Looking for Fingerprints: Bioanalytical Characterization of Biosimilars

In February 2012, FDA released its long-awaited draft guidance documents on biosimilars (1–3). The three documents outline the agency's thinking on the processes and requirements necessary to obtain approval for a biologic that could be demonstrated to be highly similar to an already marketed product. Central to the approval pathway is the idea that rigorous bioanalytical characterization of a proposed biosimilar would be the first step in showing that it is highly similar to a reference product. According to the draft guidance, the more comprehensive and robust the comparative structural and functional characterization of the innovator and proposed biosimilar, the more useful such characterization would be in determining what additional nonclinical and clinical studies might be needed for approval (1).

FOTOSEARCH/GETTY IMAGES

Looking for fingerprints

A key feature of the FDA draft guidance is the idea of a fingerprint-like identification of the protein, a set of quality attributes measured by orthogonal methods that in combination could be used to identify a protein and demonstrate similarity between a biosimilar and a comparator. This concept arose, in part, from experience in the approval of enoxaparin, a generic low-molecular weight heparin approved in July 2010. Enoxaparin was approved without a requirement for clinical trials, based on five criteria that the agency deemed sufficient to demonstrate that enoxaparin had the same active ingredient as the innovator product, Lovenox (4). These criteria were:

• the physical and chemical characteristics of enoxaparin

• the nature of the heparin material and the chemical process used to break up heparin chains into smaller pieces

• the nature and arrangement of components that constitute enoxaparin

• certain laboratory measurements of the product's anticoagulant activity

• certain aspects of the drug's effect in humans.

Although enoxaparin is not a protein, it is a complex biologic product that required a level of physicochemical characterization similar to that for recombinant protein products.

Language in the draft guidance suggests that the level of information that could constitute a fingerprint would go beyond the basic information required of all protein products (see sidebar, "Protein Quality Attributes"). FDA states, "It may be useful to compare differences in the quality attributes of the proposed protein product with those of the reference product using a meaningful fingerprint-like analysis algorithm that covers a large number of additional product attributes and their combinations with high sensitivity using orthogonal methods (3)." Exactly what attributes of the protein the agency would like to see characterized, and what methods should be used are not defined, in part because of the great diversity in the properties of protein products.

Protein quality attributes: many and varied

"The fingerprint will depend on the individual molecule and the relevant critical quality attributes," says Fiona M. Greer, global director of BioPharma Services Development at SGS M-Scan. "It may be possible to standardize this for a particular class of molecule but probably not feasible between types of product. For example, glycosylation may have an effect on the efficacy in one type of molecule, but in another, changes in carbohydrate are not significant. Whatever the standard attributes are, they should cover both the physicochemical (primary and higher order structure) and biological properties. Without a doubt, the inclusion of information regarding posttranslational modifications will be required."

The fingerprint analogy can only be taken so far. Although real fingerprints are unique and unchanging, the attributes of a protein are harder to pin down. Protein products demonstrate variation in posttranslational modifications that may or may not be important for the biological activity of the protein. This type of heterogeneity is an expected feature of proteins produced in cultured cells, and the approval process for any biologic product typically includes a discussion with the regulatory agency where the range of heterogeneity is defined and limits agreed upon within which the potency and safety of the product remain unaffected.

For biosimilars manufacturers, determining the extent of variability in the comparator protein is an essential part of the characterization process. Typically, this step involves obtaining multiple batches of the comparator product manufactured at different times and evaluated at different points in their viable shelf life, and determining the range of quality attributes that the product displays.

"I wouldn't recommend picking a single batch and trying to produce a protein that's 99% identical to that batch," says Jin Xu, director of protein sciences at the Massachusetts Biomanufacturing Center. The comparator will exhibit a range of physicochemical properties that are acceptable, and the biosimilar has to fall within that range, explains Xu.

Post-translational modifications

Glycosylation can account for a great deal of a protein's heterogeneity, depending on how heavily glycosylated the protein is. The population of sugar units attached to the individual glycosylation sites on any protein will depend on the host cell type used, with cells derived from different species of animals, from plants, or from microbes producing different constellations of sugar chains. For this reason, FDA recommends using the same type of cell to produce a biosimilar as was used for the innovator wherever possible. Even within a single cell culture, however, the same polypeptide will be produced in a number of different glycoforms.

Because glycosylation can have significant effects on a protein's efficacy, stability in circulation, or immunogenicity, it is a parameter that needs to be well characterized and well controlled. At a minimum, the carbohydrate content of the protein (neutral sugars, amino sugars and sialic acids) should be determined, in addition to the structure of the carbohydrate chains, the oligosaccharide pattern (antennary profile), and the glycosylation site(s) of the polypeptide chain, explains.

"There are many dimensions to glycans—you can analyze to the level of one antigenic epitope, or perform a detailed analysis, depending on the critical features of your protein." says Pauline Rudd, principal investigator at the National Institute for Bioprocessing Research and Training (NIBRT) in Dublin, Ireland, explains. FDA, she says, is interested in glycosylation variants that are known to produce immunogenicity in humans, for example, xylose or fucose from plant-derived proteins. But they will have different concerns for different proteins. In epoetins, for instance, terminal sialylation has been shown to affect stability in circulation, so for this type of product, the agency will expect that sialylation will be thoroughly described. Quantitative technologies are essential and hydrophilic interaction liquid chromatography-based technologies that separate glycans on the basis of lipophilicity, meet this criterion. Coupled with experiemntal databases such as NIBRT's Glycobase 3+ and exoglycosidase digestions they provide robust tools. Confirmation of structure by an orthogonal technology is always required. Mass spectrometry which, although only semi-quantitative, is an important tool as it separates glycan pools on the basis of mass and importantly can be used to examine fine structural details. Whatever the level of analysis, she says, glycosylation patterns must be reproducible. The developer defines limits, and the product must be within limits for every batch.

Although glycans are a critical feature of many biological products, they aren't everything, explains Rudd. There are 200 potential posttranslational modifications, including c-terminal clipping, disulfide bond formation, glycation, and deamidation. Correct disulphide bond formation is crucial for achieving the proper three dimensional structure of the protein, but cysteine residues can sometimes mispair so the disulphide bonds scramble. Rudd believes that more and more, tertiary structure analysis (i.e., analysis of 3D-structure) will be required.

Orthogonal and overlapping techniques

When developing a biosimilar, physicochemical characterization is carried out to determine the complete sequence of the innovator plus any post-translational modifications, and to evaluate the quality attributes and purity of the product. A side-by-side comparison of the innovator and a proposed biosimilar will require extensive data on both the primary structure and higher order structure of both molecules using a variety of techniques. Guidance on the choice of appropriate analytical methods can be obtained from the ICH Q6B guideline (5). A single method may provide information on multiple protein attributes. For instance, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) can provide information on size, presence of aggregates, and the variability in disulfide bond formation, while isoelectric focusing can provide information on isoform pattern that may reflect differences in glycosylation or deamidation.

The exact characterization strategy will depend on the individual molecule, but should include methods to compare size, charge, and shape of the molecules. One of the most versatile and essential methods that can be used is mass spectrometry (MS), according to Greer. MS can provide information on intact molecular weight, structure confirmation via mass mapping techniques, sequence using MS/MS and identification on disulphide bridging, heterogeneity and posttranslational modifications, including glycosylation. A host of additional techniques will also be required, particularly to compare the secondary and tertiary structure of the molecules. In this respect, circular dichroism is a useful technique to measure how the protein is folded.

The process and the product

For biologics, the cell type in which the protein is produced, the exact conditions of cell culture, and the methods used in downstream processing can all affect the disposition of the final product. A recent review of the effects of cell-culture conditions on glycosylation found that nearly every aspect of cell culture could be demonstrated to have an effect, including composition of the cell-culture medium, dissolved oxygen concentration, bioreactor pH, carbon dioxide partial pressure, temperature, shear stress, and manufacturing mode (i.e., fed-batch versus perfusion) (6). Scale-up can also affect glycosylation of the product, requiring characterization at multiple points during process development and scale up. First determining which glycosylation attributes are critical to the function of the protein, then choosing analytical methods that optimally measure those attributes, rather than trying a more scattershot approach may streamline the process (7).

More difficult than it appears

Although copying an innovator's product may appear easier than developing a novel therapeutic, the difficulties of engineering one product to match an existing product cannot be underestimated. Rudd points out that it is considerably more difficult to produce a biosimilar product than simply inserting a DNA construct into a cell line and expecting that the innovator product will be faithfully reproduced. For biologics, the process is intimately connected with the disposition of the final product, but the biosimilar manufacturer does not have knowledge of the processes used by the innovator.

Xu points out, however, that manufacturers have better understanding of and more control over processes than they did 20 years ago. Implementation of process analytical technology and use of quality-by-design principles in setting up the process will greatly facilitate production of a consistent product. For any biosimilar, he recommends engaging the analytical team early in the process so that the developer can determine how process changes will affect critical quality attributes. In some instances, there may be correlations between process parameters and protein structural attributes, allowing the protein attributes to be fine-tuned.

EMA is well ahead of the US in biosimilars development, having approved its first biosimilar in 2006, and with 14 biosimilar products now on the market. EMA has already released guidance documents that are specific for different product types, including epoetins, filgastrims, and a draft guidance for monoclonal antibodies (8–10). Such product-specific specifications will doubtless be released by FDA in time, and will help better define what a fingerprint looks like.

References

1. FDA, Draft Guidance, Scientific Considerations in Demonstrating Bioimilarity to a Reference Product (Rockville, MD, February 2012).

2. FDA, Draft Guidance, Quality Considerations in Demonstrating Bioimilarity to a Reference Product (Rockville, MD, February 2012).

3. FDA, Guidance for Industry on Biosimilars: Q & As Regarding Implementation of the BPCI Act of 2009 (Rockville, MD, February 2012).

4. FDA, Generic Enoxaparin Questions and Answers, www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/ucm220037.htm, accessed May 2012.

5. ICH, Q6B Specifications: Test Procedures and Acceptance Criteria for. Biotechnological/Biological Products (1999).

6. P. Hossler, SF Khattak, and ZJ Li, Glycobiol.19 (9), 936–949 (2009).

7. D. Fernandes, BioPharm Intl. 24 (1) (2011).

8. EMA, Guidance on Similar Medicinal Products Containing Recombinant Erythropoietins (Sept. 2010).

9 . EMA, Guidance on Biosimilar Medicinal Products Containing Recombinant Granulocyte-Colony Stimulating Factor (June 2006).

10. EMA, Draft Guideline, Similar Biological Medicinal Products Containing Monoclonal Antibodies (Sept. 2010).