OR WAIT 15 SECS
The author outlines an analytical strategy for establishing similarity in biosimilar development and approval.
In 2016, FDA announced its first three approvals of biosimilar products: Zorxio (filgrastim) and Erelzi (etanercept), made by Sandoz; and Inflectra (infliximab), made by Celltrion/Hospira (Pfizer) (1). While three approvals may not sound impressive, these approvals establish a clear path by which biosimilars can reach the United States market.
Since May 2016, applications for approval of biosimilars have been filed by Samsung Bioepis/Merck for infliximab; Coherus Biosciences for pegfilgrastim; Merck/Samsung Bioepis for insulin glargine; and Mylan/Biocon for trastuzumab. Four other filings have either not received FDA action yet or not received FDA complete response letters (1).
According to one publication (2), there are more than 1200 biosimilars in development against 16 major targets for potential US and global commercialization. The foundation for establishing similarity between a reference product and a biosimilar is a robust and sound analytical strategy demonstrating that the two products have similar primary, secondary, and tertiary structures, and similar functional and biological properties. This paper outlines some current methodologies and technologies available to develop a solid analytical strategy for use toward biosimilar development and approval.
In the very early stages of biosimilar development, it is important to carefully analyze a series of lots of the originator product to determine the protein sequence, identify and quantify enzymatic and non-enzymatic post-translational modification (PTM), analyze biological functionality, and establish variability in product quality attributes against which the biosimilar product in development will be measured. Once lots of the biosimilar product become available, these lots should be tested against the reference product for physical attributes; primary, secondary, and tertiary structural properties; purity and presence of impurities, including those related to the product and its manufacture; and biological activity, all using orthogonal analytical methods that have sufficient sensitivity. It is important to recognize that product quality attributes should be ranked in criticality from critical to less critical and noncritical. For the critical quality attributes, it is expected that statistical equivalence be demonstrated between originator and biosimilar products. For less critical attributes, comparability should be based on reference product variability, typically based on the average ± 2 to 3 standard deviations; for noncritical product attributes, a graphical comparison will suffice for comparability. A comprehensive analytical characterization package that should be considered for comparison of biosimilar and originator products is provided in Table I.
Strength (protein concentration/content)
High molecular weight species/aggregates
Non-enzymatic post-translational modification (PTM): oxidation
Non-enzymatic PTM: deamidation
N- or C-terminal truncation
From the series of analytical methods listed in Table I, one of the least complex tests to perform is intact mass analysis using a high-resolution mass spectrometer, preferably with electrospray ionization. Results from this method can reveal key product attributes, including the integrity of the molecule, post-translational modifications, and other types of protein modifications that may be present.
An example of the output from this type of analysis, often referred to as top-down mass spectrometry, is provided in Figure 1. Results shown in this figure were obtained from the analysis of adalimumab. From the observed molecular weights in comparison to the theoretical one, one can easily conclude that major glycan species associated with this antibody are G0F (148083.0 Da) and G1F (148243.6 Da). Furthermore, there are species, including one with a mass of 146637.7 Da, corresponding to antibody with a single glycan substitution. More detailed analysis of the data has also revealed that a large fraction of the detected species lack the C-terminal lysine on the heavy chain.
Figure 1. Molecular mass profile of adalimumab by electrospray ionization-mass spectrometry (ESI-MS).
While intact molecular mass data can be informative, more detailed chemical information is typically extracted from a well-resolved peptide map of the protein, especially when analyzed by liquid chromatography–tandem-mass spectrometry (LC–MS/MS). The mirror-image peptide maps, shown in Figure 2, are derived from the tryptic digest of adalimumab, followed by analysis by reverse-phase high-performance liquid chromatography (RP-HPLC) with detection at 210 nm and tandem mass spectrometry (MS/MS).
Not only can the analysis by peptide mapping with LC-MS/MS confirm the protein sequence, it can also identify and locate within the primary structure of the molecule any post-translational modifications, including enzymatic ones such as N- and O-linked glycosylation, phosphorylation, and other common modifications; and non-enzymatic modifications including oxidation, deamidation, N-terminal cyclization, etc. In addition, by running maps with and without reduction, in most cases, it is possible to assign cysteines involved in the formation of disulfide linkages. Fragmentation by electron-transfer dissociation (ETD) rather than collision-induced dissociation (CID) in MS/MS can further facilitate the localization of post-translational modification and disulfide bonds.
Figure 2. Comparison of originator and biosimilar adalimumab lots by peptide mapping with detection at 210 nm.
Visual comparison of the two maps shown in Figure 2 indicates that the two molecules are quite similar overall in primary structure. Upon closer inspection, however, there is evidence of differences in height for some of the peptide peaks and the presence or absence of some of the low-intensity peaks. These differences typically should be closely scrutinized to better understand the types of molecular changes that would trigger changes in peptide maps, their significance, and if they could affect the efficacy and safety of the product.
Glycosylation can potentially impact not only the biological activity of the glycoprotein, but also the product circulation half-life and can render the protein immunogenic. Beyond establishing the location of glycosylation onto the glycoprotein, it is important to fully characterize the glycans. To do this, N-linked glycans can be simply released enzymatically and then derivatized for analysis by chromatography with fluorescence detection and LC-MS/MS for structural and linkage confirmation. O-linked glycans are released chemically by beta elimination and then permethylated before analysis by LC-MS. The glycosylation comparability between an originator antibody and a biosimilar is shown in Figure 3, with fluorescence profiles displayed as mirror images. The profiles are quite comparable in that the major species are G0F and G1F, but there are minor differences with respect to the observed lower-abundance glycan species. Interestingly, in this case, the originator molecule contains a higher abundance of high mannose species, which have been shown to lead to faster product clearance from circulation (3).
Figure 3. Comparison of N-linked glycan profiles between originator and biosimilar antibodies.
Of course, significantly faster clearance of the originator product will translate into higher relative potency for the biosimilar if the two are dosed similarly.
A variety of the methods listed in Table I are available to address charge variants, impurities, strength, and potency through binding and cell-based assay; for brevity, no further discussion on these attributes will be provided in this article.
When applied correctly, tools for addressing primary structural comparability and potency (Table I) will produce high-quality data that can be used to establish a detailed picture of the primary structure and post-translational modifications of the protein. Unfortunately, not nearly as many tools are available to address protein/glycoprotein secondary, tertiary, and as applicable, quaternary structures. While there are some high-resolution methods such as multi-dimensional nuclear magnetic resonance (NMR) and X-ray crystallography, the cost, complexity, and time investment to carry out these methods precludes their widespread and routine use. Instead, there is reliance on lower-resolution methods that do not provide information at an atomic coordinate level, but provide structural information that is spatially averaged over the protein.
These lower-resolution spectroscopic methods, including circular dichroism (CD), fluorescence, and infrared spectroscopy, require some form of reporter moiety that relates to a structural characteristic of the protein, and ultimately can be correlated to the non-covalent bonding pattern of the folded polypeptide chains.
As the reporter moieties are not often associated with specific locations in the protein structure, the obtained information is a spatial average. By combining methods that rely on different reporters and different modes of measurement, it is possible to build an integrated picture of the protein structural motifs and establish biosimilarity. An example of CD results from side-by-side comparison of seven biosimilar lots with seven originator lots is provided in Figure 4.
Figure 4. Comparison of N-linked glycan profiles between originator and biosimilar antibodies.
CD spectroscopy of proteins is quite sensitive to the three-dimensional orientation of the peptide-bond group (far-UV CD; 190 to 250 nm), disulfide bonds (near-UV CD; 250 to 320 nm), and aromatic side chains (near-UV CD; 250 to 290 nm). Deconvolution of CD spectra in the far-UV can be used to quantitatively estimate the types of secondary structure, whereas near-UV CD spectra can provide useful information on the local folding environment surrounding aromatic residues. Separately, Fourier transform infrared spectroscopy (FTIR) should also be included in the study of protein secondary structure. The absorption bands from stretching vibrations of the C=O (amide I) and C-N (amide II) groups of the protein backbone are useful for quantifying different types of secondary structure; notably, FTIR is more sensitive than CD for beta-derived secondary structural conformations, which is useful in the characterization of beta-rich proteins like monoclonal antibodies.
Besides spectroscopic methods, calorimetric techniques such as differential scanning calorimetry (DSC) should also be applied for the characterization of biosimilars and comparability studies. DSC provides information on the structural stability of the folded polypeptide, given that the temperature (Tm) at which denaturation occurs is characteristic of the protein stability. Because the denaturation transitions should be the same for a protein drug product and its biosimilar, one can use the DSC thermograms to demonstrate that two products derived from different manufacturing processes are structurally comparable.
Other techniques that would complement optical spectroscopic methods include hydrogen/deuterium exchange mass spectrometry, antibody array mapping, and ion-mobility mass spectrometry. Unlike the spectroscopic methods, these alternatives require greater expertise, are likely to be more time consuming, and require significantly costlier instrumentation. However, they are expected to be much more informative than optical spectroscopy alone.
Protein aggregation is an undesired type of impurity that can result in enhanced product immunogenicity (4). Structural assays that have sufficient accuracy and precision to quantify such aggregates in protein drug products represent an important component in biosimilar comparability. Determination of the soluble aggregate levels in protein pharmaceuticals has historically relied on size-exclusion chromatography (SEC). In recent years, however, there has been an increased awareness that SEC can yield erroneous aggregation results (5), due to several factors: possible adsorption of the aggregates to the SEC stationary phase; analysis under non-native conditions due to specific mobile-phase requirements; physical filtration of large aggregates; and dilution effects resulting in disruption of weak aggregates (6, 7).
To avoid these SEC shortfalls, two alternate orthogonal methods-analytical ultracentrifugation (AUC) and field-flow fractionation (FFF)-have been more widely used. Both methods employ instrumentation, and more importantly, separation mechanisms that differ from SEC. Indeed, in many cases, AUC can provide evidence of large soluble aggregates present in a drug product that can go undetected by SEC, for reasons described in the previous paragraphs. Furthermore, AUC, by contrast to SEC, does not result in analyte dilution or capture on a stationary phase, and can often be done directly in the formulation matrix. With recent improvements in AUC and FFF resulting in better precision and accuracy, either of these methods can be used to accurately quantify the aggregate content in support of biosimilar comparability studies.
To validate the analytical approaches discussed, a review of the comparability strategies related to FDA-approved biosimilar products has been undertaken. A summary of the comparability approaches used by Sandoz for filgrastim and etanercept and Celltrion for infliximab is provided in Table II.
Briefly, the developers placed significant emphasis on the confirmation by LC-MS/MS of the primary structure and post-translational modifications. Additionally, extensive effort has gone into the demonstration of biosimilarity by a combination of binding and cell-based assays. For aggregation and other high molecular weight species, the developers have relied on SEC but also with confirmation by either AUC of FFF. Interestingly, a less uniform strategy is evident among these three products for determining higher-order structures. The similarity in secondary/tertiary structures for filgrastim was conducted mainly through 2D NMR analysis, while crystallography and hydrogen/deuterium (H/D) exchange MS were used for etarnercept, and a combination of CD, FTIR, and antibody array mapping was used for infliximab.
To date, three biosimilar products have been approved by FDA. Many other products are currently in review and are expected to receive approval in 2017. The regulatory path for getting to market in the US is becoming better defined; indeed, it is anticipated that there will be an acceleration in biosimilar approvals by FDA in the next five years.
As described in this paper, a variety of analytical tools are available to support the development of these biosimilars, especially with respect to chemical and biological/functional confirmation and comparability. While the three-dimensional structure of these molecules does present a significant challenge, there are several low-resolution methods that can be readily applied. However, more powerful methods are becoming available, including H/D exchange MS and ion-mobility MS, that can help with elucidating the spatial features of these molecules, and thus establishing comparability.
1. Pipeline Report-Biosimilar Drugs, US Specialty Care (July 2016).
2. R.A. Rader and E.S. Langer, “Future Manufacturing Strategies for Biosimilars,” BioProcess Int., May 17, 2016.
3. A.M. Goetze et al., Glycobiology. 21 (7) 949-959 (2011).
4. M.P. Baker et al., Self Nonself. 1 (4) 314-322 (2010).
5. J.F. Carpenter et al., J. Pharm. Sci. 99 (5) 2200-2208 (2010)
6. T. Arakawa et al., J. Pharm. Sci. 99 (4) 1674-1692 (2010).
7. K. Tsumoto et al., J. Pharm. Sci. 96 (7) 1677-1690 (2007).
Volume 30, Number 8
Pages: 38–43, 46
When referring to this article, please cite it as M. DiPaola, “Analytical Strategy in the Development of Biosimilars," BioPharm International 30 (8) 2017.