A Risk-Based Approach to Immunogenicity Concerns of Therapeutic Protein Products, Part 2: Considering Host-Specific and Product-Specific Factors Impacting Immunogenicity
Immune responses to foreign proteins are expected and may be anticipated for some self-proteins. The rapidity of development and the strength and persistence of the response depends on many factors.
Because all protein therapeutics are potentially immunogenic, FDA has developed a risk assessment strategy for novel products in development and for major changes in manufacture and clinical use of licensed products. The first part of this series (BioPharm International, November 2004) discussed FDA's assessment of critical factors regarding the consequences of immune responses to protein therapeutics. These factors were based on product origin (such as xenobiotic and mammalian cell) and the presence and biological function of endogenous protein counterparts. In this second part, host-specific and product-specific factors that positively or negatively impact immunogenicity are discussed.
RISK ASSESMENT STRATEGY
Second Element: Consider Host-Specific Factors that Impact Immunogenicity Positively or Negatively.
Immunologic status and competence of the host. This is a critical factor in considering the risk of immune responses to therapeutic proteins. The incidence of hypersensitivity responses to specific biological drugs may be strongly affected by treatment of patients with steroids and concomitant treatment with chemotherapeutic agents.2 However, the incidence of hypersensitivity responses to drugs and environmental antigens does not appear to be diminished by intensive chemotherapy, despite the reduction of lymphocytes,30 although formal studies are lacking. However, patients who are undergoing chemotherapy or chemoradiotherapy are at much lower risk of mounting binding and neutralizing-antibody responses. This may be due to the severe depletion of CD4+ T cells which are replenished slowly and are required for development of T-dependent IgG antibody responses.31 For example, 95% of immune competent cancer patients generated neutralizing antibody to GM-CSF, while only 10% of immune-compromised cancer patients did so.32 Similarly, 4% of healthy volunteers mounted neutralizing antibody responses against a pegylated, truncated MGDF, while only 0.5% of cancer patients did so.10
In view of the relatively high incidence of food allergy directed to plant allergens, consideration should be given to the possibility of engendering hypersensitivity responses in patients treated with products derived from bioengineered plants bearing an allergen to which they are sensitive. Although this is currently a theoretical concern, the agency has anticipated this possibility due to the increasing exploration of plants as substrates for production of vaccines and biological therapeutics and has offered guidance.33
Route of administration. Subcutaneous (SC) administration is known to engender immune responses far more ably than either the intramuscular (IM) or intravenous (IV) routes. This may be due to the rapid egress of the product into tissues containing a high frequency of potent antigen presenting cells (APCs); the potential for product to form or stay in aggregates in the SC space compared to dispersal of aggregates in the high-flow, fluid-rich IV environment; and to the possible depot effect of SC injection. The increased incidence of pure red cell aplasia (PRCA) mediated by neutralizing antibodies to Eprex is a good case in point; it appears that the SC route is necessary, though not sufficient, to generate neutralizing antibody.
Dose and frequency of administration. The effects of dose and frequency are not independent of other factors such as route of administration, product origin, and product-related factors that influence immunogenicity. For example, one would expect that for foreign proteins, frequent administration would enhance immunogenicity, especially when given by a more immunogenic route. In contrast, frequent administration of high doses of protein intravenously induces tolerance to Factor VIII in those patients who are least tolerant. Product-related factors, such as the presence of adjuvant and the type and level of product aggregates, may be crucial in determining whether a given dosing regimen proves immunogenic.
Level of immune tolerance to endogenous protein. The completeness of immunologic tolerance to an endogenous protein determines how easily tolerance to it can be broken by administration of a therapeutic counterpart. Thus, tolerance in both the T- and B-cell partners required for IgG-antibody production depends on numerous factors, especially the abundance of the protein.34-36 In this regard, T cells are tolerized at lower levels of soluble self-proteins than B cells,34 due to multiple mechanisms for negative selection of antigen-specific T cells in the thymus, including promiscuous expression of tissue-specific antigens.37
B cells have been shown to be anergized (physically present, but functionally inactivated) when a model soluble protein is present at 10-9 M but not when the protein is present at 10-10 M in a transgenic model of B-cell tolerance.36 These serum levels correlate with the extent of B-cell receptor occupancy by the protein, with some degree of anergy observed at 5% and full anergy at ~45% occupancy of the available surface Ig receptors. Below the 5% threshold, the B-cell repertoire differs little from control mice. Additional factors that impact B-cell anergy and tolerance include the affinity of antibody for protein; the stage of B-cell development at which the B cell is exposed to antigen; antigen format, with cell-surface-presented antigen (arrayed antigen) being highly efficient at deletion of immature B cells; and concomitant signals (for example, costimulation and adjuvant activity).36
How do these experimental data pertain to immune responses to therapeutic proteins? The finding that the immune system is less tolerant of low-abundance proteins, such as cytokines and growth factors, compared to high-abundance proteins is consistent with many lines of evidence: the presence of natural autoantibodies to cytokines and growth factors in healthy individuals;38-41 the development of antibody responses to cytokines during normal immune responses,8 and, finally, the ease with which immune responses can be generated to very-low-abundance endogenous factors by exogenous recombinant therapeutic products. For example, with thrombopoietin (TPO), present at picomolar levels, immune responses were relatively easily generated, in some cases requiring only two doses,10 whereas for albumin, the highest abundance protein, immune responses are difficult to detect.
Third Element: Consider Product-Specific Factors that Impact Positively or Negatively on Immunogenicity.
Product origin (foreign or endogenous). Immune responses to foreign proteins are expected and, as discussed above, may be anticipated for some self-proteins. However, the rapidity of development and the strength (titer) and persistence of the response depends on many factors, including previous and ongoing environmental exposure (streptokinase) and the mode of exposure (for example, an antigen introduced through the oral route may tolerize rather than immunize). Moreover, the presence of immunity-provoking factors in the product, such as product aggregates and materials with adjuvant activity, as well as the product's inherent immunomodulatory activity, are critical parameters in the generation of responses to both foreign and native protein.
Alteration in molecular structure. Neodeterminants, such as those created from the fusion of a therapeutic protein with a partner antigen, may generate immune responses. Responses to the neodeterminant may spread to conserved segments of the molecule. As reported, this pathway may be responsible for the neutralization of the GM-CSF-IL-3 fusion molecule (PIXY 321) in which a specific neutralizing-antibody response to the fusion protein predominates over neutralizing responses to either the IL-3 or GM-CSF partners;42 mapping studies, however, were not performed to confirm that the specificity of the response was to neodeterminants resulting from the fusion. For fusion molecules in which both partners are self proteins, studies to define the antigenic site of antibody responses are highly recommended. Fusion proteins involving a foreign protein and self protein are of particular concern because of the capacity of the foreign protein to act as a "helper" determinant, efficiently eliciting T-cell help to generate a response to the self-protein partner.
Glycosylation. Glycosylation may strongly modulate immunogenicity of the proteins to which they are attached, and the effects may be mediated via a number of possible pathways: immune responses may be generated directly to carbohydrate moieties consisting of unique mammalian xenogeneic sugars,43 yeast mannans,44 or plant sugars.45-46 Such foreign sugars may trigger innate immune responses through interaction with Toll-like receptors (such as TLR-2 and TLR-4 by yeast cell wall glycans or LPS,47 respectively) or through the abundant lectin-like molecules that pervade the mammalian proteome.48 Whether plant sugars from transgenic plant-produced proteins will prove immunogenic in humans is of great interest since oral administration of antigen is well known to induce tolerance, and consumption of plants is universal.
Glycosylation may also indirectly alter the immunogenicity of the protein by minimizing protein aggregation, as appears to be the case with glycosylated IFN-β products.49 And, finally, glycosylation may "shield" immunogenic protein epitopes from the immune system, which — though beneficial for therapeutic proteins — poses a particular problem for successful vaccine development.50-52
Antigenic organization (product aggregates). Protein aggregates have been appreciated as inducing immune responses to the monomeric form of the protein for over a half century.53-54 Speculation regarding how protein aggregates facilitate immune responses centers on their ability to cross-link sufficient numbers of B-cell receptors in a multivalent fashion, causing efficient B-cell activation7,55,56 and enhancing antigen processing and presentation, thereby efficiently recruiting the T-cell help critical for generating high-affinity IgG antibody.57 Since the biological activity of therapeutic proteins depends on protein conformation, the ability of protein aggregates to generate neutralizing antibody may depend on preservation of the native conformation of the molecule within product aggregates. Therefore, aggregates of denatured protein may be less potent in generating neutralizing antibody than in generating binding antibody. However, antibodies to linear determinants in the protein contact sites could theoretically neutralize the protein's bioactivity as well. They also could be generated de novo or result from epitope spreading.
Critical information is lacking regarding the types and quantities of aggregates sufficient to generate immune responses for specific protein therapeutics. This may differ dramatically, depending on product origin and the underlying level of tolerance. Product aggregates have been key to the development of immune responses to many therapeutic proteins including: type I IFNs,49,58 rHu IL-2 (which is formulated as an aggregate of an average size of 27 molecules59), and human growth hormone.60 It is obviously critical for manufacturers of therapeutic protein products to minimize aggregation and to use sensitive assays to detect the various types of aggregates that may arise.
Measurement of aggregates. It has recently been recognized that no single method of assessing aggregates provides a robust measure of protein aggregation.61 Indeed, exclusive use of size exclusion chromatography may preclude detection of higher molecular weight aggregates that fail to penetrate the gel. Therefore, consideration should be given to implementing several orthogonal methods, such as analytical ultracentrifugation,62 light scattering techniques,63 Fourier Transform Infrared spectroscopy (FTIR),64 and field flow fractionation.65,66
Formulation. Formulation components are chosen for their ability to preserve the native conformation of the proteins in order to maximize in vivo activity (by preventing protein denaturation due to hydrophobic interactions among protein molecules and between protein and hydrophobic surfaces such as glass and air) as well as to prevent protein degradation due to, for example, oxidation and deamidation.67-69 Large proteins included as excipients in the formulation, such as human serum albumin (HSA), may impact immunogenicity positively or negatively. Although they are added for their ability to inhibit hydrophobic interactions, they may impact negatively by co-aggregating with product or by forming protein adducts. Indeed IFN-α-HSA aggregates have been demonstrated to engender immune responses to IFN-α.57 Formulation may engender adjuvant activity through micelle formation, which can form in relatively high concentrations of nonionic detergents such as Polysorbate 80.67-70 Non-ionic detergents are also known to enhance leaching of organic molecules and metal ions, which have potential adjuvant activity.67-69
Adjuvants. Adjuvant activity can arise from multiple sources other than formulation. Indeed, adjuvants may arise from the presence of microbial host-cell proteins, oligonucleotides, or polysaccharides, which may exert direct adjuvant activity through Toll-like receptors or other pathogen-associated molecular pattern (PAMP) recognition molecules present on B cells and other APC populations.47 Adjuvant activity may also arise as a direct result of the biological activity of the product. Therefore, type I IFNs,58 IL-2,28,30 and GM-CSF 71 are known to upregulate (through differing mechanisms) immune responses to themselves as well as to other biological therapeutics, endogenous proteins, and even small drug molecules,30 while other products, such as anti-CD4 MAbs 72 and CTLA4-Ig,73 downregulate such responses.
Container-closure considerations. FDA has come to increasingly appreciate the fact that interactions between protein therapeutics and container closures may negatively impact product quality and immunogenicity. This is particularly apparent with the increasing use of prefilled syringes for patient self-administration.
Syringes are delivery devices with multiple surfaces and materials that can interact with the product and alter product quality. Glass and air interfaces are hydrophobic surfaces that can mediate protein denaturation and aggregation. Moreover, glass syringes and movable surfaces (such as syringe plungers) are coated with silicone oil, which facilitates injection but also provides a surface on which proteins can denature and aggregate. Rubber surfaces contain organic leachables involved in vulcanization or plasticization that can leach into the product, particularly when the surfaces are not coated. Finally, metal wire is used to create the syringe hubs in glass syringes, and residual levels of such metals also may interact with the product.
For example, one protein therapeutic was found to have increased levels of truncated product due to the leaching of metal ions from the syringe gasket, which activated a metalloprotease coeluting with the product. In another example, tungsten metal ions from the syringe barrel oxidized a therapeutic protein, causing product aggregation. Finally, organic leachables extracted from uncoated rubber stoppers by the polysorbate excipient may act as immune adjuvants, inducing neutralizing antibody to the product.
Product custody. The intent of packaging therapeutic products in prefilled syringes is to facilitate patient self-administration. The patient must store the product under appropriate conditions and administer it correctly. Given that most protein products are sensitive to heat, light, and mechanical agitation — which cause product denaturation and aggregation — it's worth considering whether, in the case of Eprex, self-administration with the attendant possibilities of storage or handling abuse, contributed to its immunogenicity. Did patients developing PRCA have excursions from normative storage or handling conditions? Clearly, the appropriate education of patients regarding product handling and administration is vital to ensure product quality.
REFERENCES (PART 2)
Note:
References 1-29 can be found in Part 1 of this article. See the November 2004 issue of
BioPharm International
.
30. Heywood GR, Rosenberg SA, Weber JS. Hypersensitivity reactions to chemo-immunotherapy with high dose IL-2. J. Natl Cancer Inst. 1995; 87:915-922; and personal communication of Alexandra S. Worobec, M.D. and Crystal Mackall, M.D.
31. Mackall CL. T-cell immunodeficiency following cytotoxic antineoplastic therapy: a review. The Oncologist 1999; 4:370-378.
32. Ragnhammar P, Friesen H, Frodin J, Lefvert A, Hassan M, Osterborg A, et al. Induction of anti-recombinant human GM-CSF (E. Coli derived) antibodies and clinical effects in nonimmunocompromised patients. Blood 1994; 84:4078-87.
33. FDA. Guidance for Industry: Drugs, Biologics, Medical Devices Derived from Bioengineered Plants for Use in Humans and Animals. Sept. 2002. Available at www.fda.gov.cber/guidelines.html.
34. Weigle WO. Analysis of autoimmunity through experimental models of thyroiditis and allergic encephalomyelitis. Advances in Immunology 1980; 30:159-273.
35. Haribhai D, Engle D, Meyer, M, Donermeyer D, White JM, Williams C. A threshold for central T-cell tolerance to an inducible serum protein. J. Immunol. 2003; 170:3007-3014.
36. Goodnow CC. Transgenic mice and analysis of B cell tolerance. Ann. Rev. Immunol. 1992; 10:489-518.
37. Derbinski J, Kyewski B. Self representation in the thymus: an extended view. Nature Reviews Immunology 2004; 4:688-698.
38. Revoltella R. Natural and therapeutically induced antibodies to cytokines. Biotherapy 1998; 10:321-31.
39. Soos J, Polsky R, Keegan S, Bugelski P, Herzyk D. Identification of natural antibodies to interleukin 18 in the sera of normal humans and three nonhuman primate species. Clinical Immunology 2003; 109:188-96.
40. Turano, A., Balsari A, Viani E, Landolfo S, Zanoni L, Gargiulo F, et al. Natural human antibodies to g interferon interfere with the immunomodulating activity of the lymphokine. PNAS1992; 89:4447-51.
41. Tiberio L, Caruso A, Pozzi A, Rivoltini L, Morelli D, Monti E, et al. The detection and biological activity of human antibodies to IL-2 in normal donors. Scand. J. Immunol. 1993; 38:472-76.
42. Miller L, et al. Abrogation of the hematological and biological activities of the interleukin 3/granulocyte macrophage colony stimulating factor fusion protein PIXY321 by neutralizing anti-PIXY antibodies in cancer patients receiving high dose carboplatin. Blood 1999; 93:3250-58.
43. Cramer DV. Natural antibodies and the host immune responses to xenografts. Xenotransplantation 2000; 7:83-92.
44. Bretthauer R, Castellino F. Glycosylation of pichia pastoris derived proteins. Biotechnol. Appl. Biochem. 1999; 30:193-200.
45. Gomord V, Faye L. Posttranslational modification of therapeutic proteins in plants. Current Opinion in Plant Biology 2004; 7:171-181.
46. Andersson K, Lidholm J. Characteristics and immunobiology of grass pollen allergens. Int. Arch. Allergy Immunol. 2003; 130:87-107.
47. Roeder A, Kirschning C, Rupec R, Schaller M, Korting H-C. Toll-like receptors and innate antifungal responses. Trends in Microbiology 2004; 12:44-49.
48. Holmskov U, Thiel S, Jensenius J. Collectins and Ficolins: humoral lectins of the innate immune defense. Ann. Rev. Immunol. 2003; 21: 547-78.
49. Runkel L, Meier W, Pepinsky RB, Karpusas M, Witty A, Kimball K, et al. Structural and functional differences between glycosylated and non-glycosylated forms of human Interferon-β. Pharmaceutical Research 1998; 15:641-9.
50. Gribben J, Devereux S, Thomas N, Keim M, Jones H, Goldstone A, et al. Development of antibodies to unprotected glycosylation sites on recombinant human GM-CSF. Lancet 1990; 335:434-437.
51. Schauer R. Victor Ginsburg's influence on my research of the role of sialic acids in biological recognition. Archives of Biochemistry and Biophysics 2004; 426:132-141.
52. Cole K, Steckbeck J, Rowles J, Derosiers R, Montelaro R. Removal of N-linked glycosylation sites in the V1 region of Simian Immunodeficiency Virus gp120 results in redirection of B cell responses to V3. J. Virol. 2004; 78:1525-39.
53. Dresser DW. Specific inhibition of antibody production. Immunology 1962; 5:378.
54. Gamble CN. The role of soluble aggregates in the primary immune response of mice to human gamma globulin. Int. Arch. Allergy 1966; 30:446-55.
55. Dintzis R, Okajima M, Middleton M, Greene G, Dintzis H. The immunogenicity of soluble haptenated polymers is determined by molecular mass and hapten valence. J. Immunol. 1989; 143:1239-44.
56. Bachmann M Rohrer U Kundig T Burki K Hengartner H, Zinkernagel RM. The influence of antigen organization on B cell responsiveness. Science 1993. 262:1448-51.
57. Cheng P, Steele C, Gu L, Song W, Pierce SK. MHC class II antigen processing in B cells: accelerated intracellular targeting of antigens. J. Immunol. 1999; 162:7171-80.
58. Braun A, Kwee L, Labow M, Alsenz J. Protein aggregates seem to play a key role among the parameters influencing the antigenicity of interferon alpha (IFN-α) in normal and transgenic mice. Pharmaceutical Research 1997; 10:1472.
59. Physicians Desk Reference. 2004. Proleukin. 1163-1167.
60. Moore W, Leppert P. Role of aggregated human growth hormone (hGH) in development of antibodies to hGH. Journal of Clinical Endocrinology and Metabolism 1980; 51:691-97.
61. Alliance Protein Website: www.ap-lab.com/characterization_methods.htm.
62. Liu J, Shire SJ. Analytical ultracentrifugation in the pharmaceutical industry. Journal of Pharmaceutical Sciences 1999; 88:1237-1241.
63. Wyatt Technology Corporation Website: www.wyatt.com/theory/index.cfm
64. Koppaka V, Murray P, Axelsen P. Early synergy between Abeta 42 and oxidatively damaged membranes in promoting amyloid fibril formation by Abeta40. J. Biol. Chem. 2003; 278: 36277-84.
65. Levin S. Field flow fractionation in biomedical analysis. Biomedical Chromatography 1991; 5:133-137.
66. Fraunhofer W, Winter G, Coester C. Asymetrical flow field flow fractionation and multiangle light scattering for analysis of gelatin nanoparticle drug carrier system. Anal. Chem. 2004; 76:1909-20.
67. Cleland J, Powell M, Shire SJ. The development of stable protein formulations: a close look at protein aggregation, deamidation, and oxidation. Critical Reviews in Therapeutic Drug Carrier Systems 1993; 10:307-77.
68. Wang W. Instability, stabilization, and formulation of liquid protein pharmaceuticals. International Journal of Pharmaceutics 1999. 185:129-88.
69. Shire SJ, Shahrokh Z, Liu J. Challenges in the development of high protein concentration formulations. J. Pharm. Sci. 2004; 93:1390-402.
70. Hermeling S, Schellekens H, Crommelin DJ, Jiskoot W. 2003. Micelle associated protein in epoetin formulations: a risk factor for immunogenicity? Pharm. Res. 2003; 20:1903-07.
71. Mellstedt H, Fagerberg J, Frodin J, Henriksson L, Hjelm-Skoog AL, Liljefors M, et al. Augmentation of the immune response with granulocyte-macrophage colony stimulating factor and other hematopoietic growth factors. Curr. Opin. Hematol. 1999; 6:169-75.
72. Knox S, Hoppe R, Maloney D, Gibbs I, Fowler S, Marquez C, et al. Treatment of cutaneous T-cell lymphoma with chimeric anti-CD4 monoclonal antibody. Blood 1996; 87:893-99.
73. Abrams JR, et al. CTLA4Ig-mediated blockade of T cell costimulation in patients with psoriasis vulgaris. J. Clin. Invest. 1999; 103:1243-52.
ACKNOWLEDGMENTS
The authors thank Drs. Karen Weiss, Keith Webber, Kathleen Clouse, Jay Lozier, and Andrew Chang for critical reading of the manuscript.
Amy S. Rosenberg, M.D., is director of the Division of Therapeutic Proteins, CDER, FDA, Bldg 29A, Room 2D-16, 8800 Rockville Pike, Bethesda, MD 20892, fax 301.480.3256, amy.rosenberg@fda.gov.
Alexandra S. Worobec, M.D., is a Medical Officer in the Division of Therapeutic Biological Oncology Products, CDER, FDA. Bldg. 10, Room 11C-208, 8800 Rockville Pike, Bethesda, MD 20892, alexandra.worobec@fda.gov.
