- Email: [email protected]
Immunoglobulin G from single plasma donor in immune globulin intravenous causes false positive pyrogen test
Biologicals 59 (2019) 12–19
Contents lists available at ScienceDirect
Biologicals journal homepage: www.elsevier.com/locate/biologicals
Immunoglobulin G from single plasma donor in immune globulin intravenous causes false positive pyrogen test
T
Clark Zervosa,∗, Thomas P. Zimmermanb, Todd Willisb, Greg Flexmanc, Jyoti Srivastavab, Rebecca Silversteinb, Melanie Williamsb,1, Pete Vandebergb, Jennifer L. Culpa, Doug Burnsa, Vickie Barhama, Amy Durhama, David A. Malinzaka a
Grifols Therapeutics LLC, 8368 US 70 Business Highway West, Clayton, NC, 27520, USA Grifols Bioscience Research Group, 85 TW Alexander Drive, Research Triangle Park, NC, 27709, USA c Grifols Worldwide Operations-USA, 8368 US 70 Business Highway West, Building 350, Clayton, NC, 27520, USA b
A R T I C LE I N FO
A B S T R A C T
Keywords: IGIV USP rabbit pyrogen test Monocyte activation test Flow cytometry White blood cells
A sudden, unprecedented failure of USP rabbit pyrogen tests for multiple 10% IGIV-C lots prompted a thorough investigation of the root cause for this phenomenon. All microbe-related testing, including Limulus amebocyte lysate test for endotoxin, proved negative, and no deficiencies were discovered in manufacturing. Plasma pool composition analysis revealed that a single plasma donor (“Donor X″) was common to all pyrogenic IGIV-C lots and that as little as one unit of “Donor X″ plasma (in a pool of ∼4500 units) was sufficient to cause IGIV-C lot failure in the USP rabbit pyrogen test. Whole plasma and Protein A-purified IgG from “Donor X″ caused a temperature increase in rabbits; however, all IgG samples tested pyrogen-negative in two in vitro cell-based pyrogen tests. Flow cytometry showed that “Donor X″ IgG bound strongly to rabbit white blood cells (WBC) but minimally to human WBC. Exclusion of “Donor X″ plasma from manufacturing marked the end of IGIV-C lots registering positive in the USP rabbit pyrogen test. This failure of multiple 10% IGIV-C lots to pass the USP rabbit pyrogen test was demonstrated to be due to the highly unusual anti-rabbit-leukocyte specificity of IgG from a single donor.
1. Introduction The rabbit pyrogen test, with some exceptions, is a required release test on all batches of biological products licensed in the United States [1] and has a long history of use as a test for pyrogenic substances. One biological product that utilizes the rabbit pyrogen test is intravenous immune globulin (IGIV), which is manufactured from pooled human plasma and used for treatment of primary immunodeficiency (PID) as well as a number of neurological indications, including Chronic Inflammatory Demyelinating Polyneuropathy (CIDP) and Idiopathic Thrombocytopenic Purpura (ITP) [2,3]. Manufacture of IGIV consists of alcohol fractionation of pooled plasma from thousands of donors, followed by additional purification steps and formulation for intravenous administration [2,3]. In the preceding 7-year product life cycle history of one IGIV product, 10% IGIV-C [4], there had never been a pyrogen incident up until 2010, when sporadic lots of IGIV-C began to register positive in the USP rabbit pyrogen test. These repeated incidents prompted an extensive investigation of the possible cause(s) for these
unprecedented occurrences. All aspects of the IGIV-C manufacturing process were scrutinized for possible microbial contamination and for possible operational irregularities. Diverse chromatographic, physicochemical, and biochemical methods were investigated for their potential ability to separate a putative pyrogenic substance from IgG in the affected IGIV-C lots. Two different in vitro cell-based pyrogen tests were used to cross-examine the USP rabbit pyrogen test results. Donor composition of plasma pools yielding the rabbit-positive IGIV-C lots was analyzed in search for a possible cause for the temperature response in rabbits. Discovery that a single donor (“Donor X″) was the apparent cause for IGIV-C lots failing the USP rabbit pyrogen test led to demonstration that IgG from this donor exhibited highly unusual interactions with rabbit leukocytes.
∗
Corresponding author. E-mail address: [email protected] (C. Zervos). 1 Current address: Bioagilytix, 2300 Englert Dr, Durham, NC 27,713, USA. https://doi.org/10.1016/j.biologicals.2019.04.001 Received 2 October 2018; Received in revised form 7 March 2019; Accepted 5 April 2019 Available online 22 April 2019 1045-1056/ © 2019 Published by Elsevier Ltd on behalf of International Alliance for Biological Standardization.
Biologicals 59 (2019) 12–19
C. Zervos, et al.
2. Materials and methods
antibody-peroxidase was detected with the chromogenic substrate tetramethylbenzidine. Sample response was compared to a commercial preparation of LTA from Streptococcus pyogenes (Sigma L3140). The LOQ was 500 ng/mL. A commercial kit (Fungitell™, Quest Diagnostics) was used to test for fungal (1–3)-β-D-glucans. A commercial test kit using silkworm larva plasma (Wako Chemicals USA, Inc., Richmond, VA) was used for the detection of peptidoglycans. Gram-negative flagellin was assayed by ELISA using anti-Flagellin antibody (rabbit polyclonal reactive to E.coli and Salmonella, Abcam ab93713) and FliC antibody (mouse monoclonal reactive to flagellin subunit FliC-1, Novus Biologicals NB100-78498). A commercial Flagellin (FliC) preparation from Salmonella typhimurium was used as a standard (Enzo ALX-522-058).
2.1. USP rabbit pyrogen test The rabbit pyrogen test was carried out according to the method defined in the U.S. Pharmacopeia (USP) [5], with adherence to the National Research Council guide for the care and use of laboratory animals [6]. 2.2. In vitro pyrogen tests 2.2.1. Limulus amebocyte lysate (LAL) test The LAL test detects endotoxins, which consist of lipopolysaccharide (LPS) from Gram-negative bacterial membranes. Determination of endotoxin by the LAL test was conducted by the kinetic-turbidimetric procedure using Endosafe® KTA LAL reagent (Charles River Endosafe, Charleston, SC), according to the manufacturer's instructions. Method principles followed the USP test for bacterial endotoxins [7].
2.3. Enzymatic incubations Samples (120 mL) of 10% IGIV-C were incubated with several different enzyme preparations selected to degrade possible microbial pyrogens. The enzymes used, their sources, the amount used, the incubation pH values, the incubation times, and the incubation temperatures were as follows: recombinant bovine pancreatic deoxyribonuclease I (Worthington Biochemical Corporation, Lakewood, NJ; 4200 units; pH 4.8; 3 h; room temperature) plus porcine spleen deoxyribonuclease II (Worthington Biochemical Corporation; 8300 units; pH 4.8; 3 h; room temperature); egg white lysozyme immobilized on agarose (Worthington Biochemical Corporation; 152,000 units; pH 7.0; 4 h; room temperature); Candida antarctica lipase immobilized on agarose (Sigma-Aldrich; 420 units; pH 7.2; 5.5 h; room temperature). The immobilized lysozyme and Candida antarctica lipase were removed by filtration prior to testing in rabbits. Lipase and deoxyribonuclease I and II were soluble and remained in the rabbit test samples. Enzymebuffer controls were run in the absence of IGIV-C. For incubation with a mixture of proteases, IVIG-C was diluted to an IgG concentration of 3.2 mg/mL, adjusted to a pH of 7.4, supplemented with plasmin (Grifols Therapeutics LLC; 14 mg) plus trypsin-agarose (Sigma-Aldrich; 50 units), and incubated for 5 h at room temperature with continuous stirring; this mixture was then supplemented with proteinase K-agarose (Sigma-Aldrich; 25 units) and incubated for 16 h at 40 °C with continuous stirring. After filtration, the filtrate was concentrated to an IgG concentration of 100 mg/mL. IGIV-C (10%, 150 mL) was adjusted to a pH of 7.1 and incubated with cysteine-activated papain from papaya latex (Sigma-Aldrich; 2060 units) for 24 h at 37 °C; iodoacetamide (Sigma-Aldrich; 2.0 mM) was then added to alkylate all free sulfhydryl groups present. A similar incubation of 10% IGIV-C was done with pepsin-agarose (Sigma-Aldrich; 25,000 units) at pH 4.2; at the end of this 24-h incubation, the pH was adjusted to 7.1 and the immobilized enzyme was removed by filtration. Degradation of IgG by these various protease digestions of IGIV-C was confirmed in all cases by SDS-PAGE.
2.2.2. Monocyte activation test (MAT) The MAT test detects endotoxins, as well as non-endotoxin pyrogens, including Gram-positive, Gram-negative and fungal pyrogens [8]. The MAT method was performed as described in PhEur 2.6.30 [9]. Reference Standard Endotoxin (USP) and sample dilutions were incubated with freshly drawn human whole blood from healthy, feverfree individuals. IL-6 secreted in response to pyrogens was measured using commercially available antibodies to IL6. Microwells were coated with monoclonal anti-IL6 antibody (Invitrogen M620) to capture IL6 from the whole blood incubations, and then detected with monoclonal anti-IL6 biotin labeled antibody (Invitrogen M21B). Europium-labeled Streptavidin (Perkin Elmer 1244–360) was then bound to the complex and after a wash step, Enhancement Solution (Perkin Elmer 1244–105) was added and the resulting fluorescence was measured using time resolved fluorescence (excitation and emission wavelengths 330 nm and 620 nm respectively). To insure the method would also respond to a non-endotoxin pyrogen, lipoteichoic acid (LTA, Sigma-Aldrich, L-2515) was used as a positive control. 2.2.3. A toll-like receptor (TLR)-transfected-cell-based pyrogen detection test system A TLR-transfected-cell-based pyrogen test was performed by the Fraunhofer Institute for Interfacial Engineering and Biotechnology (Stuttgart, Germany) [10]. The cell-based test system allows pathogenassociated molecular patterns to be identified and differentiated via toll-like receptors (TLRs). The test system is based on the transfection of NIH3T3 cells (mouse fibroblasts) with NF-κB-inducible reporter-genesecreted alkaline phosphatase and human TLR9 or combinations of human TLRs. These transfected cell lines are capable of evaluating the response of an individual human TLR or TLR combination, without the interference from other receptors. Alkaline phosphatase secreted during a positive response is measured through a colorimetric substrate reaction which is read photometrically at 405 nm, 60 min after addition of substrate.
2.4. Organic extractions Samples (200 mL) of 10% IGIV-C were extracted with an equal volume of n-hexane (Fisher Scientific) overnight at room temperature or with four volumes of cold acetonitrile (Fisher Scientific) 15 min in the cold. The organic phases were isolated, evaporated to dryness, and reconstituted for evaluation in the USP rabbit pyrogen test.
2.2.4. Other in vitro pyrogen tests Lipoteichoic acid (LTA), a pyrogenic substance from Gram-positive organisms, was assayed by ELISA using commercial monoclonal antibodies to LTA. Microwell plates were coated with capture antibody (Thermo MA1-7401, Pierce Biotechnology, Rockford, IL) diluted in carbonate buffer. Plates were then coated with an albumin blocker. Test samples were diluted 1:10 and 1:100 and added to the coated microwells. Following incubation, microwells were washed prior to addition of a second monoclonal antibody to LTA (Thermo MA1-40134, Pierce Biotechnology, Rockford, IL). Plates were then washed prior to incubation with a goat anti-mouse antibody conjugated to peroxidase (Thermo No. 31,430, Pierce Biotechnology, Rockford, IL). Bound
2.5. Size fractionation by ultrafiltration (UF) Samples (200 mL) of 10% IGIV-C were subjected to UF using 50 cm2 Pellicon XL 50 cassettes and Labscale Tangential Flow Filtration System (EMD Millipore Corporation, Billerica, MA) with Biomax membranes of molecular weight cut-off (MWCO) values of 10, 30, 50, 100, and 300 kDa. Both permeates and retentates were evaluated in the USP rabbit pyrogen test. 13
Biologicals 59 (2019) 12–19
C. Zervos, et al.
2.6. Chromatography
Table 1 Results from eight contemporary IGIV-C lots evaluated in USP rabbit pyrogen test and in the endotoxin (LAL) test.
A 5 cm × 26 cm column (510 mL bed volume) of MabSelect Protein A resin (GE Healthcare Bio-Sciences AB) was operated essentially as recommended by the manufacturer but using 0.2 M glycine (pH 3.6) as elution buffer. The flow-through/wash samples were concentrated to 150 mL (load volume) and diafiltered (DF) against five volumes of phosphate-buffered saline (PBS, pH 7.4). The eluate samples only needed UF because they were eluted in 0.2 M glycine at a pH ∼4.2. After UF/DF, the samples were passed through a 0.22 μm filter and submitted for endotoxin (LAL) and rabbit pyrogen testing. Phenyl Sepharose 6 Fast Flow chromatography and SP Sepharose Fast Flow chromatography (both resins from GE Healthcare BioSciences AB, Uppsala, Sweden) were operated under conditions recommended by the manufacturer.
IGIV-C batch
A B C D E F G H
USP rabbit pyrogen test
Endotoxin
(Total temperature, number of rabbits)
(EU/mL)
Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Failed (5.4 °C, 8 rabbits) Failed (3.8 °C, 8 rabbits) Failed (6.1 °C, 8 rabbits) Failed (7.6 °C, 8 rabbits)
< 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1
30 min, followed by centrifugation at 300g for 5 min. The centrifuge was set for slow deceleration to avoid disruption of the cell pellet. The supernate was decanted manually. The tubes were washed 3 times with 4 mL PBS using the same centrifugation settings. Supernate of the final wash was decanted, followed by addition of 1 mL of AlexaFluor 488 Goat anti-human IgG (H + L) (Life Technologies, Carlsbad, CA) diluted 1:150 in PBS. Tubes were then incubated for 30 min at room temperature in the dark. Following incubation, tubes were centrifuged and washed 3 times as previously described. Following the final wash, the cell pellet was resuspended in 1 mL of PBS for analysis by flow cytometry. (FACScan, Becton Dickinson Biosciences, San Jose, CA).
2.7. White blood cells (WBC) Experiments with rabbit and human WBC were designed to evaluate interactions between these cells and “Donor X″ plasma. Rabbit whole blood was collected from New Zealand White male rabbits. Human whole blood was collected from healthy volunteer donors. Whole blood was collected into Vacutainer® tubes containing EDTA. WBC were isolated from rabbit and human whole blood using ACCUSPIN™ SystemHistopaque®-1077 (Sigma-Aldrich) according to the manufacturer's package insert. Suspensions of rabbit and human WBC were prepared in PBS supplemented with 1% bovine serum albumin (Sigma-Aldrich).
3. Results 2.8. Agglutination and fluorescent microscopy studies 3.1. USP rabbit pyrogen test results Rabbit and human WBC suspensions were washed 3 times in 8 mL PBS. Each wash step consisted of gentle mixing, followed by a 10 min centrifugation at 800g. The supernate was decanted, and the WBC pellet was resuspended by gentle vortexing in PBS. Following the last wash, the WBC pellet was resuspended in 2 mL PBS. A cell count was performed on an automated cell counter, and the cell count was adjusted to approximately 5 × 107/mL. WBC suspensions were incubated with “Donor X″ or control plasma in a 96-well microplate. A volume of 100 μL WBC suspension was combined with 200 μL of “Donor X″ or control plasma. The contents of each microplate well were gently mixed using a pipette. The microplates were incubated at 37 °C for 30 min, followed by centrifugation at 300g for 5 min. The centrifuge was set for slow deceleration to avoid disruption of the cell pellet. The supernate was decanted manually. The microplates were washed 3 times with 250 μL PBS using the same centrifugation settings. Supernate of the final wash was decanted, followed by addition of 200 μL of AlexaFluor 488 Goat anti-human IgG (H + L) (Life Technologies, Carlsbad, CA) diluted 1:150 in PBS. Microplates were then incubated for 30 min at room temperature in the dark. Following incubation, the microplate was centrifuged and washed 3 times as previously described. Each well was first examined for agglutination using visible light and was then assessed for fluorescence using an Axiovert 35 fluorescent microscope (Carl Zeiss, Thornwood, NY).
Table 1 summarizes results from four IGIV-C lots which passed the USP rabbit pyrogen test and from four lots which failed. The rabbit temperature increases observed with the latter four lots, which exceeded the US Pharmacopeia allowance for pyrogenic activity, resulted in rejection of the product. 3.2. Investigation of manufacturing conditions An extensive investigation was conducted of the manufacturing process for all lots which failed the USP rabbit pyrogen test. Processing conditions for each lot were examined closely and comprehensively scrutinized for potential sources of bioburden and endotoxin and for any possible deviation or event that could contribute to the rabbit results. Nothing remarkable was discovered in the manufacturing process. 3.3. In vitro pyrogen tests 3.3.1. LAL test Spike-recovery studies performed in the relevant sample matrix demonstrated that the LAL test had a limit of detection of 0.1 endotoxin units (EU)/mL. IGIV-C lots which had passed or failed the USP rabbit pyrogen test were subjected to the LAL test. LAL test results consistently showed a negative response (Table 1), indicating that pyrogenic endotoxin from Gram-negative bacteria was not the cause of the temperature increase in rabbits.
2.9. Flow cytometry Rabbit and human WBC suspensions were washed 3 times in 8 mL PBS. Each wash step consisted of gentle mixing, followed by a 10-min centrifugation at 800g. The supernate was decanted, and the WBC pellet was resuspended by gentle vortexing in PBS. Following the last wash, the WBC pellet was resuspended in 2 mL PBS. A cell count was performed on an automated cell counter, and the cell count was adjusted to approximately 5 × 106 WBC/mL. WBC suspensions were incubated with “Donor X″ or control plasma in Falcon® polystyrene test tubes suitable for flow cytometry analysis. A volume of 500 μL WBC suspension was incubated with 1 mL “Donor X″ or control plasma at 37 °C for
3.3.2. MAT Spiking studies demonstrated that the MAT method was sufficiently sensitive to detect 0.5 EU/mL equivalents of pyrogen (LPS or LTA) in IGIV-C. For endotoxins, the threshold response of the USP rabbit pyrogen test was expected to be approximately 5 EU/kg [11]. Therefore, IGIV-C which elicits a measurable temperature response in rabbits should contain at least 1 EU/mL equivalents of pyrogenic substance. The MAT should thus be able to distinguish pyrogenic and non-pyrogenic IGIV-C samples, if the pyrogen were from a microbial source. 14
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Protein A affinity resins specifically bind the Fc region of IgG subclasses 1, 2, and 4. Isocratic elution chromatography with three different lots of pyrogenic IGIV-C consistently showed that the causative agent in the column load material bound to this affinity resin and was co-eluted with IgG. The small amount of IgG (2.4%) appearing in the flow-through fraction was predominantly IgG3 and did not exhibit detectable pyrogenic activity in rabbits.
IGIV-C lots which had passed or failed the USP rabbit pyrogen test were subjected to the MAT. Samples were tested at 1:2, 1:5, and 1:10 dilutions. None of the samples, at any of these dilutions, gave a positive, pyrogen-indicating response in the assay. 3.3.3. TLR-transfected-cell-based pyrogen detection test system Eight IGIV-C batches, four non-pyrogenic and four rabbit-positive, were diluted to a final IgG concentration of 0.001, 0.02, 0.2, 10, and 20 mg/mL. Each sample was analyzed with transfected cell lines containing human TLR combinations of TLR1/2, TLR2/6, TLR4/CD14, and TLR9 [10]. The same samples were also evaluated using the control NIH3T2 SEAP cell-line without transfected TLRs. All eight IGIV-C batches tested at all final IgG concentrations showed a comparable response between the control cell-line and human TLR-transfected cell lines, demonstrating a consistent negative response.
3.8. Donor analysis of failed IGIV-C lots Analysis of donor center contributions to the IGIV-C lots manufactured during this period of time, based upon a volume-weighted algorithm and relative to temperature increases in the USP rabbit pyrogen test, illustrated a linear relationship to a single plasma donation center (r = 0.58, p = 0.0002, N = 36). A directed audit of this donor center revealed no unusual events or circumstances that could have contributed systemically to the elevated rabbit temperature response observed with the failed IGIV-C lots. Analysis of shipment and manufacturing records identified 72 individual donors from this particular donor center who were represented in each of the failed IGIV-C lots. Although a focused review of these donor files and targeted medical reviews did not reveal anything remarkable, by a process of experimental elimination only one of these 72 donors (“Donor X″) was found to yield IgG that caused a temperature increase in rabbits. Analysis of the correlation between rabbit temperature increases and plasma concentration from all donors in affected lots of IGIV-C, independent of purification method (manufacturing or lab scale), yielded a uniquely strong linear relationship for “Donor X” (r = 0.91, p = 0.0001, N = 193), thus implicating “Donor X″ as the sole cause of the rabbit temperature responses. Direct injection of diluted plasma (10 μL) from “Donor X″ caused a temperature increase in rabbits. By contrast, plasma obtained from the parents, three siblings and three children of “Donor X″ was inert in rabbits. All samples tested in rabbits were shown to be negative in the LAL test. From this analysis it was found that as little as one unit of plasma from “Donor X″, in a plasma pool of approximately 4500 units of plasma, was sufficient to cause the resulting IGIV-C lots to fail in the USP rabbit pyrogen test. The first plasma donation by “Donor X″ coincided with the first failed IGIV-C lot. Subsequent exclusion of “Donor X″ plasma from manufacturing plasma pools marked the end of IGIV-C failures in the USP rabbit pyrogen test. Chromatography of a control plasma pool (devoid of plasma from “Donor X″) on the Protein A column yielded eluate that was nonpyrogenic in rabbits. When this control plasma pool was supplemented with 10% “Donor X″ plasma prior to chromatography, the resulting Protein A column eluate was highly pyrogenic in rabbits. The rabbit temperature response was found to be proportional to the concentration of IgG from “Donor X″, whether plasma was diluted and injected directly, purified on a Protein A column, or purified in manufacturing or in a lab-scale model of the manufacturing process. The relationship was linear and consistent over the “Donor X″ donation period studied.
3.3.4. Other in vitro pyrogen testing LTA was undetectable in all IGIV-C batches by ELISA. No significant differences in β-glucan or peptidoglycan levels were observed between rabbit-positive and rabbit-negative batches of IGIV-C. 3.4. Organic extractions Because several microbial pyrogens are of a lipophilic nature and of relatively low molecular weight, attempts were made to extract the causative agent from pyrogenic IGIV-C lots. Neither n-hexane nor acetonitrile was able to extract a pyrogenic substance from affected lots of IGIV-C. In a parallel investigation, chloroform and methanol extracts of the four pyrogenic IGIV-C lots were analyzed by a metabolomics approach using a proprietary library of approximately 50 pyrogenic agents (Metabolon, Research Triangle Park, NC). No pyrogens were detected. 3.5. Enzymatic degradation studies Different enzymes that specifically target susceptible chemical bonds in potential microbial pyrogens were used as diagnostic tools to probe the chemical nature of the unidentified pyrogenic substance present in IGIV-C lots. DNase I plus DNase II (targeting potential microbial DNA), lysozyme (targeting potential bacterial peptidoglycans), and lipase (targeting potential bacterial lipopolysaccharide) were all found to be ineffective in reducing the rabbit temperature response to pyrogenic IGIV-C lots. A mixture of three proteases (plasmin, trypsin and proteinase K) completely eliminated the rabbit temperature response to affected IGIVC lots. Similarly, incubation of rabbit-positive IGIV-C with either papain or pepsin resulted in elimination of the rabbit temperature response. These results indicated the likely proteinaceous nature of the active substance causing a temperature increase in rabbits. 3.6. Size fractionation by ultrafiltration
3.9. Agglutination of WBC Rabbit-pyrogenic IGIV-C was fractionated by UF using membranes with molecular weight cut-offs (MWCO) of 10, 30, 50, 100 and 300 kDa. No pyrogenic material passed through membranes with 100 kDa or lower MWCO. However, most of the pyrogenic activity, along with the IgG, permeated the 300 kDa MWCO membrane.
Significant agglutination was observed in test wells containing “Donor X″ plasma and rabbit WBC (Fig. 1A, C, E). No agglutination was observed in wells containing control plasma and rabbit WBC (Fig. 1B, D). These results were reproduced in numerous assays and indicate that “Donor X″ IgG binds to rabbit WBC. In contrast, no agglutination of human WBC was observed with “Donor X″ or control plasma (Fig. 2A and B), indicating that “Donor X″ IgG does not bind to human WBC. During several of the agglutination experiments, a cytopathic effect (CPE) was observed in wells containing rabbit WBC and “Donor X″ plasma. CPE was evaluated in one of these experiments, and 50% CPE was observed in both wells containing “Donor X″ plasma and rabbit WBC. No CPE was observed in wells containing control plasma and rabbit WBC, or in any wells containing human WBC with “Donor X″ or
3.7. Chromatographic investigations Hydrophobic interaction chromatography and ion-exchange chromatography were investigated for their ability to separate a putative pyrogenic substance from IgG in pyrogenic IGIV-C. With Phenyl Sepharose chromatography, the rabbit-positive substance was co-eluted with IgG in the column flow-through. With SP Sepharose chromatography, the rabbit pyrogenic activity was also co-eluted with IgG. 15
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 1. Photomicrographs A, C, and E of test wells containing rabbit WBCs with “Donor X″ plasma showing marked agglutination and strong fluorescence. Photomicrographs B and D of test wells containing rabbit WBCs with control plasma showing minimal fluorescence. Photomicrograph F is a non-fluorescent light source image of D, showing non-agglutinated rabbit WBCs present in the control plasma test well.
presence of “Donor X″ or control IgG. Similarly, rabbit WBC incubated with only fluorochrome-labeled anti-human-IgG produced minimal or no fluorescence. This indicates that the fluorescence observed above with rabbit WBC is specific to IgG present in “Donor X″ plasma.
control plasma. The observation that “Donor X″ plasma is cytopathic to rabbit WBC supported specific binding of “Donor X″ IgG to these cells. 3.10. Fluorescence microscopy Rabbit WBC were strongly fluorescent in wells containing “Donor X″ plasma and fluorochrome-labeled anti-human IgG, compared to a relatively weak degree of fluorescence in wells containing control plasma (Fig. 1). A weak degree of fluorescence was observed for human WBC with both “Donor X″ and control plasma (Fig. 2). Weak fluorescence was also observed for human WBC incubated with only the fluorochrome-labeled anti-human-IgG (no “Donor X″ or control plasma added). This finding indicates that the fluorescence observed with human WBC represents non-specific binding, independent of the
3.11. Flow cytometry Fig. 3 is a histogram overlay showing relative fluorescence intensity of two different rabbit WBC samples that were incubated with “Donor X″ or control plasma. Two additional samples were included as assay controls, the first of which was unstained rabbit WBC (cell control). The second assay control was rabbit WBC treated with fluorochrome-labeled anti-human-IgG alone. Median fluorescence was used to compare the relative fluorescence intensity of the cell populations. The median 16
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 2. Photomicrographs A of test well containing human WBCs with “Donor X″ plasma showing minimal fluorescence. Photomicrograph B of test well containing human WBCs with control plasma showing minimal fluorescence. Photomicrographs C and D are non-fluorescent light source images of A and D showing nonagglutinated human WBCs present in the test wells.
fluorescence of the “Donor X″ sample was 3264, which is significantly higher than that of the control plasma sample with a median fluorescence of 499. Although there is overlap between the “Donor X″ and control populations, there is a distinct difference in fluorescence
intensity. These results correlate with microscopic results, discussed previously, in which rabbit WBC treated with “Donor X″ plasma exhibited significantly stronger fluorescence that those treated with control plasma (Fig. 1) and exhibited a CPE with “Donor X″ plasma.
Fig. 3. Overlay histogram showing relative fluorescence of rabbit WBC treated with “Donor X” (“RWBC-Donor X″, green stripes) or control plasma (“RWBC-Control Plasma”, blue stripes) and then stained with fluorochrome-labeled anti-human-IgG. For comparison, also shown is the relative fluorescence for unstained rabbit WBC (“RWBC-Cell Control”, red lines) and for rabbit WBC treated with fluorochrome-labeled anti-human-IgG alone (“RWBC-Anti-Human-IgG Only”, yellow hatches). 17
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 4. Overlay histogram showing relative fluorescence of human WBC treated with “Donor X” (“HWBC-Donor X″, green stripes) or control plasma (“HWBC-Control Plasma”, blue stripes) and then stained with fluorochrome-labeled anti-human-IgG. For comparison, also shown is the relative fluorescence for unstained human WBC (“HWBC-Cell Control”, red lines) and for human WBC treated with fluorochrome-labeled anti-human-IgG alone (“HWBC-Anti-Human-IgG Only”, yellow hatches).
Fig. 4 is a histogram overlay showing relative fluorescence intensity of two different human WBC samples that were incubated with “Donor X″ or control plasma. Two additional samples were included as assay controls, the first of which was unstained human WBC (cell control). The second assay control was human WBC treated with fluorochromelabeled-anti-human-IgG alone. The similarity between “Donor X″ and control histograms indicates no significant difference in fluorescence intensity, with median fluorescence of 48 and 37, respectively. These results correlate with microscopic study results, discussed previously, in which human WBC treated with “Donor X″ and control plasma exhibited only weak fluorescence (Fig. 2A and B). The human WBC sample stained with only fluorochrome-labeled anti-human-IgG also showed significant overlap with “Donor X″ and control plasma samples, indicating a significant degree of non-specific binding by the secondary antibody.
coagulated plasma/kg of rabbit body weight was sufficient to give a strong pyrogenic response. This rabbit reactivity of “Donor X″ plasma was absent in serum samples obtained from the donor's parents, siblings and children and, therefore, appeared to be the result of an acquired immunity rather than a dominant allele-based effect. In a similar manner, IgG purified on a Protein A column from “Donor X″-free plasma supplemented with 10% “Donor X″ plasma gave a strong febrile response in rabbits. Ultimately, exclusion of “Donor X″ plasma from manufacturing pools coincided with an end to IGIV-C lots that failed the USP rabbit pyrogen test. Two types of experimental evidence demonstrated that “Donor X″ IgG bound to rabbit WBC but not to human WBC. First, IgG from “Donor X″ plasma caused agglutination of rabbit WBC (Fig. 1), and a significant cytopathic effect, but did not agglutinate human WBC or cause a cytopathic effect with these cells; by contrast, control (i.e., non“Donor X”) human plasma did not cause agglutination of either rabbit or human WBC (Figs. 1 and 2). Secondly, flow cytometry was used to show that “Donor X″ IgG bound to a much greater extent to rabbit WBC than to human WBC (Figs. 3 and 4). By comparison, control human IgG bound weakly to both rabbit and human WBC. These results are interpreted to indicate that an unusual specificity present in “Donor X″ IgG towards an antigen on rabbit WBC triggers release of a pyrogenic cytokine from these cells that, in turn, triggers a febrile response in rabbits. Because “Donor X″ IgG did not bind appreciably to human WBC, “Donor X″ IgG would not be expected to cause release of pyrogenic cytokines from human WBC or to be pyrogenic in humans. This expectation is supported by the negative results obtained with “Donor X″ IgG in the MAT. The likely absence of human pyrogenic activity in these affected lots of IGIV-C is also supported by their negative results in the LAL assay. The USP rabbit pyrogen test for parenteral pharmaceuticals, introduced in the 1940s, has been mandated by regulatory agencies for decades to limit the risk of a febrile reaction in patients. The rationale for this test is that rabbits are sensitive to the presence of pyrogens, so that an increase in body temperature in rabbits can signal the presence of a pyrogen in the test material that may adversely affect human patients. Unless an alternative test has been validated and approved by a regulatory agency [12,13], every lot of parenteral pharmaceutical manufactured must be shown to be devoid of a febrile response in the USP rabbit pyrogen test prior to its release for human use. With increasing understanding of the cellular and molecular mechanism of the febrile response, in vitro assays based upon the use of human peripheral blood monocytes have been shown to demonstrate high sensitivity for the detection of a broad range of pyrogens [14].
4. Discussion The sudden, unprecedented failure of USP rabbit pyrogen tests for multiple lots of 10% IGIV-C (Table 1) prompted an extensive investigation to identify the root cause for these failed lots and to take corrective action. Various complementary strategies were pursued simultaneously: analysis for microbial contamination; scrutiny of manufacturing practices and chemicals; physicochemical, biochemical and chromatographic attempts to separate a pyrogenic substance from IgG in affected lots of IGIV-C; and determination of donor composition of plasma pools that generated IGIV-C lots that failed the USP rabbit pyrogen test. In no case was the cause of the temperature increase in rabbits found to be separable from IgG by diverse physicochemical, biochemical and chromatographic techniques. This rabbit-pyrogenic activity exhibited a molecular weight similar to that of IgG, was degraded by proteases, and was inseparable from IgG by different types of chromatography. Perhaps the most convincing evidence that this rabbitpyrogenic activity was intrinsic to IgG was its binding and co-elution from Protein A affinity columns. Importantly, there were no rabbit pyrogen test failures in other products (non-IgG) derived from plasma pools containing “Donor X″ plasma. Discovery of the direct relationship between a single plasma donor (“Donor X″) and all of the IGIV-C lots that failed in the USP rabbit pyrogen test allowed for a more focused investigation of the mechanism by which IgG from this unique donor caused a temperature increase in rabbits. Whole plasma from “Donor X″ was highly active in causing a febrile response in rabbits: injection of as little as 10 μL of citrate-anti18
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Acknowledgements
Monocyte activation tests (MAT), in their various forms, offer an attractive alternative to the USP rabbit pyrogen test. In addition to the use of human blood in MAT, human TLR-transfected cell lines have been shown to exhibit high sensitivity to their TLR-specific pyrogens [10], and the human monocytic cell line Mono Mac 6 has been reported to detect different pyrogens with good sensitivity [15]. Indeed, results from our “Donor X″ investigation illustrate how the use of alternative in vitro pyrogen assays, such as LAL and MAT, may be more appropriate than the rabbit pyrogen test, and ultimately prevent the unnecessary rejection of acceptable product. The febrile response elicited in rabbits by IgG from “Donor X″ was shown to be a rabbit-specific phenomenon, caused by the reactivity of “Donor X″ IgG with rabbit leukocytes. This latter use of in vitro pyrogen release assays would be buttressed by routine bioburden testing of process intermediates that is now commonplace. Minimal binding of “Donor X″ IgG to human WBC, as well as a lack of cytokine release from human WBC incubated in vitro with “Donor X″derived IGIV-C (in the MAT), supports the conclusion that IGIV-C containing “Donor X″ IgG was highly unlikely to be pyrogenic in humans. Moreover, results from this study highlight the utility of in vitro pyrogen assays (LAL assay and MAT), not only in pyrogen investigations, but also as a suitable alternative to the USP rabbit pyrogen test for routine IGIV product release testing. Indeed, replacement of the rabbit pyrogen test with in vitro methods is encouraged by European regulatory authorities for plasma-derived products [13].
We would like to thank Chris Barbour, Rusty Barbour, Jeff Huffman, James Wooten, Stuart Roberts and Robert Parker for assistance in sample preparation; Catherine Russ, Katherine Tull, Kyle Rogers, Tabitha Leinbach, Eileen Hawkins, Tabitha Marcum, Dava Quinn. Elizabeth Pittman, John Terry, Lisha Harris, Penny Massey, Amber Gothe and James Buie for study execution; Michelle Woznichak, and Gisela Fleming for study design and data analysis; Jonathan Kent, Michelle Faucette, Bill Burns and Chad Ennis for assisting with investigation of manufacturing operations; Donna Lee, Lisa Musmanno, and Carl Gray for quality oversight and investigation support; Tom Lynch, John Parrish, and James Kennamer for leadership and advice. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.biologicals.2019.04.001. References [1] Code of Federal Regulations, Title 21, food and drugs, chapter 1, Food and Drug Administration, Department of Health and Human Services, Part 610, Subpart B, General Provisions, 610.13. [2] Orange JS, et al. Use of intravenous immunoglobulin in human disease: a review of evidence by members of the primary immunodeficiency committee of the American academy of allergy, asthma and immunology. J Allergy Clin Immunol 2006;117:S525–53. [3] Lunemann JD, Nimmerjahn F, Dalakas MC. Intravenous immunoglobulin in neurology—mode of action and clinical efficacy. Nat Rev Neurol 2015;11:80–9. [4] Lebing W, Remington KM, Schreiner C, Paul HI. Properties of a new intravenous immunoglobulin (IGIV-C, 10%) produced by virus inactivation with caprylate and column chromatography. Vox Sang 2003;84:193–201. [5] USP 36-NF 31 vol. 1. Rockville, MD: U.S. Pharmacopeial Convention; 2013. p. 130–1. [6] National Research Council. Guide for the care and use of laboratory animals. [7] USP 36-NF 31 vol. 1. Rockville, MD: U.S. Pharmacopeial Convention; 2013. p. 90–4. [8] Hasiwa N, et al. Evidence for the detection of non-endotoxin pyrogens by the whole blood monocyte activation test. ALTEX 2013;30:169–208. [9] Monocyte activation test (2.6.30). Pharmeuropa 2008;20(3):505–11. [10] Burger-Kentischer A, Abele IS, Finkelmeier D, Wiesmuller KH, Rupp S. A new cellbased innate immune receptor assay for the examination of receptor activity, ligand specificity, signaling pathways and the detection of pyrogens. J Immunol Methods 2010;358:93–103. [11] Akers MJ, Larrimore DS, Guazzo DA. Parenteral quality control: sterility, Pyrogen,Particulate and package integrity. third ed. New York: Marcel Dekker; 2002. [12] Guidance for industry, pyrogen and endotoxins testing: questions and answers. U.S. Department of Health and Human Services, FDA, CDER, CBER, CVM, CDRH, ORA; 2012. [13] Guideline on the replacement of rabbit pyrogen testing by an alternative test for plasma derived medicinal products. European Medicines Agency, CHMP; 2009. [14] Schindler S, von Aulock S, Daneshian M, Hartung T. Development, validation and applications of the monocyte activation test for pyrogens based on human whole blood. ALTEX 2009;26:265–77. [15] Moesby L, Hansen EW, Christensen JD. Ultrasonication of pyrogenic microorganisms improves the detection of pyrogens in the Mono Mac 6 assay. Eur J Pharm Sci 2000;11:51–7.
Authors contribution Clark Zervos and Vickie Barham: Experimental design, conducted rabbit pyrogen testing, fluorescence microscopy and flow cytometry. Thomas Zimmerman: Supervised chromatography and ultrafiltration fractionation studies, conducted enzymatic degradation studies, organic extractions and wrote draft of manuscript. Rebecca Silverstein: Conducted Protein A chromatography. Todd Willis, Pete Vandeberg and Melanie Williams: Conducted and directed in vitro pyrogen testing. Greg Flexman and Jyoti Srivastava: Traced “Donor X″ to a specific plasma donation center and to a particular donor. Doug Burns, Amy Durham and Jennifer Culp: Investigated manufacturing facility, procedures, records and supplies. David Malinzak: Experimental design, fluorescence microscopy. Conflicts of interest All authors except MW and DM are employees of Grifols Therapeutics LLC and have no other conflicts of interest to declare. MW and DM were employed by Grifols Therapeutics LLC at the time this research was performed and have no other conflicts of interest to declare.
19
Contents lists available at ScienceDirect
Biologicals journal homepage: www.elsevier.com/locate/biologicals
Immunoglobulin G from single plasma donor in immune globulin intravenous causes false positive pyrogen test
T
Clark Zervosa,∗, Thomas P. Zimmermanb, Todd Willisb, Greg Flexmanc, Jyoti Srivastavab, Rebecca Silversteinb, Melanie Williamsb,1, Pete Vandebergb, Jennifer L. Culpa, Doug Burnsa, Vickie Barhama, Amy Durhama, David A. Malinzaka a
Grifols Therapeutics LLC, 8368 US 70 Business Highway West, Clayton, NC, 27520, USA Grifols Bioscience Research Group, 85 TW Alexander Drive, Research Triangle Park, NC, 27709, USA c Grifols Worldwide Operations-USA, 8368 US 70 Business Highway West, Building 350, Clayton, NC, 27520, USA b
A R T I C LE I N FO
A B S T R A C T
Keywords: IGIV USP rabbit pyrogen test Monocyte activation test Flow cytometry White blood cells
A sudden, unprecedented failure of USP rabbit pyrogen tests for multiple 10% IGIV-C lots prompted a thorough investigation of the root cause for this phenomenon. All microbe-related testing, including Limulus amebocyte lysate test for endotoxin, proved negative, and no deficiencies were discovered in manufacturing. Plasma pool composition analysis revealed that a single plasma donor (“Donor X″) was common to all pyrogenic IGIV-C lots and that as little as one unit of “Donor X″ plasma (in a pool of ∼4500 units) was sufficient to cause IGIV-C lot failure in the USP rabbit pyrogen test. Whole plasma and Protein A-purified IgG from “Donor X″ caused a temperature increase in rabbits; however, all IgG samples tested pyrogen-negative in two in vitro cell-based pyrogen tests. Flow cytometry showed that “Donor X″ IgG bound strongly to rabbit white blood cells (WBC) but minimally to human WBC. Exclusion of “Donor X″ plasma from manufacturing marked the end of IGIV-C lots registering positive in the USP rabbit pyrogen test. This failure of multiple 10% IGIV-C lots to pass the USP rabbit pyrogen test was demonstrated to be due to the highly unusual anti-rabbit-leukocyte specificity of IgG from a single donor.
1. Introduction The rabbit pyrogen test, with some exceptions, is a required release test on all batches of biological products licensed in the United States [1] and has a long history of use as a test for pyrogenic substances. One biological product that utilizes the rabbit pyrogen test is intravenous immune globulin (IGIV), which is manufactured from pooled human plasma and used for treatment of primary immunodeficiency (PID) as well as a number of neurological indications, including Chronic Inflammatory Demyelinating Polyneuropathy (CIDP) and Idiopathic Thrombocytopenic Purpura (ITP) [2,3]. Manufacture of IGIV consists of alcohol fractionation of pooled plasma from thousands of donors, followed by additional purification steps and formulation for intravenous administration [2,3]. In the preceding 7-year product life cycle history of one IGIV product, 10% IGIV-C [4], there had never been a pyrogen incident up until 2010, when sporadic lots of IGIV-C began to register positive in the USP rabbit pyrogen test. These repeated incidents prompted an extensive investigation of the possible cause(s) for these
unprecedented occurrences. All aspects of the IGIV-C manufacturing process were scrutinized for possible microbial contamination and for possible operational irregularities. Diverse chromatographic, physicochemical, and biochemical methods were investigated for their potential ability to separate a putative pyrogenic substance from IgG in the affected IGIV-C lots. Two different in vitro cell-based pyrogen tests were used to cross-examine the USP rabbit pyrogen test results. Donor composition of plasma pools yielding the rabbit-positive IGIV-C lots was analyzed in search for a possible cause for the temperature response in rabbits. Discovery that a single donor (“Donor X″) was the apparent cause for IGIV-C lots failing the USP rabbit pyrogen test led to demonstration that IgG from this donor exhibited highly unusual interactions with rabbit leukocytes.
∗
Corresponding author. E-mail address: [email protected] (C. Zervos). 1 Current address: Bioagilytix, 2300 Englert Dr, Durham, NC 27,713, USA. https://doi.org/10.1016/j.biologicals.2019.04.001 Received 2 October 2018; Received in revised form 7 March 2019; Accepted 5 April 2019 Available online 22 April 2019 1045-1056/ © 2019 Published by Elsevier Ltd on behalf of International Alliance for Biological Standardization.
Biologicals 59 (2019) 12–19
C. Zervos, et al.
2. Materials and methods
antibody-peroxidase was detected with the chromogenic substrate tetramethylbenzidine. Sample response was compared to a commercial preparation of LTA from Streptococcus pyogenes (Sigma L3140). The LOQ was 500 ng/mL. A commercial kit (Fungitell™, Quest Diagnostics) was used to test for fungal (1–3)-β-D-glucans. A commercial test kit using silkworm larva plasma (Wako Chemicals USA, Inc., Richmond, VA) was used for the detection of peptidoglycans. Gram-negative flagellin was assayed by ELISA using anti-Flagellin antibody (rabbit polyclonal reactive to E.coli and Salmonella, Abcam ab93713) and FliC antibody (mouse monoclonal reactive to flagellin subunit FliC-1, Novus Biologicals NB100-78498). A commercial Flagellin (FliC) preparation from Salmonella typhimurium was used as a standard (Enzo ALX-522-058).
2.1. USP rabbit pyrogen test The rabbit pyrogen test was carried out according to the method defined in the U.S. Pharmacopeia (USP) [5], with adherence to the National Research Council guide for the care and use of laboratory animals [6]. 2.2. In vitro pyrogen tests 2.2.1. Limulus amebocyte lysate (LAL) test The LAL test detects endotoxins, which consist of lipopolysaccharide (LPS) from Gram-negative bacterial membranes. Determination of endotoxin by the LAL test was conducted by the kinetic-turbidimetric procedure using Endosafe® KTA LAL reagent (Charles River Endosafe, Charleston, SC), according to the manufacturer's instructions. Method principles followed the USP test for bacterial endotoxins [7].
2.3. Enzymatic incubations Samples (120 mL) of 10% IGIV-C were incubated with several different enzyme preparations selected to degrade possible microbial pyrogens. The enzymes used, their sources, the amount used, the incubation pH values, the incubation times, and the incubation temperatures were as follows: recombinant bovine pancreatic deoxyribonuclease I (Worthington Biochemical Corporation, Lakewood, NJ; 4200 units; pH 4.8; 3 h; room temperature) plus porcine spleen deoxyribonuclease II (Worthington Biochemical Corporation; 8300 units; pH 4.8; 3 h; room temperature); egg white lysozyme immobilized on agarose (Worthington Biochemical Corporation; 152,000 units; pH 7.0; 4 h; room temperature); Candida antarctica lipase immobilized on agarose (Sigma-Aldrich; 420 units; pH 7.2; 5.5 h; room temperature). The immobilized lysozyme and Candida antarctica lipase were removed by filtration prior to testing in rabbits. Lipase and deoxyribonuclease I and II were soluble and remained in the rabbit test samples. Enzymebuffer controls were run in the absence of IGIV-C. For incubation with a mixture of proteases, IVIG-C was diluted to an IgG concentration of 3.2 mg/mL, adjusted to a pH of 7.4, supplemented with plasmin (Grifols Therapeutics LLC; 14 mg) plus trypsin-agarose (Sigma-Aldrich; 50 units), and incubated for 5 h at room temperature with continuous stirring; this mixture was then supplemented with proteinase K-agarose (Sigma-Aldrich; 25 units) and incubated for 16 h at 40 °C with continuous stirring. After filtration, the filtrate was concentrated to an IgG concentration of 100 mg/mL. IGIV-C (10%, 150 mL) was adjusted to a pH of 7.1 and incubated with cysteine-activated papain from papaya latex (Sigma-Aldrich; 2060 units) for 24 h at 37 °C; iodoacetamide (Sigma-Aldrich; 2.0 mM) was then added to alkylate all free sulfhydryl groups present. A similar incubation of 10% IGIV-C was done with pepsin-agarose (Sigma-Aldrich; 25,000 units) at pH 4.2; at the end of this 24-h incubation, the pH was adjusted to 7.1 and the immobilized enzyme was removed by filtration. Degradation of IgG by these various protease digestions of IGIV-C was confirmed in all cases by SDS-PAGE.
2.2.2. Monocyte activation test (MAT) The MAT test detects endotoxins, as well as non-endotoxin pyrogens, including Gram-positive, Gram-negative and fungal pyrogens [8]. The MAT method was performed as described in PhEur 2.6.30 [9]. Reference Standard Endotoxin (USP) and sample dilutions were incubated with freshly drawn human whole blood from healthy, feverfree individuals. IL-6 secreted in response to pyrogens was measured using commercially available antibodies to IL6. Microwells were coated with monoclonal anti-IL6 antibody (Invitrogen M620) to capture IL6 from the whole blood incubations, and then detected with monoclonal anti-IL6 biotin labeled antibody (Invitrogen M21B). Europium-labeled Streptavidin (Perkin Elmer 1244–360) was then bound to the complex and after a wash step, Enhancement Solution (Perkin Elmer 1244–105) was added and the resulting fluorescence was measured using time resolved fluorescence (excitation and emission wavelengths 330 nm and 620 nm respectively). To insure the method would also respond to a non-endotoxin pyrogen, lipoteichoic acid (LTA, Sigma-Aldrich, L-2515) was used as a positive control. 2.2.3. A toll-like receptor (TLR)-transfected-cell-based pyrogen detection test system A TLR-transfected-cell-based pyrogen test was performed by the Fraunhofer Institute for Interfacial Engineering and Biotechnology (Stuttgart, Germany) [10]. The cell-based test system allows pathogenassociated molecular patterns to be identified and differentiated via toll-like receptors (TLRs). The test system is based on the transfection of NIH3T3 cells (mouse fibroblasts) with NF-κB-inducible reporter-genesecreted alkaline phosphatase and human TLR9 or combinations of human TLRs. These transfected cell lines are capable of evaluating the response of an individual human TLR or TLR combination, without the interference from other receptors. Alkaline phosphatase secreted during a positive response is measured through a colorimetric substrate reaction which is read photometrically at 405 nm, 60 min after addition of substrate.
2.4. Organic extractions Samples (200 mL) of 10% IGIV-C were extracted with an equal volume of n-hexane (Fisher Scientific) overnight at room temperature or with four volumes of cold acetonitrile (Fisher Scientific) 15 min in the cold. The organic phases were isolated, evaporated to dryness, and reconstituted for evaluation in the USP rabbit pyrogen test.
2.2.4. Other in vitro pyrogen tests Lipoteichoic acid (LTA), a pyrogenic substance from Gram-positive organisms, was assayed by ELISA using commercial monoclonal antibodies to LTA. Microwell plates were coated with capture antibody (Thermo MA1-7401, Pierce Biotechnology, Rockford, IL) diluted in carbonate buffer. Plates were then coated with an albumin blocker. Test samples were diluted 1:10 and 1:100 and added to the coated microwells. Following incubation, microwells were washed prior to addition of a second monoclonal antibody to LTA (Thermo MA1-40134, Pierce Biotechnology, Rockford, IL). Plates were then washed prior to incubation with a goat anti-mouse antibody conjugated to peroxidase (Thermo No. 31,430, Pierce Biotechnology, Rockford, IL). Bound
2.5. Size fractionation by ultrafiltration (UF) Samples (200 mL) of 10% IGIV-C were subjected to UF using 50 cm2 Pellicon XL 50 cassettes and Labscale Tangential Flow Filtration System (EMD Millipore Corporation, Billerica, MA) with Biomax membranes of molecular weight cut-off (MWCO) values of 10, 30, 50, 100, and 300 kDa. Both permeates and retentates were evaluated in the USP rabbit pyrogen test. 13
Biologicals 59 (2019) 12–19
C. Zervos, et al.
2.6. Chromatography
Table 1 Results from eight contemporary IGIV-C lots evaluated in USP rabbit pyrogen test and in the endotoxin (LAL) test.
A 5 cm × 26 cm column (510 mL bed volume) of MabSelect Protein A resin (GE Healthcare Bio-Sciences AB) was operated essentially as recommended by the manufacturer but using 0.2 M glycine (pH 3.6) as elution buffer. The flow-through/wash samples were concentrated to 150 mL (load volume) and diafiltered (DF) against five volumes of phosphate-buffered saline (PBS, pH 7.4). The eluate samples only needed UF because they were eluted in 0.2 M glycine at a pH ∼4.2. After UF/DF, the samples were passed through a 0.22 μm filter and submitted for endotoxin (LAL) and rabbit pyrogen testing. Phenyl Sepharose 6 Fast Flow chromatography and SP Sepharose Fast Flow chromatography (both resins from GE Healthcare BioSciences AB, Uppsala, Sweden) were operated under conditions recommended by the manufacturer.
IGIV-C batch
A B C D E F G H
USP rabbit pyrogen test
Endotoxin
(Total temperature, number of rabbits)
(EU/mL)
Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Passed (0.0 °C, 3 rabbits) Failed (5.4 °C, 8 rabbits) Failed (3.8 °C, 8 rabbits) Failed (6.1 °C, 8 rabbits) Failed (7.6 °C, 8 rabbits)
< 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1 < 0.1
30 min, followed by centrifugation at 300g for 5 min. The centrifuge was set for slow deceleration to avoid disruption of the cell pellet. The supernate was decanted manually. The tubes were washed 3 times with 4 mL PBS using the same centrifugation settings. Supernate of the final wash was decanted, followed by addition of 1 mL of AlexaFluor 488 Goat anti-human IgG (H + L) (Life Technologies, Carlsbad, CA) diluted 1:150 in PBS. Tubes were then incubated for 30 min at room temperature in the dark. Following incubation, tubes were centrifuged and washed 3 times as previously described. Following the final wash, the cell pellet was resuspended in 1 mL of PBS for analysis by flow cytometry. (FACScan, Becton Dickinson Biosciences, San Jose, CA).
2.7. White blood cells (WBC) Experiments with rabbit and human WBC were designed to evaluate interactions between these cells and “Donor X″ plasma. Rabbit whole blood was collected from New Zealand White male rabbits. Human whole blood was collected from healthy volunteer donors. Whole blood was collected into Vacutainer® tubes containing EDTA. WBC were isolated from rabbit and human whole blood using ACCUSPIN™ SystemHistopaque®-1077 (Sigma-Aldrich) according to the manufacturer's package insert. Suspensions of rabbit and human WBC were prepared in PBS supplemented with 1% bovine serum albumin (Sigma-Aldrich).
3. Results 2.8. Agglutination and fluorescent microscopy studies 3.1. USP rabbit pyrogen test results Rabbit and human WBC suspensions were washed 3 times in 8 mL PBS. Each wash step consisted of gentle mixing, followed by a 10 min centrifugation at 800g. The supernate was decanted, and the WBC pellet was resuspended by gentle vortexing in PBS. Following the last wash, the WBC pellet was resuspended in 2 mL PBS. A cell count was performed on an automated cell counter, and the cell count was adjusted to approximately 5 × 107/mL. WBC suspensions were incubated with “Donor X″ or control plasma in a 96-well microplate. A volume of 100 μL WBC suspension was combined with 200 μL of “Donor X″ or control plasma. The contents of each microplate well were gently mixed using a pipette. The microplates were incubated at 37 °C for 30 min, followed by centrifugation at 300g for 5 min. The centrifuge was set for slow deceleration to avoid disruption of the cell pellet. The supernate was decanted manually. The microplates were washed 3 times with 250 μL PBS using the same centrifugation settings. Supernate of the final wash was decanted, followed by addition of 200 μL of AlexaFluor 488 Goat anti-human IgG (H + L) (Life Technologies, Carlsbad, CA) diluted 1:150 in PBS. Microplates were then incubated for 30 min at room temperature in the dark. Following incubation, the microplate was centrifuged and washed 3 times as previously described. Each well was first examined for agglutination using visible light and was then assessed for fluorescence using an Axiovert 35 fluorescent microscope (Carl Zeiss, Thornwood, NY).
Table 1 summarizes results from four IGIV-C lots which passed the USP rabbit pyrogen test and from four lots which failed. The rabbit temperature increases observed with the latter four lots, which exceeded the US Pharmacopeia allowance for pyrogenic activity, resulted in rejection of the product. 3.2. Investigation of manufacturing conditions An extensive investigation was conducted of the manufacturing process for all lots which failed the USP rabbit pyrogen test. Processing conditions for each lot were examined closely and comprehensively scrutinized for potential sources of bioburden and endotoxin and for any possible deviation or event that could contribute to the rabbit results. Nothing remarkable was discovered in the manufacturing process. 3.3. In vitro pyrogen tests 3.3.1. LAL test Spike-recovery studies performed in the relevant sample matrix demonstrated that the LAL test had a limit of detection of 0.1 endotoxin units (EU)/mL. IGIV-C lots which had passed or failed the USP rabbit pyrogen test were subjected to the LAL test. LAL test results consistently showed a negative response (Table 1), indicating that pyrogenic endotoxin from Gram-negative bacteria was not the cause of the temperature increase in rabbits.
2.9. Flow cytometry Rabbit and human WBC suspensions were washed 3 times in 8 mL PBS. Each wash step consisted of gentle mixing, followed by a 10-min centrifugation at 800g. The supernate was decanted, and the WBC pellet was resuspended by gentle vortexing in PBS. Following the last wash, the WBC pellet was resuspended in 2 mL PBS. A cell count was performed on an automated cell counter, and the cell count was adjusted to approximately 5 × 106 WBC/mL. WBC suspensions were incubated with “Donor X″ or control plasma in Falcon® polystyrene test tubes suitable for flow cytometry analysis. A volume of 500 μL WBC suspension was incubated with 1 mL “Donor X″ or control plasma at 37 °C for
3.3.2. MAT Spiking studies demonstrated that the MAT method was sufficiently sensitive to detect 0.5 EU/mL equivalents of pyrogen (LPS or LTA) in IGIV-C. For endotoxins, the threshold response of the USP rabbit pyrogen test was expected to be approximately 5 EU/kg [11]. Therefore, IGIV-C which elicits a measurable temperature response in rabbits should contain at least 1 EU/mL equivalents of pyrogenic substance. The MAT should thus be able to distinguish pyrogenic and non-pyrogenic IGIV-C samples, if the pyrogen were from a microbial source. 14
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Protein A affinity resins specifically bind the Fc region of IgG subclasses 1, 2, and 4. Isocratic elution chromatography with three different lots of pyrogenic IGIV-C consistently showed that the causative agent in the column load material bound to this affinity resin and was co-eluted with IgG. The small amount of IgG (2.4%) appearing in the flow-through fraction was predominantly IgG3 and did not exhibit detectable pyrogenic activity in rabbits.
IGIV-C lots which had passed or failed the USP rabbit pyrogen test were subjected to the MAT. Samples were tested at 1:2, 1:5, and 1:10 dilutions. None of the samples, at any of these dilutions, gave a positive, pyrogen-indicating response in the assay. 3.3.3. TLR-transfected-cell-based pyrogen detection test system Eight IGIV-C batches, four non-pyrogenic and four rabbit-positive, were diluted to a final IgG concentration of 0.001, 0.02, 0.2, 10, and 20 mg/mL. Each sample was analyzed with transfected cell lines containing human TLR combinations of TLR1/2, TLR2/6, TLR4/CD14, and TLR9 [10]. The same samples were also evaluated using the control NIH3T2 SEAP cell-line without transfected TLRs. All eight IGIV-C batches tested at all final IgG concentrations showed a comparable response between the control cell-line and human TLR-transfected cell lines, demonstrating a consistent negative response.
3.8. Donor analysis of failed IGIV-C lots Analysis of donor center contributions to the IGIV-C lots manufactured during this period of time, based upon a volume-weighted algorithm and relative to temperature increases in the USP rabbit pyrogen test, illustrated a linear relationship to a single plasma donation center (r = 0.58, p = 0.0002, N = 36). A directed audit of this donor center revealed no unusual events or circumstances that could have contributed systemically to the elevated rabbit temperature response observed with the failed IGIV-C lots. Analysis of shipment and manufacturing records identified 72 individual donors from this particular donor center who were represented in each of the failed IGIV-C lots. Although a focused review of these donor files and targeted medical reviews did not reveal anything remarkable, by a process of experimental elimination only one of these 72 donors (“Donor X″) was found to yield IgG that caused a temperature increase in rabbits. Analysis of the correlation between rabbit temperature increases and plasma concentration from all donors in affected lots of IGIV-C, independent of purification method (manufacturing or lab scale), yielded a uniquely strong linear relationship for “Donor X” (r = 0.91, p = 0.0001, N = 193), thus implicating “Donor X″ as the sole cause of the rabbit temperature responses. Direct injection of diluted plasma (10 μL) from “Donor X″ caused a temperature increase in rabbits. By contrast, plasma obtained from the parents, three siblings and three children of “Donor X″ was inert in rabbits. All samples tested in rabbits were shown to be negative in the LAL test. From this analysis it was found that as little as one unit of plasma from “Donor X″, in a plasma pool of approximately 4500 units of plasma, was sufficient to cause the resulting IGIV-C lots to fail in the USP rabbit pyrogen test. The first plasma donation by “Donor X″ coincided with the first failed IGIV-C lot. Subsequent exclusion of “Donor X″ plasma from manufacturing plasma pools marked the end of IGIV-C failures in the USP rabbit pyrogen test. Chromatography of a control plasma pool (devoid of plasma from “Donor X″) on the Protein A column yielded eluate that was nonpyrogenic in rabbits. When this control plasma pool was supplemented with 10% “Donor X″ plasma prior to chromatography, the resulting Protein A column eluate was highly pyrogenic in rabbits. The rabbit temperature response was found to be proportional to the concentration of IgG from “Donor X″, whether plasma was diluted and injected directly, purified on a Protein A column, or purified in manufacturing or in a lab-scale model of the manufacturing process. The relationship was linear and consistent over the “Donor X″ donation period studied.
3.3.4. Other in vitro pyrogen testing LTA was undetectable in all IGIV-C batches by ELISA. No significant differences in β-glucan or peptidoglycan levels were observed between rabbit-positive and rabbit-negative batches of IGIV-C. 3.4. Organic extractions Because several microbial pyrogens are of a lipophilic nature and of relatively low molecular weight, attempts were made to extract the causative agent from pyrogenic IGIV-C lots. Neither n-hexane nor acetonitrile was able to extract a pyrogenic substance from affected lots of IGIV-C. In a parallel investigation, chloroform and methanol extracts of the four pyrogenic IGIV-C lots were analyzed by a metabolomics approach using a proprietary library of approximately 50 pyrogenic agents (Metabolon, Research Triangle Park, NC). No pyrogens were detected. 3.5. Enzymatic degradation studies Different enzymes that specifically target susceptible chemical bonds in potential microbial pyrogens were used as diagnostic tools to probe the chemical nature of the unidentified pyrogenic substance present in IGIV-C lots. DNase I plus DNase II (targeting potential microbial DNA), lysozyme (targeting potential bacterial peptidoglycans), and lipase (targeting potential bacterial lipopolysaccharide) were all found to be ineffective in reducing the rabbit temperature response to pyrogenic IGIV-C lots. A mixture of three proteases (plasmin, trypsin and proteinase K) completely eliminated the rabbit temperature response to affected IGIVC lots. Similarly, incubation of rabbit-positive IGIV-C with either papain or pepsin resulted in elimination of the rabbit temperature response. These results indicated the likely proteinaceous nature of the active substance causing a temperature increase in rabbits. 3.6. Size fractionation by ultrafiltration
3.9. Agglutination of WBC Rabbit-pyrogenic IGIV-C was fractionated by UF using membranes with molecular weight cut-offs (MWCO) of 10, 30, 50, 100 and 300 kDa. No pyrogenic material passed through membranes with 100 kDa or lower MWCO. However, most of the pyrogenic activity, along with the IgG, permeated the 300 kDa MWCO membrane.
Significant agglutination was observed in test wells containing “Donor X″ plasma and rabbit WBC (Fig. 1A, C, E). No agglutination was observed in wells containing control plasma and rabbit WBC (Fig. 1B, D). These results were reproduced in numerous assays and indicate that “Donor X″ IgG binds to rabbit WBC. In contrast, no agglutination of human WBC was observed with “Donor X″ or control plasma (Fig. 2A and B), indicating that “Donor X″ IgG does not bind to human WBC. During several of the agglutination experiments, a cytopathic effect (CPE) was observed in wells containing rabbit WBC and “Donor X″ plasma. CPE was evaluated in one of these experiments, and 50% CPE was observed in both wells containing “Donor X″ plasma and rabbit WBC. No CPE was observed in wells containing control plasma and rabbit WBC, or in any wells containing human WBC with “Donor X″ or
3.7. Chromatographic investigations Hydrophobic interaction chromatography and ion-exchange chromatography were investigated for their ability to separate a putative pyrogenic substance from IgG in pyrogenic IGIV-C. With Phenyl Sepharose chromatography, the rabbit-positive substance was co-eluted with IgG in the column flow-through. With SP Sepharose chromatography, the rabbit pyrogenic activity was also co-eluted with IgG. 15
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 1. Photomicrographs A, C, and E of test wells containing rabbit WBCs with “Donor X″ plasma showing marked agglutination and strong fluorescence. Photomicrographs B and D of test wells containing rabbit WBCs with control plasma showing minimal fluorescence. Photomicrograph F is a non-fluorescent light source image of D, showing non-agglutinated rabbit WBCs present in the control plasma test well.
presence of “Donor X″ or control IgG. Similarly, rabbit WBC incubated with only fluorochrome-labeled anti-human-IgG produced minimal or no fluorescence. This indicates that the fluorescence observed above with rabbit WBC is specific to IgG present in “Donor X″ plasma.
control plasma. The observation that “Donor X″ plasma is cytopathic to rabbit WBC supported specific binding of “Donor X″ IgG to these cells. 3.10. Fluorescence microscopy Rabbit WBC were strongly fluorescent in wells containing “Donor X″ plasma and fluorochrome-labeled anti-human IgG, compared to a relatively weak degree of fluorescence in wells containing control plasma (Fig. 1). A weak degree of fluorescence was observed for human WBC with both “Donor X″ and control plasma (Fig. 2). Weak fluorescence was also observed for human WBC incubated with only the fluorochrome-labeled anti-human-IgG (no “Donor X″ or control plasma added). This finding indicates that the fluorescence observed with human WBC represents non-specific binding, independent of the
3.11. Flow cytometry Fig. 3 is a histogram overlay showing relative fluorescence intensity of two different rabbit WBC samples that were incubated with “Donor X″ or control plasma. Two additional samples were included as assay controls, the first of which was unstained rabbit WBC (cell control). The second assay control was rabbit WBC treated with fluorochrome-labeled anti-human-IgG alone. Median fluorescence was used to compare the relative fluorescence intensity of the cell populations. The median 16
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 2. Photomicrographs A of test well containing human WBCs with “Donor X″ plasma showing minimal fluorescence. Photomicrograph B of test well containing human WBCs with control plasma showing minimal fluorescence. Photomicrographs C and D are non-fluorescent light source images of A and D showing nonagglutinated human WBCs present in the test wells.
fluorescence of the “Donor X″ sample was 3264, which is significantly higher than that of the control plasma sample with a median fluorescence of 499. Although there is overlap between the “Donor X″ and control populations, there is a distinct difference in fluorescence
intensity. These results correlate with microscopic results, discussed previously, in which rabbit WBC treated with “Donor X″ plasma exhibited significantly stronger fluorescence that those treated with control plasma (Fig. 1) and exhibited a CPE with “Donor X″ plasma.
Fig. 3. Overlay histogram showing relative fluorescence of rabbit WBC treated with “Donor X” (“RWBC-Donor X″, green stripes) or control plasma (“RWBC-Control Plasma”, blue stripes) and then stained with fluorochrome-labeled anti-human-IgG. For comparison, also shown is the relative fluorescence for unstained rabbit WBC (“RWBC-Cell Control”, red lines) and for rabbit WBC treated with fluorochrome-labeled anti-human-IgG alone (“RWBC-Anti-Human-IgG Only”, yellow hatches). 17
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Fig. 4. Overlay histogram showing relative fluorescence of human WBC treated with “Donor X” (“HWBC-Donor X″, green stripes) or control plasma (“HWBC-Control Plasma”, blue stripes) and then stained with fluorochrome-labeled anti-human-IgG. For comparison, also shown is the relative fluorescence for unstained human WBC (“HWBC-Cell Control”, red lines) and for human WBC treated with fluorochrome-labeled anti-human-IgG alone (“HWBC-Anti-Human-IgG Only”, yellow hatches).
Fig. 4 is a histogram overlay showing relative fluorescence intensity of two different human WBC samples that were incubated with “Donor X″ or control plasma. Two additional samples were included as assay controls, the first of which was unstained human WBC (cell control). The second assay control was human WBC treated with fluorochromelabeled-anti-human-IgG alone. The similarity between “Donor X″ and control histograms indicates no significant difference in fluorescence intensity, with median fluorescence of 48 and 37, respectively. These results correlate with microscopic study results, discussed previously, in which human WBC treated with “Donor X″ and control plasma exhibited only weak fluorescence (Fig. 2A and B). The human WBC sample stained with only fluorochrome-labeled anti-human-IgG also showed significant overlap with “Donor X″ and control plasma samples, indicating a significant degree of non-specific binding by the secondary antibody.
coagulated plasma/kg of rabbit body weight was sufficient to give a strong pyrogenic response. This rabbit reactivity of “Donor X″ plasma was absent in serum samples obtained from the donor's parents, siblings and children and, therefore, appeared to be the result of an acquired immunity rather than a dominant allele-based effect. In a similar manner, IgG purified on a Protein A column from “Donor X″-free plasma supplemented with 10% “Donor X″ plasma gave a strong febrile response in rabbits. Ultimately, exclusion of “Donor X″ plasma from manufacturing pools coincided with an end to IGIV-C lots that failed the USP rabbit pyrogen test. Two types of experimental evidence demonstrated that “Donor X″ IgG bound to rabbit WBC but not to human WBC. First, IgG from “Donor X″ plasma caused agglutination of rabbit WBC (Fig. 1), and a significant cytopathic effect, but did not agglutinate human WBC or cause a cytopathic effect with these cells; by contrast, control (i.e., non“Donor X”) human plasma did not cause agglutination of either rabbit or human WBC (Figs. 1 and 2). Secondly, flow cytometry was used to show that “Donor X″ IgG bound to a much greater extent to rabbit WBC than to human WBC (Figs. 3 and 4). By comparison, control human IgG bound weakly to both rabbit and human WBC. These results are interpreted to indicate that an unusual specificity present in “Donor X″ IgG towards an antigen on rabbit WBC triggers release of a pyrogenic cytokine from these cells that, in turn, triggers a febrile response in rabbits. Because “Donor X″ IgG did not bind appreciably to human WBC, “Donor X″ IgG would not be expected to cause release of pyrogenic cytokines from human WBC or to be pyrogenic in humans. This expectation is supported by the negative results obtained with “Donor X″ IgG in the MAT. The likely absence of human pyrogenic activity in these affected lots of IGIV-C is also supported by their negative results in the LAL assay. The USP rabbit pyrogen test for parenteral pharmaceuticals, introduced in the 1940s, has been mandated by regulatory agencies for decades to limit the risk of a febrile reaction in patients. The rationale for this test is that rabbits are sensitive to the presence of pyrogens, so that an increase in body temperature in rabbits can signal the presence of a pyrogen in the test material that may adversely affect human patients. Unless an alternative test has been validated and approved by a regulatory agency [12,13], every lot of parenteral pharmaceutical manufactured must be shown to be devoid of a febrile response in the USP rabbit pyrogen test prior to its release for human use. With increasing understanding of the cellular and molecular mechanism of the febrile response, in vitro assays based upon the use of human peripheral blood monocytes have been shown to demonstrate high sensitivity for the detection of a broad range of pyrogens [14].
4. Discussion The sudden, unprecedented failure of USP rabbit pyrogen tests for multiple lots of 10% IGIV-C (Table 1) prompted an extensive investigation to identify the root cause for these failed lots and to take corrective action. Various complementary strategies were pursued simultaneously: analysis for microbial contamination; scrutiny of manufacturing practices and chemicals; physicochemical, biochemical and chromatographic attempts to separate a pyrogenic substance from IgG in affected lots of IGIV-C; and determination of donor composition of plasma pools that generated IGIV-C lots that failed the USP rabbit pyrogen test. In no case was the cause of the temperature increase in rabbits found to be separable from IgG by diverse physicochemical, biochemical and chromatographic techniques. This rabbit-pyrogenic activity exhibited a molecular weight similar to that of IgG, was degraded by proteases, and was inseparable from IgG by different types of chromatography. Perhaps the most convincing evidence that this rabbitpyrogenic activity was intrinsic to IgG was its binding and co-elution from Protein A affinity columns. Importantly, there were no rabbit pyrogen test failures in other products (non-IgG) derived from plasma pools containing “Donor X″ plasma. Discovery of the direct relationship between a single plasma donor (“Donor X″) and all of the IGIV-C lots that failed in the USP rabbit pyrogen test allowed for a more focused investigation of the mechanism by which IgG from this unique donor caused a temperature increase in rabbits. Whole plasma from “Donor X″ was highly active in causing a febrile response in rabbits: injection of as little as 10 μL of citrate-anti18
Biologicals 59 (2019) 12–19
C. Zervos, et al.
Acknowledgements
Monocyte activation tests (MAT), in their various forms, offer an attractive alternative to the USP rabbit pyrogen test. In addition to the use of human blood in MAT, human TLR-transfected cell lines have been shown to exhibit high sensitivity to their TLR-specific pyrogens [10], and the human monocytic cell line Mono Mac 6 has been reported to detect different pyrogens with good sensitivity [15]. Indeed, results from our “Donor X″ investigation illustrate how the use of alternative in vitro pyrogen assays, such as LAL and MAT, may be more appropriate than the rabbit pyrogen test, and ultimately prevent the unnecessary rejection of acceptable product. The febrile response elicited in rabbits by IgG from “Donor X″ was shown to be a rabbit-specific phenomenon, caused by the reactivity of “Donor X″ IgG with rabbit leukocytes. This latter use of in vitro pyrogen release assays would be buttressed by routine bioburden testing of process intermediates that is now commonplace. Minimal binding of “Donor X″ IgG to human WBC, as well as a lack of cytokine release from human WBC incubated in vitro with “Donor X″derived IGIV-C (in the MAT), supports the conclusion that IGIV-C containing “Donor X″ IgG was highly unlikely to be pyrogenic in humans. Moreover, results from this study highlight the utility of in vitro pyrogen assays (LAL assay and MAT), not only in pyrogen investigations, but also as a suitable alternative to the USP rabbit pyrogen test for routine IGIV product release testing. Indeed, replacement of the rabbit pyrogen test with in vitro methods is encouraged by European regulatory authorities for plasma-derived products [13].
We would like to thank Chris Barbour, Rusty Barbour, Jeff Huffman, James Wooten, Stuart Roberts and Robert Parker for assistance in sample preparation; Catherine Russ, Katherine Tull, Kyle Rogers, Tabitha Leinbach, Eileen Hawkins, Tabitha Marcum, Dava Quinn. Elizabeth Pittman, John Terry, Lisha Harris, Penny Massey, Amber Gothe and James Buie for study execution; Michelle Woznichak, and Gisela Fleming for study design and data analysis; Jonathan Kent, Michelle Faucette, Bill Burns and Chad Ennis for assisting with investigation of manufacturing operations; Donna Lee, Lisa Musmanno, and Carl Gray for quality oversight and investigation support; Tom Lynch, John Parrish, and James Kennamer for leadership and advice. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.biologicals.2019.04.001. References [1] Code of Federal Regulations, Title 21, food and drugs, chapter 1, Food and Drug Administration, Department of Health and Human Services, Part 610, Subpart B, General Provisions, 610.13. [2] Orange JS, et al. Use of intravenous immunoglobulin in human disease: a review of evidence by members of the primary immunodeficiency committee of the American academy of allergy, asthma and immunology. J Allergy Clin Immunol 2006;117:S525–53. [3] Lunemann JD, Nimmerjahn F, Dalakas MC. Intravenous immunoglobulin in neurology—mode of action and clinical efficacy. Nat Rev Neurol 2015;11:80–9. [4] Lebing W, Remington KM, Schreiner C, Paul HI. Properties of a new intravenous immunoglobulin (IGIV-C, 10%) produced by virus inactivation with caprylate and column chromatography. Vox Sang 2003;84:193–201. [5] USP 36-NF 31 vol. 1. Rockville, MD: U.S. Pharmacopeial Convention; 2013. p. 130–1. [6] National Research Council. Guide for the care and use of laboratory animals. [7] USP 36-NF 31 vol. 1. Rockville, MD: U.S. Pharmacopeial Convention; 2013. p. 90–4. [8] Hasiwa N, et al. Evidence for the detection of non-endotoxin pyrogens by the whole blood monocyte activation test. ALTEX 2013;30:169–208. [9] Monocyte activation test (2.6.30). Pharmeuropa 2008;20(3):505–11. [10] Burger-Kentischer A, Abele IS, Finkelmeier D, Wiesmuller KH, Rupp S. A new cellbased innate immune receptor assay for the examination of receptor activity, ligand specificity, signaling pathways and the detection of pyrogens. J Immunol Methods 2010;358:93–103. [11] Akers MJ, Larrimore DS, Guazzo DA. Parenteral quality control: sterility, Pyrogen,Particulate and package integrity. third ed. New York: Marcel Dekker; 2002. [12] Guidance for industry, pyrogen and endotoxins testing: questions and answers. U.S. Department of Health and Human Services, FDA, CDER, CBER, CVM, CDRH, ORA; 2012. [13] Guideline on the replacement of rabbit pyrogen testing by an alternative test for plasma derived medicinal products. European Medicines Agency, CHMP; 2009. [14] Schindler S, von Aulock S, Daneshian M, Hartung T. Development, validation and applications of the monocyte activation test for pyrogens based on human whole blood. ALTEX 2009;26:265–77. [15] Moesby L, Hansen EW, Christensen JD. Ultrasonication of pyrogenic microorganisms improves the detection of pyrogens in the Mono Mac 6 assay. Eur J Pharm Sci 2000;11:51–7.
Authors contribution Clark Zervos and Vickie Barham: Experimental design, conducted rabbit pyrogen testing, fluorescence microscopy and flow cytometry. Thomas Zimmerman: Supervised chromatography and ultrafiltration fractionation studies, conducted enzymatic degradation studies, organic extractions and wrote draft of manuscript. Rebecca Silverstein: Conducted Protein A chromatography. Todd Willis, Pete Vandeberg and Melanie Williams: Conducted and directed in vitro pyrogen testing. Greg Flexman and Jyoti Srivastava: Traced “Donor X″ to a specific plasma donation center and to a particular donor. Doug Burns, Amy Durham and Jennifer Culp: Investigated manufacturing facility, procedures, records and supplies. David Malinzak: Experimental design, fluorescence microscopy. Conflicts of interest All authors except MW and DM are employees of Grifols Therapeutics LLC and have no other conflicts of interest to declare. MW and DM were employed by Grifols Therapeutics LLC at the time this research was performed and have no other conflicts of interest to declare.
19