A flow diagram of both the GC5107 and GC5101 manufacturing processes are shown in Figure 1. The processes are identical up to and including the nanofiltration step. For both products, cryo-poor pooled plasma was used as the starting material. Cold ethanol fractionation produced a paste containing fraction I + II + III, a supernatant containing fraction I + III, and a fraction II paste. The fraction II paste was dissolved in 0.6% sodium chloride solution, the pH was adjusted to 5.0, and the solution subjected to clarifying depth filtration (ultrafiltration/diafiltration, nominal size: 0.1 mm). Then 1 M sodium acetate was added to the diafiltrated IgG solution to a final concentration of 5.0 mM, the pH was adjusted to 6.0, and the solution was subjected to anion exchange (AEX) chromatography. The unbound fraction was collected, the pH was adjusted to 5.0 and then subjected to the first viral elimination step, solvent/detergent (S/D) treatment. Tri(n-butyl)-phosphate (TNBP) and polysorbate 80 (Tween 80) were added at concentrations of 0.3% and 1.0%, respectively, and the mixture was incubated for 8 h at 25°C. Sodium acetate 1 M was added to the filtrate, which was then subjected to an additional clarifying depth filtration just prior to CEX chromatography to absorb IgG. Following CEX chromatography, the filtrate was washed with equilibrium buffer (20 mM sodium acetate, pH 5.0) and eluted with elution buffer (20 mM sodium acetate with 0.5 M sodium chloride, pH 4.5) to acquire IgG. The salt content was eliminated by ultra/diafiltration (UF/DF). Nanofiltration (Ultipor®VF DV 20, Pall Lifescience, Switzerland) was then performed as the second viral elimination step. Following nanofiltration, in contrast to GC5101, GC5107 undergoes an additional ultrafiltration step prior to formulation. For GC5107 the nanofiltration solution was concentrated using a Millipore Pellicon 2 cassette. Glycine was added to the ultrafiltration solution as a stabilizer prior to adjusting protein concentration and pH. The final product was obtained following sterile filtration. Procoagulant activities were assessed in each of the intermediate and final products.

For the spiking study, commercial FXIa [Hematologic Technologies Industries (HTI), USA] was added to the pre-CEX intermediate products (post-S/D treatment) at concentrations of 2.8 µg/ml and 14.0 µg/ml. FXI content was assessed in the pre- and post-CEX products using the Human Total FXI ELISA Kit (Innovative Research, USA).

FXI/FXIa content in the intermediate and final products was analyzed using Western blot analysis and enzyme-linked immunosorbent assay (ELISA). Procoagulant activity in the final product was analyzed using the thrombin generation assay (TGA), chromogenic FXIa assay, and NaPTT assay. Each method was verified based on ICH Guideline Q2 (R1) (20).

For Western blot analysis, samples were subject to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and electroblotted onto a polyvinylidene difluoride (PVDF) membrane by semi-dry transfer (Bio-Rad, USA). Following overnight blocking at 4°C, primary antibodies were applied. Antibodies included anti-prothrombin [factor II (FII)] mouse monoclonal IgG (Merck), anti-factor VII (FVII) rabbit monoclonal IgG (Abcam), anti-factor IX (FIX) mouse monoclonal IgG (Santa Cruz Biotechnology, USA), anti-factor X (FX) rabbit polyclonal IgG (Abcam), and anti-FXI antibody provided in the Human Total FXI ELISA Kit (Innovative Research, USA). Horseradish peroxidase-conjugated secondary antibodies (Innovative Research, USA), were incubated for 1 h and signals were detected with chemiluminescence reagents (Intron, USA). Band density of FXI/FXIa was calculated using Image Lab 5.2.

ELISA assays were performed using the Human Total FXI ELISA Kit (Innovative Research, USA).

The TGA was conducted in accordance with the CBER Ig-Thrombin Generation Test Protocol (automated version) (21). The final reaction volume consisted of 50% factor XI-deficient plasma (HTI, USA), 1.25% 64 mM substrate (Bachem, Switzerland), 2.5% 0.5 mM phospholipid (Rossix, Sweden), 2.5% 6 mM tissue factor (Dade Innovin®, USA), 8.75% 114.5 mM CaCl₂ (Sigma Aldrich, USA), and 35% diluted test sample. FXIa product [National Institute of Biological Standards and Control (NIBSC), 13/100] was diluted to concentrations of 5.0, 2.5, 1.25, 0.625, 0.313, and 0.156 mIU/ml as the standard curve. Standard FXIa was spiked into a sample as a positive control.

Activated thrombin in the test samples was measured at 37°C for 1 h at 40-second intervals using a fluorescent reader (Infinite F-500, Tecan, Switzerland; Ex: 380 nm, Em: 430 nm). The maximum thrombin value was calculated using the Origin Program. Then, a 4-parameter curve was plotted using the Softmax program with the concentration of the standard solution and the maximum thrombin value of the standard solution. The FXIa content of the sample is an average value converted from each dilution factor (duplicated at 3 concentrations for each sample). The limit of detection is 1.56 mIU/ml, confirmed from analytical method validation.

FXIa functional activity in the final product was measured using the Rox Factor XIa kit (Rossix, Sweden). Samples were analyzed according to the kit instructions. In this assay, FXIa present in the test sample activates FIX to FIXa in the presence of calcium ions (CaCl₂). Generated FIXa then activates FX (FXa) in the presence of FVIII, phospholipids and CaCl₂. FXa then hydrolyzes a chromogenic substrate (Z-D-Arg-Gly-Arg-pNA). The resulting product (p-nitroaniline) was measured at a wavelength of 405 nm (reference wavelength: 490 nm). The FXIa calibrator, a kit component, was used as the standard. Results shown represent one assay per sample and each assay was run in duplicate. There were no controls in the assay. The assay used the spiking method for suitable measurement of FXIa.

The NaPTT assay was used to indicate the presence of activated coagulation factors in the final product and was performed according to the European Pharmacopeia 7.0 (22). Samples were diluted 1:5 and 1:10 in Tris-albumin buffer. Tris-albumin buffer and standard- (Factor XIa, NIBSC) spiked Tris-albumin buffer were used as negative and positive controls, respectively. Diluted samples (or controls) were added to tubes containing platelet-poor plasma (Technoclone, Austria) and FXI-deficient plasma (HTI, USA), followed by phospholipid diluted 250-fold (0.25 mM into 1 µM) (Rossix, Sweden) and CaCl₂. Clotting time (time that elapsed between the addition of CaCl₂ and the formation of a clot) was measured using an automated coagulator (ACL-TOP 500, Werfen, USA). The test was considered valid if the clotting time of the negative control was 200–350 s (as suggested by the European Pharmacopeia), the clotting time of the positive control with platelet-poor plasma was <200 s (set for system suitability in the assay), and the negative control with FXI-deficient plasma was >200 s. Minimal and discriminating FXIa concentration of the positive control was determined (100 mIU/ml).

Results of the Western blot analysis performed on intermediate and final products are shown in Figure 2. While FII, FVII, FIX and FX were removed during fractionation (lanes 3–7), residual FXI/FXIa persisted in the fractionation products after AEX but was reduced to undetectable levels by CEX chromatography (Figure 2A). Band intensity analysis of the FXI/FXIa Western blot revealed that residual FXI/FXIa remained present at approximately 16% residual ratio in the intermediate products until after CEX chromatography, when it was reduced to undetectable levels (Figure 2B).

Total FXI levels in each intermediate product in 3 manufacturing lots were analyzed using ELISA. The results are shown in Figure 3A. The FXI log reduction factor at the fractionation I + III step was 1.2, and ≥3.9 at the CEX + UF/DF 2 step (Figures 3B,C). The entire GC5107 manufacturing process resulted in a FXI log reduction factor of ≥5.4. Following the CEX + UF/DF 2 step, the FXI level was less than the limits of quantification.

Because starting plasma can vary in FXIa content, a spiking study was performed to verify the capacity of the GC5107 manufacturing process to remove even large quantities of FXIa. For this study, commercially available FXIa was added to the pre-CEX, post-viral inactivation intermediate product as shown in Figure 4. The average FXI/FXIa concentration of the cryo-poor plasma samples at the start of the GC5107 manufacturing process was approximately 2.8 µg/ml (data not shown). Addition of commercial FXIa at concentrations of 2.8 µg/ml and 14.0 µg/ml (approximately 5 times the concentration in the starting cryo-poor plasma) represents 32.0 times and 158.3 times higher concentrations than that of normal samples, respectively. Figure 4 shows that CEX chromatography was capable of removing >99% of FXI, even in samples containing 158 times greater FXI levels than that of normal samples.

Levels of FXI/FXIa protein concentration and FXIa activity before and after CEX chromatography are shown in Table 1. The concentration of FXI/FXIa protein that remained after CEX chromatography was low, even in the sample spiked with FXIa at levels more than 150 times the amount present in normal specimens. FXI residual ratio calculations were performed to measure the efficiency of the process for removing FXIa. These demonstrated that only 0.1% of the FXI present prior to CEX chromatography remained in the product after this step (Table 1).

In order to determine whether FXIa activity remained after CEX chromatography, procoagulant activity in the pre-and post-CEX samples was assessed using TGA assays. Though procoagulant activity was present in all pre-CEX samples (at extremely high levels in the spiked samples), it was reduced to below detection limits after CEX chromatography, even in the sample spiked with more than 150 times the amount of FXIa present in normal samples (Table 1). The CEX chromatography step had no effect on IgG recovery.

Nine consecutive batches of GC5107 were produced on a commercial scale using the process described in Figure 1. In order to gauge the potential for GC5107 to induce thrombogenic events, procoagulant activity in the final products was assessed using the TGA, chromogenic FXIa and NaPTT assays. These results are shown in Table 2. The results of the TGA and chromogenic FXIa assays in all 9 batches were below the detection limits (<1.56 mIU/ml and <0.16 mIU/ml, respectively). The coagulation time measured using the NaPTT assay was >250 s in all batches and were thus considered to have no procoagulant activity.