2018 PDA Biosimilars Workshop Outlines Global Regulators’ Expectations
The 2018 PDA Biosimilars Workshop held in Washington, D.C. last September provided an opportunity for further discussion on the topic of biosimilars, enabling industry representatives, academics and global regulators to convene around an increasingly critical topic.
The first session featured perspectives from global regulators on challenges encountered with submission of biosimilar marketing applications. Their presentations focused primarily on issues around manufacturing development, commercial production, and control strategies.
Emanuela Lacana, PhD, Associate Director for Biosimilar and Biologics Policy, CDER, U.S. FDA, started the session with an insight into challenges with achieving a first round BLA approval.These include analytical similarity, control strategy and routine manufacturing. For analytical similarity, inadequate reference standard qualification and characterization of critical attributes (e.g., impurities that affect potency, glycan structures, etc.) are common problems. She reminded attendees that qualified reference standards are critical for analytical similarity assessment and release of biosimilar product lots.
Martijn van der Plas, Senior of Biological Medicines, Dutch Medicines Evaluation Board, then shared EMA expectations for biosimilarity. Biosimilar product development usually involves about 30 lots of reference product, compared to about ten lots for biosimilar product. In the case of biosimilarity, however, lot independence is not always guaranteed. Acceptance criteria for analytical similarity should be predefined, although applications frequently apply post hoc justification. It is recognized that criteria for comparability remain challenging and that there is currently no EU guidance defining expectations. Next, Chantal Depatie, PhD, Senior Biologist, Health Canada, indicated that a revised guidance on biosimilarity was released in December 2016 to reflect experience gained by Health Canada. Depatie suggested biosimilar sponsors provide a justification for a non-Canadian sourced version of the reference product as a proxy for the Canadian drug in comparative studies.
Following these three speakers, a panel discussion raised a number of points of concern. First, how does the biosimilar specification setting process compare to a traditional approach for innovator products? The panel pointed out that since clinical study lots are limited, it might be acceptable to rely on some clinical reference product data for some attributes, provided similarity is appropriately demonstrated. It is important not to miss attributes which require attention. The panel’s recommendation is not to limit the number of critical quality attribute (CQA) assessments. They suggested developing a meaningful classification system and evaluate high and low criticality attributes using different criteria, rather than trying to get rid of CQAs. Some attributes might be linked to each other and it is important to assess them in this context.
Per the panel, it might also be acceptable to pool different reference product presentations for the similarity assessment, if appropriate risk assessments are performed on significant potential differences resulting from drug product manufacturing.
Other questions directed to the problem include the following (panel responses are below):
How are nominal versus label claim differences in protein content currently viewed? It is a serious problem, and it happens. There have been precedents, notably concerning differences in clinical study outcomes when the label concentration was used for clinical studies as opposed to measured concentration.
How are differences observed in forced degradation studies to be interpreted? If there are differences, they must be explained, according to the panelists.
Why is there a concern regarding a “binary” CQA/non-CQA definition, if risk assessment system is providing additional granularity? Different “colors” and “shades” are important. In the context of similarity, there has been some confusion between how the criticality assessment is done for similarity as compared to the control strategy. Assessment of criticality should be conducted based on what is known about the product’s attributes. Tiering approach for similarity is related to the statistical method and was not intended to have a direct connection to criticality of the attributes.
Was there ever a situation when a biosimilar product used a similar or even wider specification than the innovator product? Usually knowledge of the reference product is not used for this purpose. Biosimilar specifications are usually set with clinical and manufacturing information from the biosimilar product.
FDA Hones in on Data Quality
The second session, moderated by Christopher Downey, Review Chief, CDER, FDA, addressed high-level technical challenges and how to avoid pitfalls frequently encountered during biosimilar candidate development, including data quality expectations, the creation of the final control strategy and strategic choices necessary for candidate selection and development. Jee Chung, PhD, Biologist, CDER, FDA,summarized current data quality concerns for analytical methods and preapproval inspections. For analytical methods, she emphasized criticality for suitable method qualification, validation and transfer. She explained the importance of reference standard qualification. Further, the Agency has had concerns regarding interpretation of analytical similarity whenever multiple different reference standards are used during analytical similarity testing.
Next, Catherine Srebalus Barnes, Pfizer,provided detailed information on development processes for biosimilars. The best overall approach for biosimilar product development is to draft the CQA assessment first then refine it later. Additional major considerations and/or suggestions.
For the procurement of reference product lots, she recommends that sponsors sample fewer lots over a longer time period rather than acquiring a large number of lots in a limited time frame. A formal prospective procurement plan and testing strategy is recommended. Specifications can be derived from reference product results for end of shelf life. When wider limits are used versus analytical similarity results,regulators expect tight controls.
Following this session, audience members had a number of questions for the speakers. For one, is a final analytical similarity assessment post-PPQ a common practice? Not necessarily, concurred the panelists, but PPQ lots should be included as final and representative material. Further, the FDA representatives stated that specifications are set at a certain timepoint. Things can change based on a new indication. It is not only about testing the reference product range. Sponsors should not automatically assume/use this argument.
A long and intense discussion then followed on 21 CFR part 11 compliance expectations for characterization testing data. The FDA panelists reiterated that analytical similarity replaces a portion of the clinical data, therefore, expectations for data fidelity/integrity are also relatively high for characterization results. Thus, it is important to understand the role of different datasets and to assess any potential issues with data integrity in non-GMP laboratories as FDA has concerns that data may otherwise not be accurate and complete. 21 CFR Part 11 is important here because clinical studies are reduced for biosimilars compared to innovator drugs and the analytical similarity assessment carries a great deal of weight in the overall similarity assessment.
Sponsors should therefore consider all aspects of the available data and for the basis of the reference product range (e.g., process variability versus unintended shift). Sponsors should assess whether it might be appropriate to establish a specification and a control range (to assure OOT results would be appropriately investigated).
Barnes suggested for post-approval comparability to first use ICH Q5E: Comparability of Biotechnological/Biological Products Subject to Changes in Their Manufacturing Process together (“hybrid” approach) with applying relevant analytical similarity data. Barbara Rellahan, PhD, Director,Product Quality (Biosimilars), Amgen replied that her company has not included any data for analytical similarity for post-approval changes (no reference product data included). Lacana commented that in the absence of current FDA guidance, FDA cannot comment at this point of time. She suggested that sponsors should consult with FDA for major changes until a guidance becomes available. Marjorie Shapiro, Chief, Laboratory of Molecular and Development Immunology, PhD, FDA, reminded participants that a draft guidance on post-approval changes forbiologics is planned publication this year.Theoretically, the reference product and bioimilar remains linked, connected via the Q5E change sequence through post-approval changes.
In the closing Q&A, numerous topics came up for discussion:
Do we have a common vision of the role of statistical tools in the demonstration of similarity? The consensus among panelists was that statistics should not be used alone in making an assessment for analytical similarity. Statistics are supportive but not necessarily decisive.
In addition, the signal-to-noise ratio (“sensitivity”) is critical for deciding whether an attribute is equal or different and a confidence interval for a specific signal/difference should describe the level of certainty/uncertainty. An appropriate sample size should be used to obtain the desired confidence level in the results to be compared. For the criteria for equivalence tests, panelists agreed that the acceptance criteria should ideally reflect a practical significance.
What will be a direction for the European Union? Will there be a formal guidance outlining these challenges? Currently, Van Der Plas, explained that the European Union is not ready for prescriptive guidance, but this approach can change, if for example, pharmacovigilance data show new signals.
Of course, there was much more Q&A,enough to fill several volumes! For this reason, the organizers of the 2018 PDA Biosimilars Workshop invite interested parties to consider attending the 2019 PDA Biosimilars and Vaccines Conference. All of these meetings have proven to be fruitful opportunities for discussion and the organizers plan to consolidate this material into other publications and extended articles. PDA’s biosimilars conferences and workshops remain a way for those interested in the latest regulatory expectations for these innovative products to remain in the loop.