromatography (SE-HPLC) to isolate the heterotetramer from other protein species. Protein standards were extensively analyzed by N-terminal sequencing, reduced and nonreduced SDS-PAGE, SE-HPLC, and amino acid analysis to confirm identity, purity and concentration. Enzymatic activity and ability to form the expected heterotetrameric complex (in the case of the purified A₂- and B-subunits) were also demonstrated.

While in principle, the amount of FXIII measured by the Total A₂ ELISA should quantify all FXIII regardless of whether it is in the dimeric A₂ or heterotetrameric A₂B₂ state, the Total A₂ assay underdetects the heterotetrameric form of FXIII. Thus mass balance between the assays is not possible. Despite this caveat, the analytical methods do allow for an estimation of the pharmacokinetic parameters for the heterotetrameric species alone and allow for the monitoring of uncomplexed B species of FXIII. This information is important in establishing the relationship between dosed rFXIII, formed heterotetramer (rFXIII complexed with cnB subunit), and the possible induction of cnB.

The data obtained from the Total A₂ and FXIII A₂B₂ tetramer ELISA data were subjected to noncompartmental toxicokinetic analysis using WinNonlin (Pharsight Inc., Cary, NC). In all cases, the postdose plasma concentrations were corrected for individual baseline FXIII plasma levels prior to toxicokinetic analysis.

Anti-FXIII Antibody Analyses

Plasma samples were collected from all study animals included in repeated-dose studies and were evaluated for the presence of anti-FXIII antibodies. These samples were collected during the predosing period to establish baseline titers, at periodic intervals during the dosing period selected to represent trough circulating rFXIII concentrations, and at periodic intervals during the 4-week postdosing recovery period. This sampling schedule was designed to minimize possible interference of circulating rFXIII with antibody detection during the dosing and 4-week dose-free period, and to detect antibodies formed at least 2 weeks after initiation of dosing.

The anti-FXIII antibody assay used rFXIII A₂ on the solid phase to capture any anti-FXIII Ig present in the test sample. Affinity purified rabbit-anti-rFXIII antibodies spiked into cynomolgus serum was used as the quality control sample. Plates were washed to remove unbound protein, and protein G-HRP (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA), which recognizes cynomolgus monkey Ig, was used to detect immunoglobulin captured from the test or rabbit-anti-rFXIII antibodies in quality control samples. The plates were washed and TMB (BioFX Laboratories, Owens Mills, MD) was added as a detection reagent. Plates were read on a Spectramax 190 Microplate Spectrophotometer (Molecular Devices, Sunnyvale, CA). The polyclonal rabbit anti-rFXIII antibody was produced by ZGI by immunizing rabbits with rFXIII. The assay, validated for use with cynomolgus monkey sera or plasma, has an interassay precision of <10%, an intra-assay precision of <5%, is robust to multiple analysts. Sample stability was demonstrated through multiple freeze-thaw cycles and through storage at 4°C for up to four days. Assay sensitivity was established by spiking 5 μg/mL of affinity-purified rabbit anti-rFXIII antibody into naïve cynomolgus sera or plasma, which quantified at a titer of 2.3 (or a dilution between 1:100 and 1:500). By comparison, analysis of plasma and sera from rFXIII-treated animals yielded detectable titers between 2.1 and 2.6 in plasma, suggesting a sufficient assay sensitivity.

Evaluation of Fibrinogen Cross-Linking

The protein matrix of plasma is too complex for fibrinogen and cross-linked fibrin(ogen)-containing complexes to be easily identified by SDS-PAGE. In order to isolate fibrinogen and cross-linked fibrin(ogen)-containing complexes from plasma, the samples were treated with thrombin (80 U/mL), iodoacetamide (10 mM) and EDTA (10 mM) and the fibrinogen was intentionally converted to fibrin. The resulting fibrin clots were pelleted through centrifugation, washed with EDTA/iodoacetamide (which inhibit additional cross-linking from occurring during sample handling), dissolved with 8 M urea containing 40 mM DTT, and analyzed by reduced SDS-PAGE [1.0 mm 4–12% NuPAGE Bis-Tris gel, MOPS running buffer (Invitrogen, Carlsbad, CA)]. In addition to study samples, the following control samples were included: rFXIII standard (a mixture of zymogen and thrombin-treated rFXIII), cross-linked cynomolgus fibrin (isolated from re-calcified sodium citrate-anticoagulated plasma treated with thrombin), noncross-linked cynomolgus fibrin (isolated from EDTA treated plasma treated with thrombin), and cross-linked cynomolgus fibrinogen (isolated from heparinized cynomolgus plasma treated with thrombin-activated rFXIII and the thrombin inhibitor PPAK (D-Phe-L-Pro-L-Arg-chloromethyl ketone, Enzyme Systems Products, Livermore, CA). SDS-PAGE gels were stained with Coomassie Brilliant Blue R staining solution (Sigma, St. Louis, MO) and destained in 40% methanol/10% acetic acid.

Identification of Cross-Linked Proteins

Western blot analysis was performed on a subset of the samples to identify components of high molecular weight complexes. Immediately after gel electrophoresis, the protein bands of unstained gels were transferred to 0.2 μm nitrocellulose (Invitrogen). Blots were blocked and exposed to various detection antibodies based on known FXIII substrates including mouse monoclonal anti-human fibrinogen α-chain, mouse monoclonal anti-human fibrinogen β-chain, mouse monoclonal anti-human fibrinogen γ-chain (Accurate Chemical, Westbury, NY); rabbit anti-human fibronectin, rabbit anti-human vitronectin, rabbit anti-human α ₂-antiplasmin (Calbiochem, La Jolla, CA); and goat anti-human α ₂-macroglobulin (Affinity Biologics, Inc., Hamilton, Ontario). The following secondary antibodies were used: sheep anti-mouse Ig-HRP, donkey anti-rabbit Ig-HRP (Amersham Pharmacia Biotech, UK Ltd.), and rabbit anti-goat Ig-HRP (Rockland, Gilbertsville, PA). For samples analyzed by N-terminal sequencing, the protein bands of unstained SDS-PAGE gels were transferred to 0.2 μm PVDF membranes (Invitrogen). Bands were visualized by staining with 20% Coomassie Brilliant Blue staining solution (Sigma) in 40% methanol and destaining with 50% methanol. When possible, identified bands were compared with estimated molecular weight standards or control proteins to establish their identify, excised and analyzed by Edman sequencing on an ABI 494 Protein Sequencer (Applied Biosystems, Foster City, CA), or subjected to Western blot detection.

Results

Pharmacokinetics Following a Single Dose of rFXIII

Using the available bioanalytical assays, plasma concentration versus time profiles were generated for the free FXIII-B subunit, the Total A₂ (including both the free rA₂ and the complexed rA₂) and the FXIII A₂B₂ tetramer (Figure 2). As a rule, pharmacokinetic analyses of control animals demonstrated no change in FXIII molecular species.

Free FXIII-B Subunit

Following slow bolus intravenous injection of saturating quantities of rFXIII A₂ (e.g., 5 mg/kg), the free B subunit bound to the rFXIII A₂ and no free B subunit was detected in the plasma until the 24 or 72 hours postdose sampling time. The time to recovery of free B subunit was directly related to administered dose, taking longer in animals that received higher rFXIII doses. For example, in animals dosed with 5 mg/kg rFXIII, average free B subunit concentrations of approximately 1 μg/mL at 72 hours postdose (and later) were comparable to those observed at baseline (0 hours), but no free B subunit was detected at 30 minutes, 2 hours, or 8 hours postdose, and reduced concentrations were observed at 24 hours postdose (Figure 2). The return of free B subunit to baseline concentrations by 72 hours in these animals and a C_max, _A2B2 observed between 24 and 72 hours determined by the FXIII A₂B₂ tetramer assay suggests that by 72 hours all remaining rFXIII was in the form of the rA₂ dimer complexed with cynomolgus B subunit, and suggests induction of B subunit production following rFXIII dose administration. Using population-based PK modeling, the turnover for free B subunit in cynomolgus monkeys was estimated to be 10.5 ug kg⁻¹ hr⁻¹ (Dodds et al., 2004a).

Kinetics of rFXIII A₂ Complexed with Cynomolgus FXIII-B Subunit

Following intravenous injection of rFXIII to cynomolgus monkeys, the profiles resulting from analysis with the Total A₂ ELISA showed a bi-exponential decay for all tested doses (0.5–12.8 mg/kg rFXIII). During the first 24 hours postdose, total plasma A₂ levels rapidly decreased, presumably reflecting the elimination of excess, un-complexed rA₂ from circulation. The second phase, from 24 hours until the last measured time point, had a much more shallow slope and is assumed to reflect the half-life of the formed heterotetramer. Noncompartmental analyses showed a linear relationship between the initial concentration (C₀) and the administered dose. The estimated terminal half-life was not significantly different across all doses administered (135–195 hours). However, the areas under the plasma concentration vs. time curves (AUC) for these Total A₂ profiles did not increase proportionally with dose, and the clearance (CL) and volume of distribution at steady state (V_ss) increased with increasing dose (Table 3).

The profile resulting from analysis with the FXIII A₂B₂ ELISA demonstrated a time to maximum concentration (T_max) of approximately 24–72 hours followed by monoexponential decay for all tested doses. Terminal half-lives were estimated to be 125–225 hours. This is consistent with the range of terminal half-lives estimated for the Total A₂ ELISA results and thus confirms that the terminal half-life represents the elimination of the rA₂cnB₂ in both the total A₂ and the FXIII A₂B₂ assays.

Pharmacokinetics of rFXIII Following Repeated Doses

The pharmacokinetics of rFXIII, the formed tetramer rA₂cnB₂ and the free B subunit were characterized following repeat doses of rFXIII in cynomolgus monkeys. Following repeated intravenous doses of 0.3, 3.0, and 6.0 mg/kg rFXIII administered daily for 14 days, FXIII A₂B₂ troughs measured predose on days 1, 2, 8, 11, and 14, showed an increase from baseline values (endogenous cynomolgus A₂B₂) of approximately 4-fold by day 8 for the animals dosed daily with 0.3 mg/kg and approximately 8–10-fold by day 8 for the animals dosed daily with 3.0 and 6.0 mg/kg, respectively (data not shown). These data demonstrate that the formation of the heterotetramer is partially limited by the availability/turnover of free B subunit, however, a 4- to 10-fold increase over base-line A₂B₂ levels suggests that FXIII-B subunit is induced following the administration of excess rFXIII.

Formation of Anti-rFXIII Antibodies

Positive antibody titers were observed in 0 of 4, 1 of 4, and 1 of 6 of the animals receiving 0.3, 3.0, and 6.0 mg/kg rFXIII daily for 14 days, respectively. The observed mean trough and peak Total A₂ and FXIII A₂B₂ plasma concentrations showed no apparent decrease for any of the antibody positive animals (data not shown).

Single-Dose Toxicity Studies

A study of rFXIII was conducted in cynomolgus monkeys to establish the dose-response profile and to identify premonitory indicators of acute toxicity. The dose-response relationship for rFXIII was quite steep, and a threshold of toxicity was defined wherein no clinical effects were observed after a single intravenous slow bolus dose of up to 21.2 mg/kg rFXIII (n = 7), but lethality was observed in 6 of 7 animals dosed between 22.5–30 mg/kg rFXIII. Animals in poor health were monitored and received lactated Ringer’s solution and other supportive care as deemed appropriate by the veterinarian, however, moribund animals did not respond to supportive care and were sacrificed for humane reasons. At doses associated with morbidity, presentation of clinical signs was variable, with either acute (i.e., starting as early as 4–6 hours after dosing) or more gradual deterioration in condition (i.e., developing over 3–5 days). Among moribund animals, clinical observations included evidence of poor clotting, epistaxis, ecchymosis, and neurobehavioral effects including poor food and water consumption, malaise, and loss of consciousness.

Moribund animals had a consistent decrease in platelet count, with critically low platelet counts (<50,000/μL) observed in some animals. Serum chemistry changes in animals with a declining health condition were consistent with a histologically confirmed coagulopathy. Serum chemistry alterations indicative of stress and organ/tissue damage included increased blood urea nitrogen (>30 mg/dL), creatinine (>1.1 mg/dL), lactate dehydrogenase (LDH, > 882 U/L), aspartate aminotransferase (AST, >94 U/L), alanine aminotransferase (ALT, >127 U/L) levels, and C-reactive protein (CRP >3 mg/dL). Prolongations in apparent prothrombin time (APTT, >42.3 seconds) or prothrombin time (PT, >14.3 seconds) were observed only as late-stage events occurring immediately prior to humane sacrifice in some animals.

Cardiovascular safety pharmacology data consisting of blood pressure, body temperature and heart rate were collected during the toxicology studies of rFXIII. Cardiovascular safety pharmacology was evaluated because of the pharmacological activity of rFXIII as the terminal enzyme in the clotting cascade involved in stabilizing fibrin clots. No general trends in measured blood pressure, body temperature, or heart rate were observed related to rFXIII dose level or time after dose. However, among acutely moribund animals it was often difficult to collect blood samples immediately prior to animal euthanasia, suggesting low blood pressure during the more advanced stages of toxicity.

Gross pathology observations generally consisted of red discoloration or red foci indicative of focal hemorrhage in a variety of tissues including adrenal glands, lung, kidneys, heart, liver, and the gastrointestinal tract. Most of the histologic findings (e.g., intravascular congestion, embolisation, and subsequent coagulative necrosis) correlated with the expected sequelae of overexposure to FXIII. Microscopic evidence of a generalized embolisation and/or ischemia was found with varying degrees of severity in many organs and tissues including adrenal glands, kidneys, eye, lung, heart, gastrointestinal tract, pancreas, spleen, liver, brain, neurohypophysis (pituitary), and bone marrow. The primary target tissues most severely and consistently affected by this occlusive coagulopathy were adrenal gland, kidney, eye, lung, and heart.

Acellular granular emboli were observed within small blood vessels throughout the adrenal glands and kidneys, most prominently in the glomerular capillaries of the kidney (Figure 3a). These emboli were frequently associated with multifocal to coalescing hemorrhage (Figure 3b) and coagulative necrosis. In the eye, vessel occlusion within capillaries and small vessels of the retina and choroid were mixtures of classic fibrin thrombi and emboli formed of acellular granular material (Figure 3c) and in the heart, the acellular granular emboli found within the capillaries of the myocardium were frequently associated with adjacent areas of hemorrhage and coagulative necrosis (Figure 3d). Distribution in all tissues varied from multifocal, coalescing to diffuse.

Repeated Dose Studies

Four experiments evaluated the effect of repeated slow bolus intravenous rFXIII injection (Table 1). No clinical or anatomic pathologies were observed in animals that received twice-weekly slow bolus injections of 5.0 mg/kg rFXIII for 4 weeks or daily intravenous slow bolus injections of up to 6.0 mg/kg rFXIII for 14 days. Similarly, intravenous administration of three doses of control (0.0 mg/kg) or rFXIII (5.0, 8.0, or 12.5 mg/kg) each separated by 14 days was well tolerated in adult male and female cynomolgus monkeys, and no effects were seen in clinical observations, clinical pathology evaluations, or gross evaluations of tissues and organs.

Minor histological changes observed in the glomeruli in 1 of 6 animals dosed with 12.5 mg/kg and sacrificed two days after the last dose (but not 4 weeks later) suggested possible formation of transient emboli. This glomerulopathy was characterized by loss of defined glomerular capillaries due to increased eosinophilic material and cellularity in the glomerular tuft, and in some glomeruli, periglomerular fibrosis and cellular infiltrates. In contrast to these results, 2 intravenous injections of 12.5–17.5 mg/kg rFXIII separated by 72 hours resulted in toxicity within 1 day after the second dose despite the absence of any indication of effect after the first dose. The clinical observations in the moribund animals after repeated rFXIII dose administration was similar to that observed after a single dose, and consisted of a generalized occlusive coagulopathy diagnosed through clinical and anatomic pathology examinations.

Development of Cross-Linked Protein Complexes

Plasma samples from a subset of rFXIII-treated monkeys treated over a range of rFXIII doses were examined for evidence of cross-linked protein species. Evaluation of gels generated by reduced SDS-PAGE revealed the presence of intermediate and high molecular weight bands (Figure 4). Complexes in amounts over background levels were observed in a dose-dependent manner, and persisted in circulation despite clearance of rFXIII (as demonstrated by both pharmacokinetics and appearance on the gel). Using silver-stained gels, complexes were observed as early as 5 minutes after dosing with 5 mg/kg rFXIII (Figure 4), and were seen faintly in less sensitive Coomassie-stained gels as early as 30 minutes after dosing with 10 mg/kg rFXIII (data not shown). A band migrating at approximately 95 kD was identified by N-terminal sequencing as a dimer of the gamma chains of fibrinogen, which demonstrate FXIII cross-linking activity. Multiple high molecular weight bands larger than approximately 130 kD were also observed. Using Western blotting, these bands were identified as various conjugates of the alpha-and gamma-chains of fibrinogen, alone or in combination with α ₂-plasmin inhibitor α ₂-macroglobulin, and fibronectin, which are known substrates for FXIII transglutaminase activity (data not shown).

Discussion

The accumulated evidence of both kinetic and dynamic activity of rFXIII in cynomolgus monkeys consistent with the enzymatic properties of a transglutaminase justify the use of cynomolgus monkeys as a suitable species for nonclinical studies of rFXIII. The in vitro evidence that rFXIII correctly cross-links cynomolgus monkey fibrin to form gamma-dimers following thrombin-stimulated clotting, and the additional evidence of observed circulating cross-linked fibrin (and other higher-order aggregates) after intravenous injection of high doses of rFXIII to cynomolgus monkeys demonstrates pharmacodynamic activity of rFXIII in cynomolgus monkeys. The in vivo pharmacology of FXIII has been investigated in a limited number of animal models. These include animal models of FXIII deficiency (Lauer and Dickneite, 2001; Lee et al., 2001), in vivo biochemistry models (Shainoff and Dardik, 1990; Uchino et al., 1991), and models of exogenous FXIII administration in specific disease states (e.g., Isogai et al., 1990; Hirahara et al., 1993; D’Argenio et al., 2000). The available studies demonstrate a conserved biologic activity of FXIII across a number of mammalian species including mice, rats, rabbits, guinea pigs, nonhuman primates, and humans.

The in vitro and in vivo evidence that rFXIII A₂ binds to cynomolgus FXIII-B subunit resulting in a molecule of the correct molecular weight and circulating half-life shows comparability between the rFXIII kinetics in cynomolgus monkeys and endogenous FXIII in humans. Recombinant FXIII when complexed with endogenous B subunit has a terminal half-life of 5–9 days in cynomolgus monkeys, which is consistent with the terminal half-life estimated in congenitally deficient patients receiving the plasma purified FXIII product Fibrogammin (~6–12 days). Uncomplexed rFXIII (circulating in excess of free B subunit) has a short circulating half-life of 4–7 hours (data not shown). This observation is consistent with the theory that the FXIII-B subunit serves to stabilize the heterotetramer. However, in all cases where rFXIII is dosed in excess of available free B subunit, our estimates for Vss, _totalA2 are larger than the typical blood volume for a 3–4 kg cynomolgus monkey (approx. 270 mL). There appears to be a well-controlled feedback with regards to FXIII-A₂ and Free B subunit levels.

It is possible that the rapid clearance and large volume of distribution of total A₂ may be associated with the activation of excess rA₂ and its binding to other cellular components and/or compartmentalization out of circulating plasma. Importantly, completed clinical trials in both healthy volunteers and congenitally deficient patients show that when therapeutically relevant doses of rA₂ were administered (not in excess of available of free B subunit), the volume of distribution was similar to total blood volume (Lovejoy et al., 2004; Reynolds et al., 2004). Population-based models have been developed to describe the pharmacokinetics of rFXIII in both healthy volunteers and congenitally deficient patients (Dodds et al., 2004b).

The steep dose-response associated with rFXIII toxicity is consistent with its activity as a central mediator between coagulation and fibrinolysis, and can be understood in the context of the homeostatic controls that regulate the balance between bleeding and thrombosis. The likely sequence of events leading to death of animals undergoing acute rFXIII toxicity was an initial formation of acellular granular emboli possibly consisting of aggregated polymers of fibrin, fibrinogen, α ₂-plasmin inhibitor α ₂-macroglobulin, and/or fibronectin. Accumulation of these complexes in small vessels could obstruct blood flow and result in local tissue ischemia and coagulative necrosis in the surrounding tissues. The necrosis could subsequently cause the release of systemic inflammatory mediators including tissue factor, thereby eliciting shock, inflammation and activation of the clotting cascade, which would result in the more typical fibrin thrombi formation in larger vessels.

Hematological changes consistent with disseminated intravascular coagulation (DIC) were detected in samples taken from moribund animals shortly before death. These changes included decreased platelet counts, prolongations in PT and APTT, and increases in D-dimer and fibrin degradation products. Therefore, DIC was most likely a terminal event. The histological picture and proposed sequence of events are consistent with the mechanism of action of rFXIII and the collected clinical pathology data. Thus, the widespread embolisation resulting in death of animals at high doses represents exaggerated pharmacology of the test article.

The observed occlusive coagulopathy following rFXIII toxicity is distinct from that of DIC. Classical DIC ultimately involves loss of control over the production of systemic thrombin and systemic plasmin via imbalances in the clotting and fibrinolytic systems (Bick et al., 1999). In contrast, the rFXIII-mediated toxicity is attributed to the formation of cross-linked protein complexes (including fibrinogen polymers) in plasma by activated rFXIII that ultimately occludes small vessels and results in a secondary systemic coagulation cascade activation. Evidence in support of this hypothesis includes the lack of an observed procoagulant activity of rFXIII added to plasma in vitro and evidence of high molecular weight complexes in the absence of histopathologically observed emboli/vessel occlusion. Thus, although both rFXIII-and thrombin-mediated plasma coagulation can ultimately cause comparable clinical and morphological pathology in vivo, the pathogenesis of coagulopathy is distinct. Moreover, diagnostic efforts to identify thrombin activation or consumption of enzymes involved in clotting system activation would not be expected to detect early rFXIII-mediated alterations in the disease process because the rFXIII-mediated coagulopathy does not appear to involve traditional mechanisms of clotting system activation.

One aspect of the observed rFXIII toxicity is the persistence of circulating cross-linked protein complexes. Siebenlist and Mosesson (1996) have observed low levels of cross-linked fibrinogen and of fibrinogen-α ₂-plasmin ihnibitor hetero-complexes (Siebenlist et al., 2000) in human plasma suggesting that some low-level burden of these molecular species is normal in humans. The complexes observed upon rFXIII administration in cynomolgus monkeys appear to be composed of expected substrates for FXIII transglutaminase activity. Similarly, Shainoff and Dardik (1990) have demonstrated that long-lived circulating fibrinogen-fibrinogen dimers and oligomers are formed after exposing plasma to thrombin-activated FXIII (FXIIIa^*) in vitro and in vivo. Additionally, Blomback et al. (1987) and more recently Siebenlist et al. (2001) describe the ability of activated FXIII to create a cross-linked fibrinogen gel in the absence of fibrin formation. The complexes observed with supraphysiological rFXIII dose administration thus appear to reflect exaggerated levels of complexes that are normally present in circulation. Whereas rFXIII in excess of endogenous FXIII-B has a short circulating half-life, cross-linked complexes appear to have longer half-lives. Additionally, whereas some burden of circulating complexes may be observed in the peripheral circulation, it is likely that formation of even larger complexes occurs in the local microvasculature and is not accessible to peripheral sampling. In small diameter vessels, such complexes may precipitate or aggregate, and obstruct blood flow.

Activated FXIII (FXIIIa), whether endogenous or recombinant, is unlikely to be present in the systemic circulation. Once activated, rFXIII binds strongly to fibrin (Procyk et al., 1993) and binds to a variety of cells in the vasculature, similar to endogenous FXIII. Two published studies report efforts to detect FXIIIa in plasma. Nelson and Lerner (1978) claimed to have detected circulating FXIIIa, but the technical details of their assay suggest that pFXIII was being activated during their assay, making their results difficult to interpret. Subsequently, Rodeghiero et al. (1981) reported only trace FXIIIa in both plasma from normal patients and plasma from patients with disseminated intravascular coagulation. In the absence of a direct detection assay for FXIIIa in plasma, detection of circulating cross-linked protein complexes involving fibrinogen, fibronectin, or α ₂-PI may offer the best option for detecting FXIII activation.

Given the potency of activated FXIII, the toxicity of rFXIII could be mediated by a small fraction of the administered drug that became activated. The primary physiological control over FXIII activity appears to be through the regulation of thrombin generation rather than through a specific biological inhibitor. While EDTA can be used to inhibit FXIII activity ex vivo, there is no evidence of an endogenous physiological inhibitor of FXIIIa, and FXIIIa remains active even when bound within a clot (Muszbek et al., 1999; Robinson et al., 2000). Likewise, FXIIIa remains active when bound to thrombin-stimulated platelets (Greenberg and Shuman, 1984). The normal decay in FXIII activity probably results from oxidation of the active site cysteine (Robinson et al., 2000).

The pathway that results in the activation of rFXIII at high doses is open to speculation. While thrombin is understood to be the typical regulator of FXIII activity during coagulation system activation, a nonproteolytic activation can occur in the presence of elevated level of intracellular calcium (Muszbek et al., 1999) and salt concentrations (Polgar et al., 1990). FXIII-B plays a protective role with respect to the nonproteolytic activation of the cFXIII, and cellular FXIII A₂ is more readily activated through the nonproteolytic pathway than pFXIII A₂B₂ (Polgar et al., 1990). Consequently, activity associated with non-proteolytic activation of rFXIII in vivo likely occurs when uncomplexed rFXIII is present in excess, such as the conditions provoked with supraphysiological rFXIII administration and depletion of free FXIII-B subunit.

The results presented here represent novel finding regarding the pharmacokinetics and pharmacodynamics of rFXIII in cynomolgus monkeys. We demonstrate that rFXIII-mediated toxicity results from exaggerated pharmacological activity of the molecule at supraphysiological concentrations. The absence of observed toxicological effect with repeated intravenous doses up to 8 mg/kg (1136 U/kg) was used to support clinical dosing of 0.014 to 0.35 mg/kg (2–50 U/kg) in healthy volunteers and congenital FXIII-deficient patients.