Irreproducible preclinical research is a global, expensive, and well-recognized problem that contributes to delays and increased costs of drug discovery (1, 2). One published study conservatively estimated the total prevalence of irreproducible preclinical research to exceed 50% with a cost of $28 billion per year in the United States alone (3), while other researchers have estimated that an astounding 85% of biomedical research is wasted as a result of correctable problems (4). Excluding scientific misconduct, which is not a major source of irreproducibility (5), lack of reproducibility typically results from cumulative errors or flaws in one or more of the following, non-discrete categories of the research and publication process: biological reagents and reference materials, study design, laboratory protocols, and data analysis and reporting (3). Although each contributes to a systemic problem that requires extensive changes in the overall scientific culture milieu, taking immediate steps to address biological reagent issues--specifically the use of cultured cell lines--is a relatively straightforward fix that will improve the credibility, reproducibility, and translation of preclinical research. It will also make more efficient use of scarce biomedical research resources.

	Immortalized cancer cell lines isolated from various human and other mammalian tissues have been used for decades across multiple areas of biomedical research (6). Their use is central to most drug-discovery projects, from initial target validation studies, through clinical candidate selection, to subsequent translational studies (7). It is essential for drug-discovery scientists to have routine access to a wide variety of high-quality, well-characterized, and contaminant-free mammalian cell lines. For these applications, accurate determination of species, sex, and tissue of origin (i.e., identity) is crucial to interpretation, validity, and translation of research results (8). Cell lines are cultured, passaged, and processed in and among laboratories with widely varying quality control (QC) procedures, while sharing cell lines is endemic--particularly in academia. For these reasons, misidentification errors, including intraspecies (most commonly by HeLa cells [9]) and interspecies (nonhuman) cross-contamination, as well as labeling and cell naming errors, occur frequently and can persist for years (10). Changes in the genotype and phenotype of cells (i.e., drift) as a result of over-passaging of cell lines or poor culture technique continue to be persistent problems (11).

	Microbial contamination of continuous cell cultures by a wide variety of microorganisms is also problematic in cell culture laboratories worldwide, particularly by mycoplasma. The latter can be difficult to detect, grow to high densities without adverse effect on cell morphology, and can impact a wide variety of cell functions, including changing response to therapeutics (12). Commercial tests are available to detect the most common forms of mycoplasma, but no affordable test exists to detect all Mollicute species.

	Although multiple organizations (e.g., National Institutes of Health [NIH]) promote or recommend best practices for handling biospecimens and other biologicals, including cell lines (7, 13), none are universally followed by a majority of biomedical researchers. Expanding the development and use of best practices and consensus-based standards for obtaining and maintaining authenticated and contaminant-free cell lines should also include smaller repositories, otherwise biological materials in the public domain will likely become compromised over time (8). 

 summarizes factors that can contribute to irreproducible cell-based research.

					Cross-contamination by other cell lines

Use and cost of misidentified and contaminated cell lines


	How widespread is the problem? One key review examined the prevalence of contaminated cell lines from 1968 to 2007 and reported combined cell line misidentification and contamination rates ranging from 18% to 36%, with only a small improvement over time (14). A more recent estimate places the cross-contamination rate at 20% (15). While intraspecies contamination receives the majority of attention, approximately 6% of cell cultures are thought to be affected by interspecies cross-contamination (16). To complicate matters, a study of more than 200 biomedical papers found that only 43% of cell lines could be unambiguously identified by their description (e.g., authors provided a name and source for the line such as a repository) (17). This type of problem-coupled with journal-imposed space limitations-is symptomatic of a widespread lack of consensus on the level of detail required to allow adequate documentation of materials and methods in the literature to facilitate external replication of study results.

	A search of NIH RePORTER identified 9000 projects and sub-projects that use cell lines at a total estimated taxpayer expenditure of $3.7 billion. If 18-36% of these research projects use misidentified or contaminated cell lines, potentially $660 million to $1.33 billion in research dollars could be affected. Based on two well-known misidentified cell lines, HEp-2 and INT 407, more than 7000 articles have been published that may have inappropriately used one or both cell lines at a total estimated cost of more than $700 million (15). Estimates of the prevalence of mycoplasma contamination of cell cultures vary widely, from 15% to as high as 35% (12, 18). An assessment of mycoplasma contamination in the National Center for Biotechnology Information Sequence Read Archive conservatively found that 11% of projects were contaminated (19). The authors estimated that hundreds of millions of dollars of NIH-funded research using continuous cell lines had been potentially affected.

	Although cell lines contaminated with mycoplasma or other microorganism constitute an ongoing and expensive concern in cell banks and cell culture labs around the world, the remainder of this paper focuses primarily on misidentification and intraspecies cross-contamination of human cell lines used in preclinical biomedical research.


	The identity of a cell line (i.e., authentication) can be determined by comparing the genetic signature (profiling or fingerprinting) with established databases (e.g., American Type Culture Collection [ATCC] in USA, Japanese Collection of Research Bioresources [JCRB] in Japan, Deutsche Sammlung von Mikroorganismen und Zellkulturen [DSMZ] in Germany) to discover  misidentified cells (20). It is important to emphasize, however, that profiling comprises only one component of understanding the complex molecular and phenotypic properties of a cell line, which is not a uniform, clonal population (8). To fully characterize a cell line requires detailed genomic, proteomic, and phenotypic analyses, which remains implausible and costly for most cell banks, let alone typical research laboratories. For this reason, cell line authentication and QC measures such as mycoplasma detection constitute an essential first step to establish and maintain the integrity of cell cultures and to enhance reproducibility of results using cultured cells. An American National Standards Institute (ANSI)-accredited, low-cost (approximately $150 fee for service or $15-30 in-house) standard for cell line authentication based on short tandem repeat (STR) profiling has been available for several years (21). 

 provides a timeline of key events in cell line-based research and authentication.

	Another DNA profiling test of cell-line identity uses single nucleotide polymorphism (SNP) variations between members of the same species within a specific locus, which are conserved during evolution (10, 22). Although commercial kits are becoming available, at present there is no ANSI-approved standard or centralized, online database for SNP-based cell line authentication. 

 compares and contrasts the pros and cons of STR and SNP profiling assays.

					a STR: www.promega.com/products/pm/genetic-identity/population-statistics/power-of-discrimination/

http://agenabio.com/products/panels/sample-id/

	Despite the widespread availability of the STR standard and its low cost, there is little evidence that authentication is routinely used in the life sciences-particularly among academic researchers (23). One widely cited survey reported that only one-third of laboratories tested their cell lines for identity (24). A 

 editorial reported that only 19% of papers using cell lines published in the latter months of 2013 conducted (or at least reported conducting) cell-line authentication (25). Although the International Cell Line Authentication Committee (ICLAC) online database of Cross-contaminated or Misidentified Cell Lines is widely considered to be the “go-to” reference in the field (26), less than half of the respondents from a 2014 Sigma-Aldrich survey were familiar with the database, and only 11% searched the database during 2013 (27).

	Many scientists remain unaware or unconvinced of the need to carefully establish and maintain cell cultures, and many do not authenticate their cell lines often enough or at all (8). More worrying is a lack of understanding of how to interpret DNA profiling results. The current 

 honor system that assumes all scientists use proper cell culture practices and authenticate their cells lines, as well as a pervasive presumption that misidentified or contaminated cells is a problem “for others” or is inconsequential for the final conclusions. This culture persists because most scientific journals-with few exceptions-do not require authentication as a condition of acceptance of research for publication (8). Merely reporting or attesting that cell lines were authenticated or checked against a database of misidentified or cross-contaminated database is not sufficient. To date, compliance levels and the impact of reporting guidelines to improve study reproducibility have been disappointing (28), but multidisciplinary efforts continue to promote transparency, openness, self-correction, and reproducibility in science reporting (29, 30).

	Although expanding the commercial availability of inexpensive assays and fee-for-service providers will help make authentication more universal, a systematic approach with commitment by all key stakeholders that embraces the importance of targeted training and education is needed. At present, there is little or no standardized training on cell-culture best practices and authentication in basic biological research groups, although these do exist in GLP and GMP labs.


	To effect meaningful change, enhancing the reproducibility and translation of biomedical research using best practices for cultured cell lines and authentication must build upon ongoing multi-stakeholder efforts to raise awareness of the issues and solutions (6). The Global Biological Standards Institute (GBSI) #authenticate campaign (www.gbsi.org/authenticate) facilitates this kind of engagement (31). NIH’s proposed Principles and Guidelines for Reporting Preclinical Research, which were developed and are endorsed by many journals and research societies, recommend establishing best practice guidelines for cell lines, such as the need to authenticate cell lines, report the source of the cell lines, and communicate their mycoplasma contamination status (32). More recently, NIH announced clarifications to their expectations from the scientific community regarding the rigor of research proposed in grant applications, as well as additions to the review criteria used to evaluate proposals; these clarifications were developed in an attempt to enhance reproducibility (33). Notably, the latter changes, which will take effect in early 2016, include expectations for authentication of key biological and/or chemical resources, such as cell lines, antibodies, and other biologics.

	Additional systematic changes are needed beyond raising awareness, expanding use of reporting guidelines, and revising proposal preparation and review criteria. Development, implementation, and dedicated funding to support targeted training and education is absolutely essential. Toward that end, in 2014 NIH launched an extramural grant initiative, “Training Modules to Enhance Data Reproducibility (R25)” (34).

	Understanding existing barriers that prevent implementation of universal cell authentication is central to changing this state of affairs. GBSI conducted an online survey to determine why cell authentication and the STR standard specifically is not used more broadly, the results of which will be shared in 2015. As a central component of a broader educational program to improve the credibility, reproducibility, and translation of the life-science research, GBSI is developing an exportable “active learning” training module to reduce cell line misidentification, mislabeling, and contamination errors.


	Targeted training, education, and access to reliable and affordable assays are crucial to change the culture of cell authentication. In conjunction with effective policies and expanded use of standards and best practices for cell culturing and authentication, knowledge of why and how often to perform cell authentication will improve; hundreds of millions of dollars in annual research expenditures will be used more efficiently; and the translation of discoveries from bench to clinical trial to bedside diagnostics and therapies will be accelerated. Considering the billions of dollars spent on cell-based research each year, expanded awareness and adoption of authentication protocols through targeted training is a relatively inexpensive way to considerably increase our annual return on biomedical research investment.

. 116, pp. 116-126 (2015).
	2. F.S. Collins, L.A. Tabak, 

 505, pp. 612-613 (2014).
	3. L.P. Freedman et al., 

. 13:e1002165 (2015).
	4. I. Chalmers and P. Glasziou, 

 505, pp. 483 (2014).
	6. J.R. Lorsch et al., 

 346, pp. 1452-1453 (2014).
	7. J.D. Wrigley et al., 

 19, pp. 1518-1529 (2014).
	8. L.P. Freedman et al., 

. 12, pp. 493-497 (2015).
	9. D.A. Kniss and T.L. Summerfield, 

. 1933719114522518 [doi] (2014).
	10. M. Yu et al., 

 520, pp. 307-311 (2015).
	11. A. Capes-Davis et al., 

 132, pp. 2510-2519 (2013).
	12. S. Rottem et al.,“Contamination of Tissue Cultures by Mycoplasmas,” in 

, L. Ceccherini-Nelli, Ed. (INTECHOPEN, Rijeka, Croatia, DOI: 10.5772/51518, 2012), pp. 35-58.
	13. R.J. Geraghty et al., 

10.1038/bjc.2014.166 [doi] (2014).
	14. P. Hughes et al., 

 43, pp. 575, 577-578, 581-572 passim (2007).
	15. J. Neimark, 

 347, pp. 938-940 (2015).
	16. W.G. Dirks and H.G. Drexler, 

 946, pp. 27-38 (2013).
	17. N.A. Vasilevsky et al., 

. 11, pp. 483-488 (2014).
	19. A.O. Olarerin-George and J.B. Hogenesch, bioRxiv, Cold Spring Harbor Laboratory, pp. 1-24 (2014).
	20. Y.A. Reid, 

. 731, pp. 35-43 (2011).
	21. ATCC/SDO. ASN-0002: Authentication of Human Cell Lines: Standardization of STR Profiling, ATCC-Standards Development Organization (SDO), Manassas, Virginia, USA (2012).
	22. F. Castro et al., 

 132, pp. 308-314 (2013).
	23. L.K. Riley and B.A. Bauer, “Authentication and Quality Control Guidelines for Human Transplantable Tumors and Cell Lines,” presentation at FELASA 2013 (Barcelona, Spain, 2013).
	24. G.C. Buehring et al., 

 40, pp. 211-215 (2004).
	25. Editors of Nature, “An update on data reporting standards,” 

 16, pp. 385-385 (2014).
	26. ICLAC, “Database of Cross-Contaminated or Misidentified Cell Lines,” http://iclac.org/databases/cross-contaminations/, accessed Aug. 12, 2015.
	27. Sigma-Aldrich, “The Second Annual State of Translational Research: 2014 Survey Report,” pp. 1-32 (St. Louis, Missouri, USA, 2014).
	28. D. Baker et al., 

. 12:e1001756 (2014).
	29. B.A. Nosek et al., 

, 348, pp. 1422-1425 (2015).
	30. B. Alberts et al., 

, 348, pp. 1420-1422 (2015).
	31. GBSI, “#authenticate,” 

Principles and Guidelines for Reporting Preclinical Research

,” accessed Aug. 12, 2015.
	33. NIH, “

Enhancing Reproducibility through Rigor and Transparency

,” accessed Aug. 12, 2015.
	34. NIH, “

Training Modules to Enhance Data Reproducibility (R25)


	Leonard P. Freedman is president, Global Biological Standards Institute; Mark C. Gibson is policy research analyst, Global Biological Standards Institute; and Richard M. Neve is senior research scientist, Gilead Sciences, Inc.

: When referring to this article, please cite it as L.P. Freedman, M.C. Gibson, and R.M. Neve, "Changing the Culture of Cell Culture: Applying Best Practices and Authentication to Ensure Scientific Reproducibility," 

Articles

Cover Story

Rafe Swan/Getty Images; Dan WardIrreproducible preclinical research is a global, expensive, and well-recognized problem that contributes to delays a

Changing the Culture of Cell Culture: Applying Best Practices and Authentication to Ensure Scientific Reproducibility

	Although guidance can be found in available best practice documents (1–3), limited information exists about how exactly larger biological manufacturers are transferring analytical methods. The author conducted this brief survey of how manufacturers transfer analytical methods as it was expected that a significant variation exists. Specifically, the preferred pre- and post-licensure analytical method transfer (AMT) options for the execution model, sample size, and acceptance criteria were surveyed. A total of eight large pharma/biotech manufacturers, representing the three major regulatory regions of the United States, the European Union (EU), and Asia, provided answers. All eight manufacturers have global operations, multiple commercial biological products, and more than 3000 employees. This survey was intended to evaluate how these manufacturers handle internal versus external AMTs as well as pre- and post-licensure AMTs. 

	The manufacturers’ representatives were contacted directly by the author via a detailed email survey request and instructions. At total of 15 manufacturers were contacted and eight responded with complete answers. Subject matter experts, responsible for AMTs, typically answered the survey questions. Only one email response was received from each of the eight participants.


	AMTs may occur at any point in a method and product lifecycle. An AMT transfer may be associated with transfer of the entire manufacturing process during product development, may occur after product licensure, and/or may be a portion of a larger technology transfer process. Or, an AMT may be required to implement the use of a new laboratory for quality-control release and/or stability testing, either within or outside the company (e.g., contract laboratory). Using Parenteral Drug Association (PDA) 

as the reference point for AMT models, execution matrices, sample size calculations, and acceptance criteria, the following options and conditions were provided so that each manufacturer’s information could be part of the evaluation (1). Most of the surveyed manufacturers follow 

for AMT execution matrices and sample-size calculations. The strategy used for AMTs can vary, and several possible options, per PDA 

	A. Co-validation-Sending and receiving laboratories participate in the analytical method validation study execution. This may be used early in the lifecycle of a test method when appropriate.

	B. Comparative study-AMT study performed concurrently by sending and receiving laboratories. Acceptance criteria determine the equivalence of the two laboratories. Historical and validation data may be used when appropriate for parts of the method transfer study. The sending laboratory typically has collected a significant amount of historical data for test method performance results in addition to test results for the samples to be tested at the receiving laboratory. Acceptance criteria are typically set based on relevant product or material specifications with consideration of the previous validation/qualification studies and/or recent routine quality control (QC) testing data. See the following B-1 and B-2 options:

		B-1. Fixed AMT execution matrix-The fixed AMT execution matrix does not integrate known test method result variation and has therefore an identical set of comparative data generated between both laboratories for each method transfer executed. A fixed execution matrix can be more advantageous when transferring multiple products to/from multiple locations.

		B-2. Variable execution matrix-The variable execution matrix does consider test method result variation and may require a larger data comparison set for highly variable test methods. A variable execution matrix may be advantageous when transferring bioassays with a relatively high degree of test-result variation.

	C. Performance verification-The receiving laboratory may already perform the method for a similar product or for another type of sample for the same product. In this case, a formal method transfer may not be required. Any reduced prospective study considered should be properly justified.

	D. Waiver-The receiving laboratory may already perform the method for a similar product or for another type of sample for the same product. In this case, a transfer may be waived. Any waived study should be properly justified using available data.

 provides recommendations to use an intermediate precision-type AMT execution matrix and detailed instructions for risk-based acceptance criteria. In addition, for the option of using a variable execution matrix, sample-size calculations and a detailed AMT example are given. Although the AMT process is conceptually similar for pre- and post-validation transfers as well as for internal vs. external transfers, PDA 

, as well as other available guidance by the International Society for Pharmaceutical Engineering (ISPE) and the US Pharmacopeial Convention (USP), is more focused on post-validation AMT.


	Because limited specific guidance exists for pre-validation AMTs, and they may not be visible to the chemistry, manufacturing, and controls (CMC) assessors, the survey questions were posed to manufacturers to understand the variation among manufacturers frequently transferring analytical methods. A total of five possible AMT cases that include pre- and post-validation transfer stages and possible sending units (SU) and receiving units (RU) were evaluated for the preferred AMT model use. To ensure that the data can be fully evaluated and compared, the questions were also provided in tabular format. Survey results for the four questions are summarized in 

	1. Which different model, sample size(s), and/or acceptance criteria do you use for:

		a. Clinical vs. commercial AMT studies?

		b. Internal vs. external (to/from contract manufacturing organization [CMO]) AMT studies?

		c. To/from CMO vs. in-country batch release lab AMT studies? (In-country batch release lab = non-US regulatory-approved labs to release commercial product into their respective countries/regions)

	2. What are the total number of samples/replicates tested by both labs, SU and RU? Are you using a fixed AMT execution matrix (e.g., n=6, n=12, n=18, n=24, or n=36 at each site, SU and RU) or a variable execution matrix (calculated sample size which considers test method variation?

	3. Was an intermediate precision-type comparison used (at least two critical variation factors selected)?

	4. What is the acceptance criterion for RU used: fixed or risk based (e.g., fixed: ≤ 1.0 SD difference between SU and RU; risk based: calculated primarily by considering specification and process capability and maximum acceptable result drift from SU to RU and result variation at RU)?

	Results for preferred use of AMT model(s) were somewhat surprising in that two of the eight manufacturers make use of all or most of the available models regardless of the stage of product development. Six of the eight limit the model options to only two, with two of the six using only model B for any of the five possible AMT cases. All manufacturers use model B, and this option is by far the most often used model. As the author expected that option B would be the most frequently used option, the question of whether the fixed (B-1) or variable (B-2) comparison models were used was therefore further evaluated, and the results are shown in 

	Six of the eight manufacturers stated that they use fixed execution matrices for pre-validation AMTs, which offer the advantage that the SU and RU always know how much testing is to be planned and executed regardless of product-type and/or product experience. A lot of the master plan’s and AMT protocol’s content can thus be readily copied to reduce the master plan and protocol generation time. For post-validation AMTs, this ratio changes in favor of using a variable execution matrix for three manufacturers. This change to a more rigorous AMT planning and execution process can be justified, as the investment of more rigorous AMT process steps is commensurable with the product development stage(s). The biological product is more valuable at the commercialization stage(s) as both successful clinical and product/process validation studies were completed. Overall, the results show significant variation among the eight survey participants with all four cases (only B-1; only B-2; only B-1 and B-2; and B-1 changing to B-2 for post-validation AMTs) represented.

	An appropriate sample size can be determined using the risk-based approach outlined in PDA 

. Based on the sample size n, the study plan should ideally be designed so that at least two independent factors (e.g., analysts and/or days) known to (potentially) impact test method results are investigated during the transfer. Statistical equivalence testing is usually performed to confirm that the transfer study results are acceptable and fall within preset limits.

	Results for sample size varied significantly in that the typical AMT sample sizes can range from a minimum of n=3 to as much as n=24. Two of the manufacturers surveyed use n=24 for all post-validation AMTs regardless of whether methods are transferred internally or externally. Most of the manufacturers surveyed tend to increase the sample size for post-validation AMTs, which again can be explained by the increase in “product value” and the level of confidence in the AMT results for late-stage/commercial products.

	In an intermediate precision matrix study, at least two independent factors (e.g., analysts and/or days) are deliberately varied to better estimate routine testing conditions. Intermediate precision at the RU could be evaluated from this data set; however, when a more detailed result interpretation is desired at the RU, a more extensive set-up may produce test result variation that is more representative of the variation that typically occurs with routine testing. Only one manufacturer surveyed indicated that it might use repeatability-type precision for pre-validation AMTs. All manufacturers surveyed use at least two independent variation factors for post-validation AMTs.

	For post-validation AMTs, more than half of the AMT acceptance criteria are risk based and calculated primarily by considering specifications and process capability and the maximum acceptable result drift (from SU to RU) and result variation at RU. The survey suggests further that some manufacturers integrate more risk-based acceptance criteria for post-validation AMTs.

	It is the author’s experience that a large data set for the product- and method-specific performance capabilities may not exist prior to method validation. The potentially large level of uncertainty in the required method performance as well as product/process capability in early product development stages can lead to setting fixed acceptance criteria. In late-stage, post-validation stages, ideally, the maximum acceptable differences between laboratories for the method performance characteristics of quantitative methods such as accuracy and intermediate precision are estimated based on product-specific historical data with respect to the specifications.

	Other approaches-such as setting the acceptance criteria based on previous validation/qualification studies and/or recent routine QC testing data with respect to the relevant product or material specifications--may also be used. When using fixed acceptance criteria such as a standard deviation-based limit for the maximum allowed difference between SU and RU, recent SU historical, product-specific data are usually not considered. Each approach has its advantage(s). The risk-based approach in PDA 

 evaluates risks to both patient and manufacturer.

 highlights the two primary sources to be considered to set risk-based acceptance criteria for analytical method performance. To be truly risk-based, from a patient and data-continuity perspective, investigators should use the product specifications and existing knowledge of product and process capabilities (1). Typically, in early-stage product development, the specifications and product/process performance can be estimated from historical data of similar products. In late-/commercial-stage product development, more product-specific historical data exist so that product-specific, risk-based acceptance criteria can be set using product-specific specifications and process/product capabilities.

	The AMT acceptance criteria should be set to balance patient and manufacturer’s expectations. First, if a high level of method capability is desired within the given specifications and expected process capability, setting tight acceptance criteria may be appropriate. Fulfilling the second driver-which involves setting protocol acceptance criteria ranges wide enough to ensure successful AMT completion-may be in direct opposition with the expectations of the patient. Acceptance criteria may be set unsuitably wide to assure that all criteria are readily passed. The method performance may therefore be considered validated, compliant, and acceptable, although the actual method performance may not be suitable with respect to specifications and/or overall process capability expectations. It is therefore important to set risk-based and balanced acceptance criteria intended to satisfy both drivers as much as possible (1).

	Uncertainty and test-result variation result in risks to patient and firm. This relationship should be understood and used to set acceptance criteria to ensure the continued suitability for use of the analytical method following an AMT. Simplified, the relationship of the primary variation sources are shown in 

. As specifications are for the observed manufacturing process variation, we should therefore control result drifting and variation at RU through risk-based acceptance criteria.


	The demonstration of equivalence in average test results (accuracy and/or matching) and similar precision (intermediate precision) performance between the laboratories is of primary interest in the evaluation of quantitative methods. A comparison of additional validation characteristics (such as quantitation or detection limits) may be considered for particular method types. Statistical tests can be used to demonstrate equivalence between laboratories. Equivalence testing by two one-sided t-tests (TOST) is generally applicable in most cases (1). The TOST results are statistically satisfactory and AMT results pass if the confidence interval for the difference in means between the two laboratories falls within an acceptable interval [-Θ, + Θ]. The interval should define the largest difference that could be accepted between the laboratories while not significantly impacting the suitability of the transfered analytical method (1).


	In summary, the survey results provide valuable insight into how large biologics manufacturers are transferring analytical methods in different situations. Some significant variation was observed among the eight manufacturers, specifically in the use of AMT models, sample size, and setting of acceptance criteria.

	Analytical method transfers are becoming more important as biological products transfers are occurring more frequently and increasingly more globally. More practical global guidance could help to harmonize current method transfer practices. This again would ensure that product approval could occur faster which is in the interest of all stakeholders.

, “Analytical Method Validation and Transfer for Biotechnology Products,” (PDA, 2012).
	2. International Society for Pharmaceutical Engineering, 

 (ISPE, Tampa, FL, 2003).
	3. USP, General Chapter <1224> “Analytical Method Transfer,” 

 (US Pharmacopeial Convention, Rockville, MD, 2012).


	Stephan O. Krause, AstraZeneca Biologics, Frederick, MD, USA

 When referring to this article, please cite it as S. Krause, “Analytical Method Transfer Conditions Used by Global Biologics Manufacturers,” 

Features

The authors present the results of a survey of biologics manufacturers to evaluate how these manufacturers transfer analytical methods.

Analytical Method Transfer Conditions Used by Global Biologics Manufacturers

	Today’s single-use technology (SUT) is more than just bioprocess bags and the silicone tubing to connect them. Most companies now use SUT bags for at least one other application, usually storage of small-volume buffers or for in-process sampling. For sample handling in particular, bags provide obvious advantages over traditional sample bottle assemblies that must be steamed in place to obtain an aseptic sample of a cell-culture vessel. However, single use now reaches into many more phases of the modern biologics manufacturing process. Bags are available with internal mixing systems from several suppliers. Disposable filter capsules containing more than 2.3 m

	Aseptic connection devices in a variety of sizes and shapes allow tubing connections in seconds. Single-use formats are available for many sensors, including those that measure pH, conductivity, and dissolved oxygen. Together, these advances in SUT have enabled wide adoption of single-use bioreactors (SUBs) across the industry. Even chromatography columns up to 60 cm in diameter are now available pre-packed and fully disposable.

	The author evaluated the potential to reduce operating costs by switching to largely disposable process equipment during the technology transfer of a commercial purification manufacturing process. The design considered replacing most of the steel tanks, filter housings, and transfer lines with SUT. Detailed design planning identified more than 80 SUT components that would be required, including roughly a dozen static bag designs from 1 L to 2000 L, mixing systems from 100 L to 1000 L, and more than 20 separate tubing assemblies, manifolds, and hoses. To support routine operations, more than 500 individual pieces of SUT would be required per batch. With only 20 batches a year, the number of SUT items that need to move annually through procurement, warehousing, release, use, and disposal quickly exceeds 10,000.

	Selecting which SUT to use is only the beginning of the implementation process. Implementation of SUT at any stage of a commercial manufacturing process requires bringing the SUT components into GMP systems as a raw material. This means generation of new part numbers, material specification documents, manufacturing procedures, release testing procedures, supply agreements, quality technical agreements, quality audits of the supplier manufacturing sites, and assessment of potential SUT extractables and leachables (E/L). The E/L testing is one of the longer duration pre-requisite activities to GMP implementation. For process development teams supporting post-approval processes, the scope and duration of this work can be easily underestimated. This workload will grow exponentially when considering the technology transfer of processes currently being developed that heavily leverage SUT in multiple steps.


	The potential for polymeric components to leach unwanted chemicals into drug products must be evaluated (1, 2). E/L data are used to assess potential risks to patients from the use of SUT in manufacturing processes (3). The scope of the assessment should include everything from small surface area polymers such as valve diaphragms to large surface area materials such as chromatography resins. A process based largely on SUT will have hundreds of SUT polymeric components to be evaluated and potentially tested for E/L. It is not feasible to expose every process component to every process solution to confirm chemical compatibility in all situations. However, risk assessment tools can be used to provide thorough assessments and prioritize high-risk component solution combinations to result in a reduced E/L study scope.

	The risk assessment tool used to design the E/L studies is a three-step process streamlined from the numerical method described by the Biopharmaceutical Process Extractables Core Team in 2002 (4). Quantitative evaluation criteria are first defined. The capability of extraction for each solution and component pair is then determined based on the criteria for both ease and extent of the extraction. Finally, the proximity to final product is determined. These three steps yield a risk action level for each SUT component solution combination. The ratings should be agreed upon by a cross-functional team, including members from the manufacturing users, technical experts, and quality assurance. Definitions and category criteria should be defined prior to making any rankings to ensure a precise, unbiased, data-driven ranking process.

	The capability of extraction is determined by rating the ease and extent of the potential extraction. The ease of extraction is rated either as difficult, average, or easy. Use of organic solvents or operating a material at the extremes of the supplier-supported chemical compatibility conditions, for example, can be considered easy to extract, whereas an aqueous solution used within the component’s recommended temperature range can be considered difficult to extract. The extent of extraction is rated either as negligible, moderate, or significant. For example, components with small surface area or short exposure duration can be considered to have a negligible extent of extraction; whereas material exposed to steam or extended durations, such as liquid storage bags, can be considered as having a significant extent of extraction. The two ratings are applied to the extraction capability matrix shown in 

, to identify an extraction capability level of L1, L2, or L3.

Figure 1: Extraction capability risk-assessment matrix. L1 has a higher capability of extraction than L3. The extraction capability level is applied to the proximity risk-assessment matrix in Figure 2. 

	This “L” rating is applied to the y-axis of the action level matrix shown in 

. The x-axis represents the proximity to finished goods and is categorized into three zones. Zone 1 represents materials used in a process step such as the final drug substance filtration or formulation step where there are no clearance steps (e.g., diafiltration step, bind and elute chromatography step) before the finished product. Zone 3 consists of component solution combinations that are substantially upstream in the process, typically separated from the drug substance by two or more clearance steps before the finished product. From this second matrix, the final risk action level is defined. SUT component solution combinations rated as risk action level 3-low risk-would not require additional E/L data to be generated. In all cases of product contact, the materials should be confirmed to meet the minimum United States Pharmacopeial Convention (USP) class VI or other compendial expectations or else additional risk mitigation steps may be appropriate. For components rated as risk action level 1-high risk-E/L data must be obtained and assessed. E/L data may not be necessary for materials rated as risk action level 2-medium risk-based on the justifications provided by the cross-functional risk assessment team. Many upstream components can typically be excluded from the E/L studies based on low-risk action levels.

Figure 2: Proximity risk-assessment matrix. Solution component combinations determined to be action level 1, or high risk, require extractable/leachable (E/L) data to be generated. Combinations determined to be action level 3, or low risk, would not need additional E/L data.

Supplier selection and dual sourcing as risk mitigation


	A manufacturing process that relies heavily on SUT requires the manufacturer to depend on the SUT suppliers. There are ways to reduce the risks resulting from this dependence. Supplier selections and the quality of your supplier relationships become crucial to ensuring consistent supply of the manufactured drug products. When identifying new SUT components to implement, leverage existing supplier relationships to expedite implementation by eliminating the need for new quality technical agreements (QTAs), supply agreements, or supplier audits. When a new supplier is required, the strength of the supplier’s quality system and their reputation in the industry are as important as the technical process needs. Understand the supplier’s validation program and lot-release process. Define requirements for endotoxin and sterility assurance. Risk assessment tools can also be used to determine the level of sterility assurance required for each component. The impact of potential bioburden ingress is not the same for all process steps, so not all components require the same level of sterility assurance. This may represent an opportunity to reduce the overall cost of goods (COGs).

	High consumable volumes necessitate consideration of both capital investment and operational expenditure costs together over a fixed period before selecting SUT technologies or suppliers. The comparative analysis of capital costs and operational expenses will facilitate important process decisions such as technology and supplier selections. Establish business agreements for ≥3 years to lock in important assumptions that impact COGs at least through the initial phases of commercial production.

	As with all raw materials, supply-chain risk can be reduced by qualifying a secondary source for SUT components. Occasionally, technology selection prevents dual sourcing. For example, several currently available disposable mixing systems utilize patented mixing technologies that are not interchangeable and require purchase of the compatible mixing bags from one supplier alone. Many SUT components, however, can be dual sourced.

	There are essentially three ways to implement dual sourcing for SUT. The first is true redundancy for 100% of the SUT components, but this presents several challenges. It is simple enough to select two qualified suppliers and approve SUT design drawings with both. However, if the SUT parts are considered critical process raw materials and the fluid handling systems are qualified as part of the process validation (PV), it may not be feasible to validate equivalence for all SUT components. Each SUT component will require creation of two material specifications, one for each source. Enterprise inventory management systems and quality control material release systems will also have to manage the two source options for each process component. The warehouse may also need a strategy to segregate the two sources and determine when each option should be delivered to the manufacturing floor. In general, it will be necessary to generate twice the number of GMP documents to incorporate all of the duplicate SUT components into GMP systems.

	A second option is to identify one supplier as primary and purchase all SUT components from that supplier for normal operations. The secondary supplier would only need to be engaged in the event of a supply-chain interruption or quality problem with the primary supplier. This strategy minimizes the number of items to be incorporated into GMP document systems. The secondary supplier, however, would need to build and maintain the templates and documentation required to produce these materials without receiving a commitment from the process owner for any significant volume purchases. This could lead to long start-up times when the material is requested. The lead time to satisfy a production request when demand is infrequent is a crucial consideration in this case.

	The third, recommended option is to make each supplier the primary source for some materials and secondary for others. This option allows both suppliers to be engaged by a constant stream of revenue and dedicate some portion of their manufacturing capacity to supporting a process. E/L testing can be performed on both films to accelerate timelines. This provides an inherent alternative if any SUT components are found to have unacceptable leachable levels late in the schedule, because it is unlikely that both tested source materials will experience the same failure. Approve design drawings for most components with both suppliers for future use. The advantage of this option compared to 100% redundancy is the reduced number of new GMP documents required. Material specifications, QC material release documents, and inventory management systems require only single examples, not one for each supplier. Manufacturing procedures can also be specific to the expected source component and will not need to incorporate provisions for two possible components being used.

	As part of the QTAs and business agreements, expectations for customer change notification and demonstration of component equivalency should be considered. An extractables study was performed by the SUT-integrator ASI comparing two platinum-cured silicone-tubing materials with similar specifications. The study found the extraction profiles were similar enough to conclude they were comparable within the variability of the analytical techniques used (5). Such data could allow the two raw materials to be deemed interchangeable so the SUT supplier could use either one without notifying the end user of a change in raw material source. This is a shift away from the traditional quality definition of supplier change notification expectations, but it may become a necessary flexibility as industry use of SUT increases. Questions around dual sourcing are being discussed by industry groups such as the BioPhorum Operations Group (BPOG) and the Bio-Process Systems Alliance (BPSA). SUT suppliers have the same interests in dual sourcing of raw materials as drug substance manufacturers, so expectations for supplier-change notification should be contractually defined to avoid future unwanted surprises.

	SUT now exists to support all phases of drug substance manufacturing. Implementation of a commercial GMP manufacturing process that heavily leverages SUT components requires a large scope of work, including legal and business agreements; procurement and supply chain system designs; GMP documentation; and E/L studies. High numbers of SUT components can also mean a significant potential for in-process leachables, but good risk-assessment tools can prioritize the potential risks and reduce the scope of the E/L testing. The SUT-dependent process also means relying heavily on the quality systems of SUT suppliers. Dual sourcing and strong QTAs can mitigate some of that risk.

 <1031>, “The Biocompatibility of Materials used in Drug Containers, Medical Devices and Implants,” USP 38-NF 33, US Pharmacopeial Convention, Rockville, MD, 2015).
	2. FDA, 

Guidance for Industry: Container Closure Systems for Packaging Human Drugs and Biologics 

(Rockville MD, May 1999).
	3. W. Ding et al., 

. 34 (6), pp. 1-11 (2014).
	4. J. Bennan et al., “Evaluation of Extractables from Product-Contact Surfaces” from the Mega Source supplement to 

 15, s22-34 (2002).
	5. J. Briggs and N. Chandarana, 

 When referring to this article, please cite it as C. Atwell, “A Risk-Based Strategy for Implementing Disposables in a Commercial Manufacturing Process,” 

Troubleshooting

The author explores a dual-supplier sourcing strategy for single-use products and its ability to mitigate business continuity risk.

A Risk-Based Strategy for Implementing Disposables in a Commercial Manufacturing Process

More than two-thirds of recombinant biopharmaceutical products on the market are glycoproteins, and every stage of their manufacture is carefully monitored and tested to ensure consistency in quality, safety, and effectiveness (1). Of the various aspects of biopharmaceutical production (such as yield, protein folding, and post-translational modifications), the host cell’s biosynthesis of attached oligosaccharides (glycans) is often the most difficult to control. Selected expression systems and even slight changes in process conditions can alter the synthesis of glycans and as a consequence, the physicochemical properties (e.g., serum half-life), safety, efficacy, and immunogenicity of the end product. Regulatory agencies worldwide require state-of-the-art glycan analyses and the demands placed on these methods have steadily increased as better technologies have been developed. Ultimately, robust, information-rich, and reproducible methods for glycan analysis must be included in regulatory filings for glycoprotein-based biotherapeutics to ensure accuracy and consistency. Method simplification and standardization will provide additional assurance that the glycan-analysis methods used are transferrable between testing sites both within and outside (e.g., contract research organizations) of the organization, ensuring better quality and efficiency in manufacturing.


	By 2008, the biotechnology company Genzyme had developed and marketed the drug Myozyme (alglucosidase alfa) for the treatment of Pompe disease, a rare and progressively debilitating disorder characterized by deficiency of lysosomal enzyme alpha-glucosidase (GAA). The company was preparing to expand the targeted treatment population from primarily children to adults. Its 160-L production facility was working at capacity, so $53 million was invested to build a 2000-L facility for Myozyme in Allston, MA (2). The company was ready to launch, but FDA rejected Genzyme’s application to sell the drug from the 2000-L plant. According to regulators, the version made in the 2000-L tank was no longer the same drug as the one produced in the 160-L tank. FDA argued that the differences in glycosylation-specifically in this case, the composition of mannose-6-phosphate-meant that the drug was no longer the biological equivalent of the original material produced in the 160-L bioreactor, and may in fact introduce unknown clinical variables. Genzyme argued that it had already conducted a clinical trial on the larger batch material, demonstrating safety and effectiveness. Ultimately, Genzyme had to market the product from the larger bioreactor under a different name.

	The incident was a watershed moment in the biopharmaceutical industry, marking the emergence of new challenges (1). First, regulatory authorities were beginning to scrutinize the glycan structures of biopharmaceutical products more carefully based on established technical guidelines (e.g., ICH Q5E, ICH Q6B, and FDA’s 

A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance

), yet there remained inconsistencies in how FDA, the European Medicines Agency (EMA), and Japanese regulators determined what is “biosimilar”. Second, products with complex glycosylation patterns have the potential to easily fall out of specification with changes in biomanufacturing processes and scale-up, so to meet the new regulatory demands, manufacturers had to start carefully characterizing product glycosylation and its relation to the biological and clinical activity of a medication, and begin monitoring relevant glycan characteristics during production (3, 4, 5).

	In the years following FDA’s decision on Myozyme, the attention given to glycan structure in biopharmaceuticals has only increased, reflecting improvement in analytical technology and a greater understanding of the role these structures play in the physical characteristics, stability, biological activity, and the clinical safety and effectiveness of a drug (6, 7). The technical guidelines for characterizing and monitoring glycans have changed little since 2008; manufacturers refer mainly to International Conference on Harmonization (ICH) documents Q5E and Q6B (3, 4). These documents list the following recommendations on characterizing glycans:

	“For glycoproteins, the carbohydrate content (neutral sugars, amino sugars, and sialic acids) is determined. In addition, the structure of the carbohydrate chains, the oligosaccharide pattern (antennary profile), and the glycosylation site(s) of the polypeptide chain is analyzed, to the extent possible.”

	Other guidelines exist, setting expectations for glycan analysis, such as FDA’s 

Guidance for Industry, Immunogenicity Assessment for Therapeutic Protein Products

, and EMA’s 2007 monograph on the characterization of monoclonal antibodies (8). The monograph says the following on glycans:

	“Glycan structures should be characterized, and particular attention should be paid to their degree of mannosylation, galactosylation, fucosylation, and sialylation. The distribution of the main glycan structures present (often G0, G1, and G2) should be determined.”

	These documents, however, present few details on how to set specification limits on glycans, or recommend technologies and procedures for consistent analytical results. The consequences for this long-standing ambiguity are that manufacturers and regulators sometimes end up with different ideas as to what constitutes a necessary specification for a glycan structure. Furthermore, companies submit reports to regulatory authorities with widely different analytical approaches. Procedures may vary even within the same organization, potentially leading to inconsistent results, analytical testing failures, and ultimately, regulatory delays.

Quality by Design vs. quality in practice


	In 2002, in response to an increasing burden on FDA of regulating product manufacturing, and a perception among companies that regulatory requirements were limiting flexibility in process optimization, FDA implemented changes through its Pharmaceutical cGMP 21

 Century Initiative and the release of FDA’s process analytical technology guidance (PAT) (5). The new approach placed greater responsibility on the manufacturers to monitor quality control through timely measurements and corrections during processing.

	Around the same time, ICH published two guidance documents: ICH Q8 

Quality Systems Approach to Pharmaceutical cGMP Regulations 

(9, 10, 11). These documents helped to further define current scientific and risk-based approaches to pharmaceutical quality control.

	The concept of quality by design (QbD) was incorporated into FDA review in 2004, which together with the aforementioned guidelines, emphasized a greater understanding of the product and its manufacturing process, and designing quality control into the process, rather than testing it after the fact (12). This approach is particularly well-suited to glycan analysis, which is typically associated with a complex set of critical quality attributes (CQAs) (such as sialylation, antennary structure, or glycan structure heterogeneity) that are important to the biological or clinical activity of the drug. The CQAs must be identified, measured during process development, and maintained within required parameters (i.e., the design space) during production.

	In the case of glycans, the measurement itself may introduce uncertainty and risk, due to a high variability of outcomes when characterizing oligosaccharide chains. An interlaboratory study presenting 11 industrial, regulatory, and academic labs with the same set of four released N-glycans demonstrated that results were not consistent between the laboratories when comparing analyses of sialylation and antennary structure (13). This particular study did not address the potentially added variability caused by sample preparation. The variability in outcomes may be due in part to the availability of numerous analytical approaches and differences between labs as to the selection of approach and limitations of available equipment. Inconsistent levels of training and expertise in glycan analysis may also have had an impact.

Basic requirements for standardized protocols of glycan analysis


	The establishment of a robust protocol for glycan analysis can help extract the maximum benefit from QbD practice; give manufacturers greater control over product quality and comparability between batches and process modifications; and ensure consistency and quality in regulatory submissions. Such a protocol should have the following features.


	A selection of known glycoproteins, glycopeptides, released glycans, and monosaccharides will help calibrate and validate any system of glycoprofiling used in the initial characterization of the product or monitoring of the manufacturing process.


	Isolated product with a known clinical safety and efficacy profile provides a reference point for comparing glycan structure of batch products under different process conditions and times.

Comprehensive identification of critical glycan attributes


	Structural features of glycans have been linked to circulating half-life of the glycoprotein in the blood (sialylation); placental transport (galactosylation); antibody-dependent cell-mediated cytotoxicity (core fucosylation); and a wide range of effector functions, bioavailability, and safety characteristics (14, 15). Critical attributes may include:

		Sialylation state, including degree and linkage type (α2-3 vs. α2-6)

		Site-specific glycosylation profiles and occupancy

		Absence of immunogenic elements such as N-glycolylneuraminic acid (Neu5Gc), deacetylated 

-acetylneuraminic acid (Neu5Ac), and Galα(1-3)Gal.

		Variations in these CQAs introduced by manufacturing can originate from selection of cell line, bioreactor conditions such as nutrient levels, pH or oxygen content, as well as inadvertent modifications during downstream purification.

Establishment of ranges of acceptable variation in complex glycosylation patterns


	Many glycoproteins, particularly those with multiple glycosylation sites, do not exist as a single species, but as a mixture of glycoforms. The natural complexity and heterogeneity of glycan structures can have important functional relevance for a protein, and even minor, low-abundance glycoform species can be crucial. For clinical purposes, each product may have a different tolerance or requirement for glycoform distribution. In particular, clarity on the extent to which low-abundance glycoforms should be identified and monitored is essential.

Adherence to best practices in sample preparation


	Selecting the most appropriate method from the wide range of published and commercial sample preparation methods can be daunting. For example, purification of glycans after release from protein may be performed by solvent precipitation, solid-phase extraction, or size-exclusion, hydrophobic-interaction, or hydrophilic-interaction chromatography. Some methods may lead to a non-stoichiometric recovery of oligosaccharides, skewing the results of glycan profiling. Recent developments in sample preparation have allowed for a reduction in preparation times and improved quantitative yields of both high- and low-abundance glycoforms (16).

Selection of glycoanalysis technologies, methods, and strategy


	There is a wide array of technologies that can be applied to glycan analysis (see 

). A series of detailed optimal workflows and best practices would help to harmonize analytical procedures between and within organizations that submit regulatory reports. Workflows would cover initial characterization through to routine monitoring and quality control. Considerations should be made with respect to the simplicity and time of analysis, as long as the required levels of accuracy and reproducibility are not compromised.

The use of orthogonal and complementary methods of analysis help compensate for systematic errors in measurement. These methods typically isolate molecules and their fragments based on different physical properties (e.g., high-performance capillary electrophoresis [HPCE] vs. hydrophilic interaction liquid chromatography [HILIC]) or analytical treatment (e.g., electrospray ionization-mass spectrometry [ESI-MS] vs. matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry [MALDI–TOF–MS]), and are compared to compensate for potential bias introduced by each analytical method.


	Pharmaceutical regulatory agencies worldwide have laid out the general principles of quality control and risk management in biopharmaceutical manufacturing. Of the many CQAs that require consideration, the variation of the N-linked and O-linked glycosylation profiles of biotherapeutic glycoproteins is one of the most complex to assess. Currently, there are numerous methods used to elucidate these structures with varying degrees of accuracy and precision. In addition, the use of these somewhat disparate methodologies makes it not always possible to directly compare results between laboratories. To meet regulatory requirements for consistent process and quality control, it would be beneficial to establish more specific and standardized guidelines for glycan analysis performance with respect to reproducibility, accuracy, and sensitivity for the characterization and routine monitoring of critical glycoforms, including those of low abundance. While such guidelines are within purview of national regulatory bodies and international consensus organizations (such as ICH), no such guidelines have been released to date. The requirements for glycan analysis described in this article could address many of the issues related to process and quality control in glycoprotein manufacturing.

. 28 (9), pp. 917-924 (2010).
	2. G. Mack, 

Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products

, EMA Document CPMP/ICH/5721/03 (Geneva, 2003).
	4. ICH Q6B 

, EMA Document CPMP/ICH/365/96 (Geneva, 1999).
	5. FDA, 

Guidance for Industry, PAT--A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance

 (Rockville, MD, Sept. 2004).
	6. R. Jefferis, 

Guideline on Development, Production, Characterization and Specifications for Monoclonal Antibodies and Related Products

, EMEA/CHMP/BWP/157653/2007 (London, 2007).
	9. FDA, 

Guidance for Industry, Q8 Pharmaceutical Development

Guidance for Industry, Q9 Quality Risk Management

, (Rockville, MD, June 2006).
	11. A.S. Rathore, A. Sharma, and D. Chillin, 

 19, pp. 48-57 (2006).
	12. A.S. Rathore, 

 27 (9), pp. 546-553 (2009).
	13. S. Thobhani et al., 

 19 (3), pp. 201-211 (2009).
	14. T. Kibe et al., 

 21 (1), pp. 57-63 (1996).
	15. A. Okazaki et al., 

 336 (5), pp. 1239-1249 (2004).
	16. M.A. Lauber et al., 


	Jennifer Fournier is product marketing manager, consumables business unit-ASR, at Waters Corporation. 
	
	

 When referring to this article, please cite it as J. Fournier, "A Review of Glycan Analysis Requirements," 

Upstream Processing

A Review of Glycan Analysis Requirements

	Peer-reviewed
	Article submitted: May 21, 2015
	Article accepted: July 27, 2015

Biotherapeutic antibodies are still the largest growing class of medicines and are used globally to treat a wide range of diseases such as cancer, asthma, and rheumatoid arthritis. Compared to traditional small-molecule drugs, their structures are complex, and they may assert their effects by binding to more than one target molecule. To date, the majority of monoclonal antibodies (mAbs) have generally targeted a particular antigen; however, there are several additional antibody formats currently in the development pipelines of pharmaceutical and biotech companies. One of these promising groups of antibodies is called bispecifics, where one antibody can simultaneously bind to two targets.

Advanced antibody engineering has resulted in the ability to manufacture new recombinant bispecific antibodies suitable for therapeutic use. FDA has approved a bispecific antibody-Amgen’s Blincyto (blinatumomab)-for the treatment of refractory B-cell precursor acute lymphoblastic leukemia (ALL). Blincyto targets cell surface proteins CD19 and CD3 simultaneously, helping to put T cells within reach of the targeted cancer cell with the goal of allowing the T cells to inject toxins into cancer cells, prompting cell death (1). Bispecific antibodies, thus, have the potential to improve cytolytic effects in cancer therapy. Bispecific antibodies are also considered an effective platform for the delivery of therapeutic antibodies across the blood-brain barrier (2). This suggests that in certain cases, bispecific antibodies may have a clinical advantage over monospecific antibodies.

	During development and quality control, a range of analytical technologies are used to characterize biotherapeutic drugs, including methods to analyze their structural integrity and activity. Target/ligand binding is an essential critical quality attribute (CQA) that needs to be monitored. The enzyme-linked immunosorbent assay (ELISA) is often used to assess ligand binding; however, it is an indirect method and the output is limited to end-point data. By contrast, label-free technologies such as surface plasmon resonance (SPR) and bio-layer interferometry allow users to follow the dynamics of the interactions so that binding and dissociation events can be directly measured. SPR is a direct binding technique that can measure sequential binding events and is a key part of any analytical toolbox, enabling the examination of dual-target specificities in a bispecific antibody within a single assay.

	In the early phases of development (3), the focus is on the drug substance and the assumed mechanism of action. Early biotherapeutic development is typically target based (4), and for a therapeutic antibody, the mechanism of action includes target binding, but may also include Fc receptor and complement binding. A lead substance shall elicit the desired functional response of the target molecule, have adequate bioavailability and biodistribution, and is evaluated from a toxicity perspective.

	As the lead candidate enters late-stage development, several CQAs have been established (i.e., properties essential for clinical safety and efficacy), including data on how it interacts with target proteins. See 

Surface plasmon resonance is helping define bispecific antibodies, the next-generation of biopharma therapeutics.

BioPharm International-10-01-2015

Changing the Culture of Cell Culture: Applying Best Practices and Authentication to Ensure Scientific Reproducibility

Analytical Method Transfer Conditions Used by Global Biologics Manufacturers

A Risk-Based Strategy for Implementing Disposables in a Commercial Manufacturing Process

A Review of Glycan Analysis Requirements

Applications of Surface Plasmon Resonance for Detection of Bispecific Antibody Activity

Dynamic Light Scattering for Non-Destructive, Rapid Analysis of Virus-Like Particles

Evolving to Meet Industry Changes

Small-Molecule API CMOs Are Thriving

Small-Scale and At-Scale Model Development and Optimization

BioPharm International, October 2015 Issue (PDF)

Market Access in China