Elucidating Biosimilars Characterization

The global market for biosimilar drugs has been forecasted to be $2.445 billion in 2013 according to a report by the British market-research firm, Visiongain (1). The growth corresponds to a 20% increase from last year’s figures and accounts for approximately 2% of the overall biologics market. Although narrowly focused on only a small number of therapy areas at present, the biosimilars market is set to expand over the next decade and beyond as a result of two major factors: the impending patent expiries on blockbuster biologics and the financial environment that is driving payers to push for wider adoption of biosimilars to manage the escalating costs of healthcare. While many companies are keen on getting a share in the biosimilars market, bringing these complex molecules from bench to launch can be a challenge, not just during the development stage but also in terms of the manufacturing process involved.

Pharmaceutical Technology spoke with Jonathan Bones, principal investigator at Ireland’s National Institute for Bioprocessing Research and Training (NIBRT); Reg Shaw, PhD, CEO of NIBRT (collectively referred to as NIBRT thereafter); Kamali Chance, PhD, head of global biosimilars regulatory strategy, Biosimilars Strategic Unit; Colin Vose, PhD, vice-president, Centre for Integrated Drug Development; Doris Weilert, PhD, senior research pharmacokineticist, Early Clinical Development, Pharmacokinetic/Pharmacodynamic; and John Patava, PhD, director, head of biosimilar intelligence and capabilities, all at Quintiles (collectively referred to as Quintiles thereafter) about the importance of characterization studies during biosimilars development and related analytical methods.

The complex nature of biosimilars

PharmTech:

Why are biosimilars not considered identical to their
original biologic products?

Quintiles: Biosimilars are not approved using the same pathway as generic medicines because of the sheer size and complexity of the molecules that make up biological medicines. Generic medicines are copies of relatively small, well-characterized molecules with low molecular weights produced using well-defined processes involving chemical synthesis. As such, the active ingredient of generic copies and the original medicines are for all intents and purposes, identical. In contrast, biosimilar medicines are of very high molecular weight and complex structures depending on the product. Biological molecules are also manufactured by processes involving living organisms, such as genetically engineered bacteria, yeast, or mammalian cells, and although these processes can be well-defined, they are subject to the variability inherent in living systems. The active ingredient for biologics can only be considered similar, not identical.

Proteins produced in different mammalian cell lines, or under different environmental conditions, may be expressed with subtle but important differences with respect to characteristics such as glycosylation. For this reason, biosimilar medicines should never be assumed to be exact copies of the originator molecule. In recognition of this inherent variability, regulators have determined that a unique pathway is required for the testing and registration of biosimilar medicines.

NIBRT: Differences regarding the approval process for small molecules as compared to recombinant protein therapeutics are reflected in the complexity that must be considered when comparing small-molecule APIs with biologics, molecules that are often many orders of magnitude larger. For example, the cholesterol-lowering agent atorvastatin has a chemical formula of C₃₃H₃₅FN₂O₅ and a molecular weight of 558.64 g/mol. Compare this compound with the monoclonal antibody trastuzumab, which has a reported chemical formula of C₆₄₇₀H₁₀₀₁₂N₁₇₂₆O₂₀₁₃S₄₂ and a molecular weight of 145531.50 g/mol, approximately 260 times the size of the small molecule API and ultimately more structurally complex.

Because we’re dealing with relatively simple chemical species when we consider small-molecule API generics, the approval process has been tried and tested effectively over recent decades and is now well established. Rather than requiring full clinical trials, generic drug manufacturers are requested to demonstrate pharmaceutical equivalence to show that their medicine contains the same active pharmaceutical ingredient at the same purity and same dose and with the same administration route as the innovator product.

Application of the same approval approach for large biological therapeutics is not so straightforward due to the complexity of these large structurally complicated molecules. Despite advances in analytical chemistry and instrumentation, the direct assessment of comparability from analytical data is limited. Using liquid chromatography-mass spectrometry (LC-MS) methods, we can determine and confirm the primary sequence, the presence and identity of post-translational modifications and perform protein structural analysis using advanced experiments such as hydrogen-deuterium exchange MS. Comparison of analytical data forms the case for proving whether two molecules are structurally similar or not; however, data from cell-based potency and bioequivalence assays will also be required. Ultimately, it will be up to the regulatory agency to decide, based upon the presented evidence, whether a biosimlar candidate is similar enough to be approved.

The inherent analytical complexity arises from the fact that recombinant proteins are expressed in cellular systems under defined bioprocess conditions rather than using stepwise chemical synthesis. The cellular machinery that expresses the recombinant protein is sensitive to the physiochemical environment of the cell, the availability of nutrients, and removal of toxicants and inhibitory compounds. All of these factors can affect the quality and quantity of the expressed protein.

As companies generating biosimilar molecules do not have access to the intellectual property regarding the innovators’ manufacturing process, the process for generating the biosimilar molecule will have, by nature, inherent differences. Therefore, it is inappropriate to say that two molecules are identical; substantive analytical, biochemical, and if necessary, clinical comparison between the biosimilar and the innovator molecules will have to be demonstrated.

Bioequivalence testingPharmTech: Can you explain the procedures for testing the bioequivalence of biosimilars and how it differs from bioequivalence testing for generic drugs?

Quintiles: The principles of pharmacokinetic bioequivalence testing for proteins are essentially very similar to those for classical medicinal-chemistry small-molecule products and are based on a comparison of peak concentration, time-to-peak concentration, area under the concentration-time curve; however, depending on the mechanism of action of the protein, such a study may need to be done in patients. Although it may be possible to carry out such a study in healthy volunteers, the products are “foreign” proteins that may produce an immune response on repeated administration. In addition, some proteins, for example, monoclonal antibodies, have half-lives of weeks. These proteins would, therefore, require sample collection for many weeks to accurately define the pharmacokinetics and a washout period of some months between treatments with the originator product and biosimilar. Thus, a classical randomized crossover design as used for small molecules is not appropriate on safety and logistics grounds. Such studies, therefore, are most likely to use a parallel-group design, in which each subject only receives one of the products being compared. Such an approach increases the variability and thus, the number of subjects needed to achieve the required power to demonstrate pharmacokinetic bioequivalence.

The importance of characterizationPharmTech: Why is it important to characterize amino-acid sequence and carbohydrate structure?

NIBRT: Characterization of the amino-acid sequence--normally using LC-MS-based approaches, either top-down or peptide-centric bottom-up methods--is important to verify that the product produced is that as expected from the engineered gene sequence. Data from such experiments facilitate the determination and comparison of the experimental versus the predicted mass of the product. Deviations are indicative of post-translational or other modifications. Such modifications, if present, can then be localized when using peptide-centric LC-MS methods, or sequence variants can also be verified at the peptide level. Data from such studies can then be used to evaluate whether the presence of a particular modification is important with regard to the structure-function relationship of the molecule, or whether for safety or efficacy purposes, it may be necessary to engineer out the modified amino acid.

Similarly for carbohydrates, characterization of the glycosylation present with regard to monosaccharide analysis and structural characterization of the N- and O-linked oligosaccharides present is important to ensure that the desired glycosylation is present and that potentially immunogenic epitopes are absent.

Characterization of the glycosylation present on recombinant proteins is particularly important as the glycans attached to the molecule can modulate the stability and ability of the molecule to elicit its desired effector response, particularly in the case of monoclonal antibodies (1). Because glycosylation is cell-line specific and also affected by the environmental conditions that a cell finds itself in, it is necessary to routinely characterize the oligosaccharides present to make sure that the expressed protein is being produced with consistent and reproducible glycosylation. Furthermore, characterization of the glycosylation present is important to verify the presence or the absence of potentially immunogenic epitopes such as galactose-α1-3-linked galactose motifs and the nonhuman monosaccharide N-glycolyl neuraminic acid.

Characterization of glycosylation is still a significant analytical challenge although recent advances in analytical instrumentation and separation chemistries have benefited the field considerably. Glycans are traditionally analyzed using either liquid-phase separation techniques such as liquid chromatography or capillary electrophoresis with optical or fluorescence detection or mass spectrometry. As oligosaccharides lack inherent chromophores or fluorophores, they must be derivatized to facilitate detection, fluorescent reagents, such as 2-amino-benzamide, 2-aminobenzoic acid, 2-aminopyridie or 2-aminoacridone, have been widely reported as derivatization reagents used in glycan analysis. When using capillary electrophoresis, charged fluorescent labels such as 1-aminopyrene-3,6,8-trisulfonic acid (APTS) or 8-aminonaphthalene-1,3,6-trisulfonic acid (ANTS) are used to impart electrophoretic mobility to the labeled oligosaccharides and also to facilitate highly sensitive laser induced fluorescence detection.

Mass spectrometry (MS) has also been widely used for glycan analysis, either using matrix-assisted laser-desorption ionization (MALDI) or electrospray ionization (ESI). A caveat with the use of MS is its inability to distinguish isobaric monosaccharide residues or isomeric oligosaccharides; therefore, it is more a compositional analysis. MS/MS based methods can facilitate glycan sequencing when performed on [M+H]+ ions formed during positive ionization. MS/MS of [M-H]- ions formed when using negative ionization is considerably more informative as it can provide diagnostic ions that facilitate the deduction of linkage and positional analysis of the oligosaccharides. There have also been significant advances in associated informatics platforms for the annotation of both LC and MS data in recent years.

Furthermore, the issues of glycan micro- and macroheterogeneity must also be considered. Performing a global glycosylation screen informs us of the identity and relative abundance of the types of oligosaccharides attached to the molecule. However, combination with glycoproteomics is often necessary to identify the sub-populations of glycans present at each glycosylation site.

The combination of analytical techniques in an orthogonal manner is recommended to impart confidence to the analytical data. It is not recommended to analyze the glycosylation using a signal technique; generally, a minimum of two approaches is recommended.

Quintiles: The amino-acid sequence of a biosimilar molecule is one of the starting points in determining similarity to the originator medicine, with the draft guidance from FDA implying that a biosimilar molecule needs to have the same amino-acid sequence as the originator medicine. Some proteins, such as monoclonal antibodies, are glycosylated (i.e., they have carbohydrate molecules attached to them). The extent and the exact structure of carbohydrates attached to a protein may affect its binding to its target receptor, its clearance from the body, and potentially, its immunogenicity. The manufacturing process (e.g., cell line used) can influence the exact nature and extent of glycosylation of the protein, and thus potentially, its activity and safety. It is not possible to definitively determine the structure of a large protein, such as a monoclonal antibody of approximately 150,000 Da, using currently available analytical techniques. This means that confirming similarity of two proteins requires the use of a wide range of analytical techniques.

Functional binding of molecules can be tested using microarray analysis. This technique enables the biosimilar developer to test the binding of the protein to a large number of targets to determine binding and eliminate potential cross reactivity. In many cases, this functional binding may be adequate if the effect of the medicine is simply to neutralize its target. Where the protein-based medicine acts through some cellular signaling pathway, there are tools for assessing these modes of action such as cell-based kinase assays.

Tools for characterizationPharmTech: What techniques are used to compare the structure of biosimilars and biologics?

NIBRT: LC-MS is an extremely valuable tool for the analysis of biosimilars and biologics due to its ability to separate many components in complex mixtures and determine the mass of those components in both a qualitative and quantitative manner. LC-MS is routinely used for the characterization of biologics using top-down approaches wherein the intact mass of the molecule is determined, middle-down approaches wherein subunits or domains of the molecule are separated and independently analyzed, and bottom-up approaches following digestion of the therapeutic protein into its constituent peptides through the action of a suitable protease. Such approaches facilitate verification of the primary sequence of the molecule, identification and characterization of post-translation modifications, especially wherein the combined use of collisional-induced dissociation (CID) and/or electron-transfer dissociation (ETD) fragmentation strategies are employed. As mentioned previously, LC, MS, and LC-MS approaches are widely used for the characterization of the glycosylation present.

The combination of generated data from the analysis of the primary sequence with additional analytical platforms is often required to provide an insight into the impact of alterations in the primary sequence or post-translation modifications on the secondary, tertiary, or quaternary structure of the protein. Traditional techniques such as X-ray crystallography and nuclear magnetic resonance (NMR) spectro-scopy can be used for protein structure determination; however, the application to comparability studies can be complicated. Other methods for studying higher order protein structure such as calorimetric methods, analytical ultracentrifugation, circular dichroism, and fluorescence are capable of providing information regarding overall alterations or differences in the protein structure, but generally, are incapable of localizing the area within the sequence of structural difference. Newer methods such as hydrogen-deuterium exchange (HDX) MS are valuable tools for detecting small changes in specific regions of the protein structure. HDX-MS is also more sensitive that other technologies requiring much smaller sample amounts and is also automated.

Ion-mobility spectrometry-mass spectrometry is also emerging as a new and powerful technology to allow for the elucidation of protein structure based upon comparison of the molecules collisional cross-sectional (CCS) area in the gas phase. If the overall protein structure, say between innovator and biosimilar, is different, the molecules may have differences in drift time and associated CCS value, which may be indicative of a conformational change between the molecules.

With regard to aggregation analysis, size-exclusion chromatography (SEC), often with multi-angle light-scattering detection, is widely used for the determination of aggregates. Other methods employed include asymmetric flow field flow fractionation (A4F) or analytical ultracentrifugation (AUC), often as an orthogonal technique to SEC to increase overall analytical confidence and interpretation.

All of the aforementioned methods provide information regarding the molecules structural similarity; however, to determine similarity of function or efficacious effect bioequivalence assays must be performed.

Safety considerationsPharmTech: What are the safety issues that must be considered when developing a biosimilar product?

Quintiles: The most significant concern, from a safety perspective, with biosimilar medicines, is the risk of eliciting an inappropriate immunogenic response. It is difficult to predict, based on in-vitro characterization alone, whether a biosimilar product will be more or less immunogenic than the originator molecule. In-vivo studies in animal models are also not particularly useful for determining immunogenicity of a protein in humans. This is especially so for recombinant human proteins, or ‘humanized’ antibodies (antibodies that have a significant part of their protein amino-acid sequence from human origin), as they are highly immunogenic in most animal models. For this reason also, animal models are not particularly useful in determining pharmacokinetics of biosimilar medicines because the immune response they stimulate accelerates their clearance from the animal.

Pharmacokinetic profile and safety testing of a biosimilar medicine, with the knowledge that this product has been well-characterized and determined to be highly similar to the originator medicine, needs to be conducted in humans. As a starting point to these studies, it should be understood that most biological medicines will lead to an immunogenic response in some patients. Immunogenicity rates of between less than 1% to more than 20% have been reported for human proteins and humanized therapeutic monoclonal antibodies (2). Therefore, clinical studies need to be designed to show, not only the similarity in efficacy of the biosimilar medicine to the originator, but also its immunological profile. This assessment is done by ensuring adequate patient exposure to the biosimilar medicine beyond the initial efficacy testing phase.

Defining biosimilarityPharmTech: How do you define ‘similar’ when comparing a biosimilar with a reference product, given there will be differences caused by the manufacturing process?

NIBRT: The definition of similarity is complicated by nature of the fact that biologic products are expressed in living expression systems and differences regarding the manufacturing processes, be it cell lines used, media used, differences in downstream processing or processing between the innovator and the biosimilar process will undoubtedly exist. Furthermore, there is currently a lack of appropriate reference standard material for the development of analytical methods for the evaluation of comparability and similarity. While batches of drug product are often used, it should be considered that the formulation of the drug product may interfere in the subsequent comparability/similarity study and attempts to deformulate the drug product may unknowingly introduce modifications into the molecule, which complicates the study from the offset. It is inappropriate to say that two molecules are identical due to the inherent complexity of the manufacturing process. Indeed, it has been demonstrated that measureable differences exist between different lots of an innovator product (3). The terms comparable, similar, and highly similar require definition by the regulatory authorities. Analytical chemistry generates the data that forms or backs up the argument regarding comparability or similarity; however, it will always be the decision of the regulators as to whether they believe the data are sufficient to justify such claims.

References
1. F. Nimmerjahn and J.V. Ravetch, Nat Rev Immunol 8 (1) 34-47 (2008).
2. FDA, “Prescribing Information,” www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm, accessed Aug. 10, 2013.
3. M. Schiestl et al., Nat Biotechnol, 29 (4) 310-312 (2011).