Although vaccines and immunotherapies are designed to engage the human immune system in fighting disease, unwanted immunogenicity can be a major problem for protein-based therapeutics. Some patients produce antidrug antibodies (ADAs), which might lead to drug inactivation or adverse effects. Even human and humanized proteins have proven to be surprisingly immunogenic in some cases, suggesting that immune tolerance requires careful consideration in biologic product design. In rushing to deliver new drugs to market, some biotherapeutics developers have overlooked factors that contribute to protein immunogenicity. Fortunately, the parameters influencing vaccine efficacy have been thoroughly studied for years, which allows biopharmaceutical companies to draw parallels when addressing immunogenicity of their protein therapeutics. Individual variations in patient tendencies to develop ADAs are probably genetic, so this is another area where personalized medicine and pharmacogenomics may help the industry progress.

The immune response to biologic products often involves B or T cells. The former produce antibodies that bind to proteins and thus reduce or eliminate their therapeutic effects. Potential complications can be life-threatening. Thus, measuring the tendency to trigger antibody formation is an important part of determining the clinical safety and efficacy of protein-based drugs. T cells help activate B cells, especially for disease cases in which a patient’s natural protein is defective in some way. That patient’s T cells could treat protein therapeutics as if they were foreign invaders because they are different from the native protein. Such a response has been noted, for example, in some hemophilia patients, whose blood factor VIII is genetically defective. They may develop ADAs when infused with a correct factor VIII therapeutic protein (e.g., Bayer’s Kogenate or Baxter’s Advate products), presenting a significant impediment to such treatment. It is, in fact, considered to be “the most important problem in hemophilia A care today” (1). Rheumatoid arthritis is another condition for which treatments are complicated by immunogenicity (2).

Regulators Respond: Regulations often come about in response to specific events that highlight the need for government oversight of a specific business activity. Late in the 20th century, notable occurrences of pure red-cell aplasia (PRCA) associated with erythropoietin treatments (3) led to closer scrutiny by regulators in several countries, and the “International Guidelines” box lists some of the results. Meanwhile, biotherapeutics developers began looking closer at the subject, too (4). Immunogenicity is now considered to be a basic aspect of biologic product safety. It is generally accepted that repeated administration of protein therapeutics can lead to ADA production in patients. So there is no dearth of guidance available to biopharmaceutical companies embarking on immunogenicity assessment of their products in development (5).

For example, guideline S6 from the International Conference on Harmonisation of Technical Requirements for the Registration of Pharmaceuticals for Human Use (ICH) describes approaches to preclinical testing, including that for immunogenicity (6). The guideline was adopted by the European Medicines Agency’s (EMA’s) Committee for Medicinal Products for Human Use (CHMP) in 2011 and by both the US Food and Drug Administration (FDA) and the Japanese Ministry of Health, Labour, and Welfare (MHLW) in 2012. Note that immunogenicity is not to be confused with immunotoxicity (the unwanted side effect of immune-system suppression), for which ICH has another specific guidance (ICH S8).

International Guidelines European Medicines Agency EMEA/CHMP/BMWP/14327/2006: Guideline on Immunogenicity Assessment of Biotechnology-Derived Therapeutic Proteins. European Medicines Agency: London, UK, 13 December 2007; www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2009/09/WC500003946.pdf.EMA/CHMP/BMWP/86289/2010: Guideline on Immunogenicity Assessment of Monoclonal Antibodies Intended for In Vivo Clinical Use. European Medicines Agency: London, UK, 24 May 2012; www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2012/06/WC500128688.pdf.EMA/275542/2013: Concept Paper on the Revision of the Guideline on Immunogenicity Assessment of Biotechnology- Derived Therapeutic Proteins (CHMP/BMWP/42832/2005) (Draft). European Medicines Agency: London, UK, 20 February 2014; www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2014/03/WC500163623.pdf. US Food and Drug Administration CBER/CDER. Draft Guidance for Industry: Assay Development for Immunogenicity Testing of Therapeutic Proteins. US Food and Drug Administration: Rockville, MD, December 2009; www.fda. gov/downloads/Drugs/…/Guidances/UCM192750.pdf. CDER. Draft Guidance for Industry: Immunogenicity-Related Considerations for the Approval of Low Molecular Weight Heparin for NDAs and ANDAs. US Food and Drug Administration: Rockville, MD, April 2014; www.fda.gov/downloads/Drugs/ GuidanceComplianceRegulatoryInformation/Guidances/ UCM392194.pdf. CBER/CDER. Guidance for Industry: Immunogenicity Assessment for Therapeutic Protein Products. US Food and Drug Administration: Rockville, MD, August 2014; www.fda.gov/ downloads/Drugs/GuidanceComplianceRegulatory Information/Guidances/UCM338856.pdf.

Cause and Effect
Many factors contribute to the immunogenicity of recombinant biologics, some related to the route and frequency of administration or a patient’s own circumstances, but most directly involving the product itself (7). Nonhuman sequences of amino acids in a protein’s make-up are well known to elicit an immune response; that’s the main reason the industry pushed for “humanizing” and developing “fully human” antibodies some two decades ago. Contaminants and adventitious agents are likely to do the same. Immunogenic glycosylation patterns have limited the use of plant- and yeast-based expression systems. Some stabilizing agents and storage media can make products more immunogenic (even acting like vaccine adjuvants), especially in combination with out-of-specification temperatures or rough handling. Protein oxidation and aggregation also are likely to cause trouble; the former can lead to the latter, which also can be caused by other means. Often no single underlying factor can be pointed to, but rather several factors will interact to cause a problem.

In addition to the erythropoeitin events referred to above, ADA issues have arisen over time with blood factors VIII and IX, interferons α and β, interleukin, hormones, and monoclonal antibodies (MAbs) — with clinical consequences ranging from none at all (for growth hormone) to loss of product efficacy (the most common result of immunogenicity), allergic anaphylaxis, and PRCA as mentioned above. In the latter case, patients’ immune systems went on to attack their own native protein as well as the introduced recombinant product. Regulatory guidance for immunogenicity continues to be revised, and class-specific guidance is emerging as research brings more cause-and-effect understanding to light.

Protein aggregates are often immunogenic. However, “in the absence of adjuvants, proteins usually are not immunogenic, in fact, often they are tolerogenic” (1). That is the basis of allergy immunotherapy, for example, through which controlled exposure to small amounts of a given allergen builds up a patient’s tolerance to it. “Tolerogenicity” is a never-ending source of consternation for vaccine developers — but something that biotherapeutics companies love to see. So immunogenicity of therapeutic protein products — especially fully human and humanized products — often can be attributed to copurified impurities (e.g., from animal-sourced products used in biomanufacturing), aggregation and conformation changes that attract T-cell attention, and even simple product storage and handling. One common cause of unwanted immunogenicity is inclusion of polysorbate (Tween) surfactants in biotherapeutic formulations (8). All these are the types of problems that manufacturing changes can solve.

“Little is known about the mechanism of aggregate induction of immunogenicity,” says Ed Maggio (CEO of Aegis Therapeutics). The most immunogenic type seem to be large, multimeric protein aggregates made up of native- like molecules with repetitive epitopes on their surfaces — leading to an assumption that more available binding sites increase risk (9). Aggregates of denatured protein also can induce antibody production, but the results shouldn’t be product-neutralizing antibodies because protein conformation is lost through denaturation. That is, ADAs that recognize unfolded linear sequences aren’t likely to latch onto properly folded versions of those same sequences; it happens, but it’s rare.

So the occurrence of aggregation itself doesn’t guarantee trouble. ADAs against aggregates may not bind to single monomers and, even if they do, they don’t always cause problems. Different immunological mechanisms have been hypothesized based on the different types of immunogenic threat, some involving T cells and some not. The biopharmaceutical industry is actively seeking reliable predictive models to address immunogenicity of protein aggregates, but doing so requires improved understanding of how it works.

When it comes to formulation-based immunogenicity, Maggio says that polysorbates 80 and 20 are the main offenders. To complicate matters, these nonionic surfactants are included in most biotherapeutic drug products to prevent aggregation and extend shelf life. They are mixtures, themselves (of fatty-acid esters) that inevitably include spontaneously generated oxidative contaminants (e.g., protein-damaging peroxides, epoxy acids, and reactive aldehydes that react with methionines, histidines, lysines, cysteines, tryptophans, tyrosines, and primary amino groups). Those impurities can form immunogenic “neoantigens.” Lot variability in the extent of polysorbate degradation can cause subsequent lot variability in biologics immunogenicity. At the very least, that can complicate bioequivalence comparisons required for regulatory approval and postmarket surveillance — or lead to lot failures, recalls, and even necessitate reformulation and requalification that can involve expensive human trials. Maggio points to studies that have found a broad range of hydroperoxide content in commercially available polysorbate products (10). The level of reactive contaminants in those preparations varies over time because auto-oxidation of polysorbates is spontaneous and progressive.

In response to such findings, suppliers have made available specifically “high-purity” polysorbate preparations. These are treated to remove peroxides and packaged with oxygen excluded from their containers (in the headspace, it is replaced with nitrogen or argon). But oxidation begins as soon as the contents come into contact with the air, Maggio points out, and such reactions will accelerate in aqueous solution (11). Again, this may or may not cause a problem — polysorbates may help or harm — so risk assessment must be based on real data from laboratory analyses.

Evaluating Immunogenicity
Because therapeutic proteins have different properties and aggregation profiles, each poses a unique immunological risk. So experts recommend a case-by-case approach to assessing such risk. That posed by the presence of aggregates, for example, comes from not only their tendency to trigger ADAs, but also the clinical consequences that those antibodies might have (9).

“The amount, size, and type of aggregates necessary to trigger immune responses are a major concern for pharmaceutical companies and regulatory agencies. Current US Pharmacopeia (USP) particulate requirements state that particles >10 μm in size should be controlled below 6,000 particles per container, but no regulations have been established for the smaller sizes. Therefore, it is possible that immunogenically relevant protein aggregates have been routinely ignored by these regulations. For every aggregate >10 μm present in a formulation, there can be considerable amounts of slightly smaller aggregates, and each may contain hundreds or thousands of protein units. Thus, given the lack of knowledge regarding the most immunogenic aggregate sizes, it is important to develop instruments, protocols, and regulations that properly address a broader range of particle sizes, especially in the subvisible range.” (9)

Circulating ADAs are the primary measurement used in defining an immune response to recombinant protein products. As mentioned above, both patient-related and product-related factors come into play, and those provide a starting point for immunogenicity risk assessment. Ideally, all potential factors should be taken into consideration early in drug development.

Information about patient immune responses to protein therapeutics is required for marketing applications. And limited data may be available at the time of product launch, so postmarket surveillance is often necessary. Methods for evaluating immunogenicity include in silico screening and protein structure analysis, analytical characterization, preclinical testing and bioanalytical assays, and ultimately clinical trials. During early development, preclinical data can help developers design later clinical studies and interpret their results.

ICH S6 suggests measuring ADAs in nonclinical studies when there is evidence of altered pharmacodynamic activity or immune-mediated reactions in animal models or human subjects (6). “Samples should be preemptively collected for potential analysis.” And when ADAs are detected, companies should assess their effects on study interpretation. The guideline says that determining the potential for neutralizing antibodies (NAbs) is warranted when ADAs are detected but there is no identified PD marker to demonstrate their sustained activity.

The  to demonstrate their sustained activity. The EMA’s 2007 guideline cautions that “the predictive value of nonclinical studies for evaluation of immunogenicity of a biological medicinal product in humans is low due to inevitable immunogenicity of human proteins in animals” (13). Although such experiments normally are not required, they can “be of value in evaluating the consequences of an immune response.” Companies must develop “adequate screening and confirmatory assays” to measure immune responses against their products. Such assays should distinguish neutralizing from nonneutralizing antibodies, both in pivotal clinical trials and postmarket studies. Data should be systematically collected from “a sufficient number of patients.” A sampling schedule is determined for each product by considering risks associated with an unwanted immune response in patients. Thus, immunogenicity issues should be addressed in a product’s overall risk management plan.

The classic immunological assay is a sandwich type of enzyme-linked immunosorbent assay (ELISA). Other approaches involved in immunogenicity testing include aggregation and particle analysis (14). Experience in applying immunochemical analysis to clinical diagnostics and disease research has led supplier companies such as Cygnus Technologies into the immunogenicity field. That company offers reagents and test kits developed to overcome many nonspecific and false-positive reactions that can confuse analysis of patient immune responses. According to the Cygnus website, well-characterized, immunospecific reagents can be used to establish baseline (pretherapy) immune response; detect titer increases after therapy; screen patient populations to identify those who may have neutralizing antibodies; correlate antibody titers to therapies; map immunogenic epitopes on protein structures; distinguish between recombinant (drug) and the natural protein forms and between immune response to a drug or associated process contaminants; and establish isotypes and subtypes of antibodies involved.

Other providers of immunogenicity assay reagents, kits, and other technologies include Antitope (an Abzena company), Meso Scale Discovery, Pall ForteBio, Perkin Elmer, and ProZyme. But like viral safety, immunogenicity testing is an area of such specialized knowledge that many drug-sponsor companies prefer to outsource the work to contract testing service providers such as those listed in the “Immunogenicity Testing Services” box. And some contract manufacturing and development organizations — e.g., Covance, Fujifilm Diosynth Biotechnologies, and Lonza — make a point of touting their own expertise in this area. Immunogenicity analysts use instruments such as the Biacore system from GE Healthcare, high-performance liquid chromatography (HPLC) systems from Agilent Technologies, Cirascan microarrayers from Aushon Biosystems, the Optim 2 fluorescence and static light- scattering system from Avacta Analytical, Bruker’s AVANCE-II nuclear magnetic resonance (NMR) spectrometer, Gyrolab automation from Gyros, the Sysmex FPIA3000 particle-sizing system and MicroCal microcalorimetry from Malvern Instruments, and dual- layer multiplexing technology from SQI Diagnostics.

Some Immunogenicity Testing Services Accuro Biologics (www.accurobio.com) in Scotland Advanced Biodesign (www.a-biodesign.com) in France Anabiotec (www.anabiotec.com) in Belgium BioAgilytix (www.bioagilytix.com) in North Carolina Charles River Laboratories (www.criver.com) in many locations across Europe and the United States Eurofins Bioanalytical Services (www.eurofins.com) in several locations around the world Haemtech Biopharma Services (www.haemtechbiopharma. com) in Vermont ImmunXperts (www.immunxperts.com) in Belgium Intertek (www.intertek.com) in many locations worldwide Pacific BioLabs (www.pacificbiolabs.com) in California ProImmune (www.proimmune.com) in Florida and in Oxford, UK Quintiles (www.quintiles.com) in multiple locales worldwide SGS (www.sgs.com) in many locations around the world TNO Triskelion (www.tnotriskelion.com) in The Netherlands

The Tiered Strategy: Many experts endorse a tiered approach to immunogenicity determination (15). For example, Jo Goodman (senior R&D manager at MedImmune’s Cambridge, UK, site) said at a 2013 conference that “It is essential to adopt an appropriate strategy for development of adequate screening and confirmatory assays to measure an immune response against a therapeutic protein.” Assays, she said, must be sensitive and detect responses at clinically relevant concentrations. Developers need to consider possible interference from “matrix factors” in patient sera, for example. Immunogenicity assays should be able to detect different types of responses (e.g., IgG or IgM) and be validated and standardized to distinguish neutralizing from nonneutralizing antibodies. An assay should be designed for the clinical population to which it will be applied and ideally capable of detecting low-affinity antibodies. All these criteria help prevent false-negative results. Use of relevant positive and negative controls will help.

The tiered approach begins with patient samples taken and screened at specified times during clinical trials (15). Positive results go through confirmatory testing, and negative results are rejected (which explains the importance of preventing false negatives). Samples that are confirmed positive go through further analysis for characterization and titer, as well as neutralization assay development. Using assays for clinical markers and assessment of clinical response in patients, analysts then assess the correlation of characterized ADAs with clinical responses.

Goodman describes immunogenicity assay technology as “a changing landscape.” With the issue first brought to light just over a decade ago, assay technology and strategies for determining immune responses have evolved over time. “Early assay types mainly involved ELISA, surface plasmon resonance (SPR), fluorescence, and radioimmunoprecipitation (RIP) techniques. And those technologies are still viable approaches. But new technology has become available, and each platform has its own advantages and disadvantages.” Most new approaches are variations on older themes rather than wholly novel analytical technologies.

Starting before clinical testing begins, service providers such as Lonza stress immunogenicity assessment by addressing T-cell epitope mapping throughout drug development. By managing potential drug immunogenicity at the earliest possible stage, Lonza says that companies can save time and money by creating the safest possible protein products.

A number of issues can arise in immunogenicity assay development and execution:

• cell-related problems in cell-based assays (e.g., cell-line stability, passaging, media/sera changes, mycoplasma or other contamination, cell banking);
• reagent trouble (e.g., stability and activity, changes in positive controls or detection antibodies, microplate and detection kit variability, and biological activity of specialized reagents);
• instrumental issues (e.g., maintenance/calibration, pipette changes, varying parameters, introduction of new brands or models of instrument);
• problems related to assay format and execution (e.g., different analysts, method drift over time, potential differences among patient samples).

The tiered approach may not work for biosimilar product development (16). Small differences in production processes can lead to protein conformational or folding changes — potentially causing aggregation and subsequent immunogenicity. Screening and confirmation assays in the tiered approach tend to measure ligand binding. Neutralizing assays are cell based and may involve proliferation or gene reporters or measure potency (e.g., antibody-dependent, cell-mediated cytotoxicity). Biosimilar developers have different questions to answer: Is my biosimilar as immunogenic as the innovator? Might it even be “biosuperior?” What about interchangebility?

Immunogenicity assays for biosimilar products may involve two different positive control antibodies: one against the innovator product and one against its biosimilar counterpart. Some developers wonder whether they can simply use an innovator’s assay to detect ADAs against a biosimilar. However, that single assay can reveal only relative immunogenicity rates between biosimilar and innovator products. “A second assay may be necessary to reveal true differences” (16). In preclinical testing, immunogenicity rates may differ because of the small number of animals used. Even if a biosimilar product appears to have lower immunogenicity than its original counterpart, the developer will need to determine the biological relevance of such results in combination with other data (e.g., pharmacokinetics, pharmacodynamics). All assays used should have comparable sensitivity, selectivity, and precision to those used by the innovator. Here, as everywhere in biopharmaceutical development, a knowledge-based risk-management approach is advised:

“The more we learn regarding the factors triggering immunogenicity, and how they affect and activate the immune system, the better we can implement assays to predict and assess immunogenicity at a preclinical stage, with improved clinical translational value.”(16)

Predict and Prevent
Before a product ever reaches clinical-phase studies and associated bioanalysis, immunogenicity can and should be considered. Even before preclinical testing, some companies are getting proactive. Some even believe they may be able to ward off trouble at the protein-design stage. Doing so could turn potential “biosimilars,” into “biobetters” — or give innovators the kind of patent edge they’re looking for. Although as yet no reliable, straightforward models for widespread prediction have been put forth, a number of approaches are in different stages of development. They may be in vitro (cell-based analysis), in vivo (animal testing), or in silico (computer modeling) in nature.

Planning and Predicting: In silico profiling is gaining interest for a number of reasons. Modeling proteins based on their amino-acid sequences aids in product characterization and helps companies make process and formulation decisions early on. And some are using it at the product discovery and lead-optimization stage to find problematic sequences that are likely to induce immunogenicity in circulation. T cells bind to specific epitopes (small linear fragments of protein antigens) displayed on the surface of antigen-presenting cells (17). Computer algorithms have been created to map the locations of such epitopes in protein three-dimensional structures using frequency analysis, support-vector machines, hidden Markov models, and neural networks.

With programs such as the EpiMatrix system from EpiVax and Antitope’s iTope software and T Cell Epitope Database information, companies can quickly screen large genomic sequences for putative epitopes — and successes among analysts involved in vaccine design and studying autoimmune disorders are naturally leading others to apply them in assessment of unwanted immunogenicity. Some developers say their software can measure the potential immunogenicity of whole proteins and their subregions. An “epitope-density approach” is gaining acceptance for comparing protein therapeutics, such as in drug discovery and biosimilar development. It may reduce the risk of failure due to immunogenicity in the clinical setting (17).

In silico approaches are still a new technology. “They are potentially useful in R&D,” says Maggio of Aegis Therapeutics “but they may not impress regulatory agencies” when incorporated into investigational new-drug (IND) applications for going into clinical trials. Robin Thorpe (scientific advisor at ImmunXperts and an expert in the regulatory field) says that regulators consider them a way to guide further investigation. “But for product development, in silico analysis can aid in selection of most appropriate versions of products for development. To prevent overestimation, it is almost always advisable to confirm results using in vitro T-cell assays.”

Many scientists may feel more comfortable with in vitro results — especially those already using cell-based assays to measure product potency, for example. Peptides some 9–25 amino acids long can be used to measure T-cell responses in vitro (17). Animal testing, too, can provide results that may be predictive of human immunogenicity — especially that associated with protein aggregation. But an important caveat here is that predictive information is used more to guide choices made in product design, formulation, and clinical testing — not to serve as safety data by itself.

“Given the number of other factors that may equally influence immunogenicity, it seems very unlikely that [in vitro or in silico] methods may one day fully predict an immune response to either monomeric or aggregated therapeutic proteins” (17). The EpiScreen assay from Antitope is one success story: a preclinical in vitro assay that has been shown to correlate with clinical data (18). Companies such as Merck and Amgen have used it in development as well as in reformulation of problem drugs (19–21). The “Expert Commentary” box on the next page goes into more detail.

In vivo testing, on the other hand, involves the entire immune system of a complex organism and thus can “better simulate the extremely complex scenario that results in complete immune responses, especially when protein aggregates are involved.” But animal models must be well chosen because most modern recombinant proteins are humanized or fully human in nature, making them foreign to animals by nature. Some transgenic and knock-in/ knock-out strains have been developed, as a result, and they do offer promise as predictive immunogenicity models. This science, however, is still in its early stages.

Preparation and Prevention: Risk management follows from risk assessment. Biopharmaceutical companies have many options for addressing the problem of immunogenicity when it arises. Factors associated with drug administration and patient-specific conditions can be difficult for biomanufacturers to control. Often such issues must be addressed through changes at the clinical level. For example, the erythropoeitin problem led to a change in the product’s mode of administration from subcutaneous to intravenous as well as revised storage and handling protocols, along with increased patient monitoring and maintenance efforts (12). This may be an area where pharmacogenomics and personalized medicine can make a difference: by identifying patients who are more or less likely to have a problem as well as those inclined to benefit from a given therapy.

For product-related factors, biomanufacturers can make formulation changes to prevent aggregation or even change the drug substance itself. When predictive analysis has identified T-cell epitope issues, deimmunization by epitope modification is one strategy for reducing protein immunogenicity. Some proteins may benefit from fusion or conjugation with other molecules (e.g., polyethylene glycol (PEG), which is also used as a formulation excipient) to help make them more “invisible” to patients’ immune systems — primarily by improving their solubility and making them less prone to aggregation.

Epitope Modification: When predictive methods identify potential immunogenicity issues early in product development, companies have more options available for addressing such problems than if they weren’t discovered until clinical testing. Humanization is such a common answer for antibodies that it’s almost a given part of product development now. Most companies add amino-acid sequence segments derived from variable regions of human antibodies, taking care not to use known T-cell epitopes. If nonhuman glycosylation patterns are the problem, then an expression system can be chosen that performs the correct posttranslational modifications.

Modern predictive methods can identify problematic T-cell epitopes on recombinant proteins in development, allowing companies to use selective mutation methods that remove them. Once a library of sequence variants has been designed and expressed, those can be screened for retention of their therapeutic activity as well as reduced immunogenicity using the same in silico, in vitro, and/or in vivo methods.

Formulation Changes: Unwanted immunogenicity isn’t always attributable to a protein’s amino acid sequence. If problems arise later in development, the drug product’s formulation is often to blame. And that usually means that aggregation is involved. Protein aggregation has consequences beyond immunogenicity, as well. Aggregated proteins lose their bioactivity and stability, often irreversibly, and they waste valuable product. Aggregation- prone proteins need to be protected carefully throughout bioprocessing, especially during shaking and shipping; freezing and thawing; drying and reconstitution; and formulation, fill, and finish. Structural modifications can help, but the range of formulation options should be considered first. When aggregation in processing is inevitable, the last resort option is to remove aggregates before drug-product formulation — but that wastes valuable product and thus cannot be very cost effective (22).

Maggio of Aegis focuses on polysorbate problems (8). “Replacing polysorbates with surfactants that do not cause progressive protein degradation (and increased immunogenicity) represents a critical need and a significant opportunity for creation of better innovator and biosimilar biotherapeutics.” He points to alkylsaccharides as a promising alternative. Historically used in both food and cosmetics, these nonionic surfactants are each made of a sugar coupled to an alkyl chain. “They are being adopted by innovator biotherapeutic companies such as Hoffmann- La Roche, which has recently licensed Aegis’s ProTek dodecylmaltoside for stabilization of products in development. Results have been published for an interferon product (23). A second “big pharma” company has entered into a ProTek license similar to Roche’s, Maggio reports, and two more are in early discussion with his company regarding both MAbs and non-MAb biotherapeutics.

Tolerization: Even fully human antibodies can be immunogenic. A new approach to reducing the problem introduces regulatory T-cell (Treg) epitopes to the protein sequence that induce immune tolerance to it. This may well complement humanization (replacing foreign sequences with human ones) and deimmunization (reducing T-effector epitopes) approaches to product design. The concept brings us back to one expert’s claim that most proteins are actually tolerogenic rather than immunogenic (1). It leads us to the question: How do we convey that property on others? Immune responses are controlled by
several different biological mechanisms, some of which could be exploited for induction of tolerance to protein therapeutics (24). Natural tolerance is the basic recognition of “self”; adaptive tolerance (the basis of allergy immunotherapy) happens with certain types of repeated exposure — when the body recognizes that a given “invader” is not harmful.

Tolerization may offer an option for some products that are already on the market. For example, humanized alemtuzumab (Genzyme’s Campath) antibodies treat leukemia and potentially multiple sclerosis. The product was humanized because the original rat-derived version induced neutralizing antibodies in many clinical trial participants (24). With up to 75% of patients still developing antibody responses to the humanized form (especially after several doses), Genzyme is considering administration of a nonbinding version to tolerize patients to the drug before they begin a course of therapy. Some experts are saying this approach has “significant potential to accelerate development of biobetter products.”

From Start to Finish — and Beyond
Immunogenicity is a key metric of product safety; that makes it a key metric of product quality. A biotherapeutic product’s life cycle can be broadly described as moving from discovery/design through lead optimization, product characterization and preformulation, process development, preclinical testing, and clinical trials to market authorization — and from there on to postmarket surveillance. Companies often think of the biologics license (BLA) or new drug application (NDA) as the final goal. And they often consider quality, safety, and efficacy to be a “three-legged stool” of separate but interdependent parameters. But risk management continues beyond market authorization, and quality could be said to include both safety and efficacy. After all, no product can be called “high quality” if it doesn’t do the job it needs to do without doing any harm — no matter its purity.

Once a bit of an afterthought, immunogenicity is now an essential part of biotherapeutic development from start to finish. The industry’s (and its regulators’) growing knowledge of this subject is making proving, predicting, and preventing it an ever more integral part of quality by design.

Expert Commentary: Predictive Analysis I spoke with Chloé Ackaert, a project manager at ImmunXperts SA in Belgium, about in silico and in vitro immunogenicity analysis. Here’s part of our conversation:Q: How does in silico immune profiling fit in with the biobetter concept? A: In silico analysis and sequence optimization is the first step, but still every optimized protein needs to undergo other analyses before it can be concluded that the changes made resulted in a better protein. Sequence optimization might need other optimizations, as well. First, one must always ensure that the protein is not impaired or altered functionally. Second, changes in amino-acid sequence may imply different structure, stability, and/or pharmacokinetics and pharmacodynamics. Finally, the formulation might need to be adapted to suit a new sequence better. This approach has a different aim and consequences from screening a large number of candidates in discovery. But both provide information based on calculations and statistics, which benefits from subsequent in vitro analysis to select the best candidate.Q: How do you distinguish between predictive analysis and immunogenicity assessment? A: Predictive is a strong word that we avoid with any ex vivo analyses. We prefer to say “early immunogenicity risk assessment.” We don’t have a crystal ball with which to predict what a certain product will do in patients. But we can generate a risk profile that puts a product into perspective with others for which the immunogenic potential is known. By combining all information from available analyses, you can make an informative profile of a candidate molecule. But unknown factors remain that can lead to a different profile once a drug is in the clinic. So we don’t predict how a molecule will behave; we assess its immunogenic risk in preclinical models. Immunogenicity assessment is generally used to describe antidrug antibody (ADA) measurement, which is currently possible only once a product is administered to a living organism.Q: What are the most promising methods? A: We advise customers to combine in silico and in vitro analysis, the first to screen a large panel of molecules for the best candidates. With those selected candidates, the best way to proceed is to perform an in vitro analysis, in most cases a T-cell proliferation assay with human immune cells. Use of those cells is very important to provide a strong surrogate marker for the potential of drug candidates to induce an ADA response later on. That puts their immunogenic potential of in perspective with that of known products (benchmark molecules). The key is to screen early to avoid problems later on. Q: What’s the most problematic aspect of this work? A: Looking at data without the right context can lead to wrong conclusions. The assays and technologies available today are of a very high quality. But the importance of using the right controls and benchmark molecules is becoming clear and urgent. A key concept that could improve the science is standardizing the currently used assays and how they present results. Currently, every service provider has its own approach, which makes assay comparison difficult. At ImmunXperts, we aim to centralize all assays available and gather information in a structured, standardized, and coherent way. This helps us merge results provided by different experts with our final aim of providing the most complete immunogenicity profile for a given drug candidate.

Expert Commentary: Predictive Analysis, Deimmunization, and HumanizationI spoke with Neil Butt, vice president of business development at Abzena plc in England, about immunogenicity analysis and antibody humanization. Here’s part of our conversation:Q: Tell me about your company’s EpiScreen assay. A: It is an ex vivo assay that quantifies T-cell responses to protein therapeutics based on stimulating peripheral blood mononuclear cells (PBMCs) isolated from human donors with test samples. The assay is performed as a service and is not availble in a kit format. Three whole proteins run in parallel can be assessed for immunogenicity in eight weeks, with additional proteins (run in parallel) taking a little longer. The assay has not been automated to date.Q: How will EpiScreen data fit in with in silico information? A: The EpiScreen assay simulates the four key events that lead to a clinical immunogenic response: antigen uptake by antigen-presenting cells (APCs), antigen processing within each APC, presentation of processed antigen in the context of the major histocompatibility complex (MHC), and recognition of the peptide–MHC complex by T-cell receptors. In silico tools take into account only the third stage of that process and are thus very overpredictive of overall immunogenicity. They can be used as a low-resolution screen, but accurate assessment requires the use of a T-cell assay. The two can be combined for deimmunizing proteins. Once a T cell epitope is identified, in silico tools can be used to design protein variants that do not contain those MHC-binding peptides. Q: And how would you say those results complement whole-organism studies? A: Currently there are no ideal animal models for predicting the immunogenicity of protein therapeutics in man. HLA class II transgenic mice can be used to identify T-cell epitopes derived from therapeutic proteins. But they are limited by the HLA diversity that can be tested. An alternative to transgenic models is engraftment of neonatal NOD scid IL2Rγnull mice with human CD34+ stem cells derived from PBMCs. But their MHC class II repertoire is limited and they are not tolerized against all human proteins. Without germline transfer of genes encoding human immune cells, each mouse has to be engrafted individually (which is expensive and time consuming). The EpiScreen assay provides an alternative, more robust technology that has been demonstrated to correlate with clinical immunogenicity. Q: How important is it for companies to approach immunogenicity analysis from all those different angles? A: Immunogenicity assessment can be incorporated at all stages of the development process, and each tool has its place. In silico tools can be used to narrow down a large number (10–100) of lead sequences to a manageable number for T cell assays. Ex vivo time-course assays can be used to assess the overall immunogenicity of a smaller subset of preferred proteins (1–10) during early lead selection. T-cell epitope mapping can be used to map the precise location, number, and potency of T-cell epitopes within a protein sequence — allowing for potential deimmunization. In silico tools then can be used for redesigning that sequence either to remove T-cell epitopes or design new therapeutic proteins without them. Later, an EpiScreen time- course assay can again be used to assess the success of deimmunization as well as the effects of formulation and aggregation on therapeutic proteins. Q: Can you elaborate on the idea of deimmunization? Does it apply only to new products/biosmiliars? Or can it help turn marketed proteins into biobetters? A: Deimmunization technologies can apply to antibodies and other proteins. Our Composite Human Antibody technology combines humanization and deimmunization technologies into one platform to make fully humanized antibodies without T-cell epitopes. For other proteins, our Composite Protein technology incorporates EpiScreen technology for accurate identification of T-cell epitopes and validation of their absence in deimmunized proteins, with iTope technology and the TCED database for designing deimmunized sequences. Both technologies apply to development of both new products and biobetters (the latter based on molecules that exhibit high levels of clinical immunogenicity). Q: Does “humanization” not apply to nonantibody glycoproteins? A: The concept is based on the adaptive nature of the immune response, allowing an antibody to be derived that will bind to virtually any antigen with high affinity. Because humans cannot be immunized experimentally with antigens to isolate a specific antibody, the procedure is performed in model animals to create and isolate hybridomas that secrete antibody with the desired specificity. Because such antibodies are nonhuman, they require engineering to make them “human-like” before they can be used clinically. For most other classes of protein, a direct homologue exists in humans that matches the activity of a protein identified in animal models. So that can simply be expressed and used as a therapeutic without humanization. An exception to this rule are replacement therapies for certain diseases in which the protein is absent due to genetic mutation. When administered to patients, it would appear foreign to their immune systems, giving such proteins the potential to be immunogenic. Similarly, some other proteins that are clinically useful and derived from microorganisms (e.g., bacterial toxins, for which no human homologue exists) tend to are highly immunogenic and must be deimmunised by removal of T-cell epitopes.

References
1 Lollar P. The Immune Response to Blood Coagulation Factor VIII. 14th Annual Immunogenicity for Biotherapeutics. Baltimore, MD, 18 March 2013. Institute for International Research: New York, NY.

2 Ryan S. Explaining Immunogenicity. Irish Med. Times 10 September 2014; www.imt.ie/clinical/2014/09/explaining- immunogenicity.html.

3 Thorpe R, Wadhwa M. Unwanted Immunogenicity of Therapeutic Biological Products: Problems and Their Consequences. BioProcess Int. 1(9) 2003: 60–62.

4 Schellekens H. Immunogenicity of Therapeutic Proteins: Clinical Implications and Future Prospects. Clin. Ther. 24(11) 2002: 1720–1740.

5 Cai X-Y, Cullen C. Current Regulatory Guidelines for Immunogenicity Assays. Immunogenicity Assay Development, Validation, and Implementation. Future Medicine: London, UK, November 2013: 6–17.

6 ICH S6(R1). Preclinical Safety Evaluation of Biotechnology-Driven Pharmaceuticals. International Conference on Harmonisation of Technical Requirements for the Registration of Pharmaceuticals for Human Use: Geneva, Switzerland, 2011; www.ich.org/fileadmin/Public_ Web_Site/ICH_Products/Guidelines/Safety/S6_R1/Step4/S6_R1_ Guideline.pdf.

7 Purcell RT, Lockey RF. Immunologic Responses to Therapeutic Biologic Agents. J. Investig. Allergol. Clin. Immunol. 18(5) 2008: 335–342.

8 Maggio ET. Polysorbates, Immunogenicity, and the Totality of the Evidence. BioProcess Int. 10(10) 2012: 44–49; www.bioprocessintl. com/sponsored-content/preclinical-immunogenicity-assessment.

9 Filipe V, et al. Chapter 10: Aggregation and Immunogenicity of Therapeutic Proteins. Aggregation of Therapeutic Proteins. Wang W, Roberts CJ, Eds. John Wiley & Sons: Hoboken, NJ, 2010; 416–433.

10 Wasylaschuk WR, et al. Evaluation of Hydroperoxides in Common Pharmaceutical Excipients. J. Pharm. Sci. 96, 2007: 106–116.

11 Ha E, Wang W, Wang YJ. Peroxide Formation in Polysorbate 80 and Protein Stability. J. Pharm. Sci. 91, 2002: 2252–2264.

12 Schellekens H. Immunologic Mechanisms of EPO-Associated Pure Red Cell Aplasia. Best Pr. Res. Clin. Haematol. 18(3) 2005: 473– 480.

13 EMEA/CHMP/BMWP/14327/2006: Guideline on Immunogenicity Assessment of Biotechnology-Derived Therapeutic Proteins. European Medicines Agency: London, UK, 13 December 2007; www. ema.europa.eu/docs/en_GB/document_library/Scientific_ guideline/2009/09/WC500003946.pdf.

14 Mire-Sluis A, et al. Analysis and Immunogenic Potential of Aggregates and Particles: A Practical Approach, Part 1. BioProcess Int. 9(10) 2011: 38–47. Part 2. BioProcess Int. 9(11) 2011: 38–43.

15 Goodman J. Using the Aushon Cirascan™ to Develop Sensitive and Drug Tolerant Immunogenicity Assays. 14th Annual Immunogenicity for Biotherapeutics, Baltimore, MD, 19th March 2013. IIR USA: New York, NY.

16 Sauerborn M. Is the Tiered Immunogenicity Testing of Biologics the Adequate Approach in Preclinical Development? Bioanalysis 5(7) 2013: 743–746.

17 De Groot AS, Martin W. Reducing Risk, Improving Outcomes: Bioengineering Less Immunogenic Protein Therapeutics. Clin. Immunol. 131(1) 2009: 189–201.

18 Baker MP, Jones TD. Identification and Removal of Immunogenicity in Therapeutic Proteins. Curr. Opin. Drug Discov. Dev. 10(2) 2007: 219–227.

19 Jaber A, Baker M. Assessment of the Immunogenicity of Different Interferon Beta-1α Formulations Using Ex Vivo T-Cell Assays. J. Pharmaceut. Biomed. Analysis 43(4) 2007: 1256–1261.

20 Jaber A, et al. The Rebif® New Formulation Story: It’s Not Trials and Error. Drugs RD 8(6) 2007: 335–348.

21 Joubert MK, et al. Highly Aggregated Antibody Therapeutics Can Enhance the In Vitro Innate and Late-stage T-Cell Immune Responses. J. Biolog. Chem. 287(30) 2012: 25266–25279.

22 Wang W, Warne NW. Chapter 8: Approaches to Managing Protein Aggregation in Product Development. Aggregation of Therapeutic Proteins. Wang W, Roberts CJ, Eds. John Wiley & Sons: Hoboken, NJ, 2010; 333–365.

23 Rifkin RA, et al. n-Dodecyl-β-d-Maltoside Inhibits Aggregation of Human Interferon-β-1b and Reduces Its Immunogenicity. J. Neuroimmune Pharmacol. 6(1) 2011: 158–162.

24 De Groot AS, et al. Beyond Humanization and De-Immunization: Tolerization As a Method for Reducing the Immunogenicity of Biologics. Exp. Rev. Clin. Pharmacol. 6(6) 2013: 651– 662.

For Further InvestigatIon
Brinks V. Immunogenicity of Biosimilar Monoclonal Antibodies. GaBI J. 2(4) 2013: 188–193.

Carr F. Fully Human Antibodies: Myths and Magic Bullets. MedNous November–December 2013: 18.

Dobrovolskaia MA. Nanoparticles and Immunogenicity. 14th Annual Immunogenicity for Biotherapeutics. Baltimore, MD, 18 March 2013. Institute for International Research: New York, NY.

Gunsior M. Implications of Immunogenicity in Drug Development (Global Bioanalytical Services white paper). Covance Laboratories: Chantilly, VA, 2010; www.covance.com/content/dam/covance/ assetLibrary/whitepapers/adawhitepaperpdf.pdf

Jawa V, et al. T-Cell Dependent Immunogenicity of Protein Therapeutics: Preclinical Assessment and Mitigation. Clin. Immunol. 149(3) 2013: 534–555.

Liang BA, Mackey T. Emerging Patient Safety Issues Under Health Care Reform: Follow-On Biologics and Immunogenicity. Ther. Clin. Risk Manag. 7, 2011: 489–493.

Nielsen M, et al. NetMHCIIpan-2.0: Improved Pan-Specific HLA- DR Predictions Using a Novel Concurrent Alignment and Weight Optimization Training Procedure. Immunome Res. 6(9) 2010.

Parenky A, et al. New FDA Draft Guidance on Immunogenicity. AAPS J. 16(3) 2014: 499–503.

Perry LCA, Jones TD, Baker MP. New Approaches to Prediction of Immune Responses to Therapeutic Proteins During Preclinical Development. Drugs RD. 9(6) 2008: 385–396.

Rup B. Biotherapeutic Immunogenicity Risk Factors: The Science, Reliability, and Concepts for Implementing Predictive Tools to Improve Their Reliability. Mastering Immunogenicity Summit, Boston, MA, 13 September 2011. ProImmune Ltd.: Oxford, UK.

Sauna ZE. The Assessment of Immune Responses Against Biological Drugs. Mastering Immunogenicity Summit, Boston, MA, 13 September 2011. ProImmune Ltd.: Oxford, UK.

Sauna ZE. The Immunogenicity of Protein Therapeutics: Time to Get Personal? Mastering Immunogenicity Summit, Boston, MA, September 2012. ProImmune Ltd.: Oxford, UK. www.proimmune.com/ ecommerce/pdf_files/sauna1.pdf.

Shankar G, et al. Assessment and Reporting of the Clinical Immunogenicity of Therapeutic Proteins and Peptides: Harmonized Terminology and Tactical Recommendations. AAPS J. 16(4) 2014: 658– 673.

Stas P, Lasters I. Strategies for Preclinical Immunogenicity Assessment of Protein Therapeutics. IDrugs 12(3) 2009: 169–173.

Van Beers MMC, Bardor M. Minimizing Immunogenicity of Biopharmaceuticals By Controlling Critical Quality Attributes of Proteins. Biotechnol. J. 7(12) 2012: 1473–1484.

Immunogenicity of Biopharmaceuticals. Van de Weert M, Møller EH, eds. Springer: New York, NY, 2008.

Weber CA, et al. T Cell Epitope: Friend or Foe? Immunogenicity of Biologics in Context. Adv. Drug Deliv. Reviews 61(11) 2009: 965–976.

Cheryl Scott is cofounder and senior technical editor of BioProcess International, PO Box 70, Dexter, OR 97431; 1-646-957-8879; cscott@bioprocessintl.com; www.bioprocessintl.com.