While other stakeholders also need clarification of the FDA policies, it is the developers who will make it possible to bring lower-cost biosimilars to make a real difference. In this paper, I am providing comments on the proposed action plans of the FDA and pointing out what I believe to be the missing elements.
The Biologics Price Competition and Innovation Act of 2009 (BPCIA) created a new category of biological products, biosimilars, which have similar safety and efficacy profiles as US-licensed reference products. The BPCIA was approved as part of the Affordable Healthcare Act.
Now, a decade later, the FDA has approved only 15 biosimilar products, most of which are not yet launched for legal reasons, and the FDA has yet to approve its first interchangeable biologic product. This pace of market entry is much slower than what the FDA had anticipated. The reasons for the slow entry of biosimilars are many, and I was able to analyze, summarize, and publish my findings recently (in the European Pharmaceutical Review, GaBI Journal, and BioProcess International) based on my firsthand experience in developing biosimilars.
I sent these publications to the FDA, and a few months later, I filed a citizen petition wherein I provided a detailed scientific thesis on why the existing guidance is impeding the development of biosimilars.
On June 21, 2018, the FDA, in an unprecedented move, withdrew its most significant and pivotal guideline, “Statistical Approaches to Evaluate Analytical Similarity” for biosimilars in line with the recommendations made in the petition. Within a few weeks of withdrawing this guidance, the FDA issued its Biosimilar Action Plan (BAP) that identified several changes coming in the biosimilars guidance.
The BAP, as currently posted by the FDA, includes a listing of 11 action items that the FDA considers to be pivotal, yet without addressing the needs of biosimilar developers in securing a faster approval by the FDA. While other stakeholders also need clarification of the FDA policies, it is the developers who will make it possible to bring lower-cost biosimilars to make a real difference. In this paper, I am providing comments on the proposed action plans of the FDA and pointing out what I believe to be the missing elements.
Below are the key elements of the FDA’s BAP and a commentary on how these elements can be made more relevant to industry.
1. Developing and implementing new FDA review tools, such as standardized review templates that are tailored to enhance public information about FDA’s evaluation of these products.
While tools tailored to educate the public about how the FDA evaluates biosimilar Biologics License Applications (BLA) may help stakeholders better understand the process of approval, they do little to help developers unless the FDA chooses to include a molecule-specific template for developers indicating the types of studies required and how they are evaluated leading to the determination of biosimilarity. The template will help developers if it clearly defines how the FDA weighs out each step of the evaluation. It is suggested that the FDA post example templates for review by the developers before finalizing their format and content.
2. Creating information resources and development tools for sponsors of biosimilar applications. This includes tools such as in silico models and simulations to correlate pharmacokinetic (PK) and pharmacodynamic (PD) responses with clinical performance. Such tools can make biosimilar drug development more efficient.
Providing information, resources, and tools for developers is the most helpful step to avoid redundancies and repeat meetings with the FDA. Since the FDA has long emphasized that one size does not fit all, common tools that are molecule-specific are most appropriate. The tools such as in silico PK/PD modeling to predict clinical performance is likely the most impactful. The goal of this exercise should be to waive in-patient studies as much as possible.
Development tools that help developers may also include enlisting the United States Pharmacopeia (USP) to craft a new class of biosimilar monographs based on identified US-licensed products; however, the FDA has advised USP not to develop any biological monographs, fearing that changes made to these monographs over time may slow down the entry of biosimilars. Ideally, a USP monograph based on a US-reference product may allow a waiver of analytical similarity testing, a significant step in expediting development of biosimilars.
3. Enhancing the Purple Book to include more information about approved biological products, including information relating to reference product exclusivity determinations.
The BPCIA provides a 12-year exclusivity to new biological drugs; the Purple Book should indicate when this exclusivity ends and identify additional exclusivities after the first licensing of the product. The FDA can allow biosimilar sponsors to request extrapolation of indications only to those indications that are not protected by a patent.
A statement regarding indications not authorized should also be included. Additionally, the FDA needs to create a policy about any future indications for a reference product that may be approved.
While developers use the Purple Book to determine the products that are available as biosimilars, there remains an explanation due from the FDA about which products do not qualify as biosimilars, as the FDA discusses in the Biosimilar Advisory Advice. Clear advice should be given about biobetters, additional indications, new modes of action, new dosage forms or dosages.
4. Actively exploring the potential for entering into new data sharing agreements with foreign regulators to facilitate the increased use of non-US-licensed comparator products in specific studies to support a biosimilar application.
Since the BPCIA requires that a biosimilar be similar to a US-licensed originator, developers are not permitted to use a non-US licensed product as a reference. As a result, creating a global dossier requires conducting 3-way studies (ie, a US-licensed product, a non-US product, and the biosimilar candidate) to develop a regulatory dossier. To reduce the burden of additional studies, and to reduce unnecessary exposure to humans, several regulatory authorities have established clear policies on bridging studies. The most stringent default requirements are imposed by the FDA to demonstrate analytical and clinical pharmacology similarity. It is also suggested that an allowed reference product be labeled as a “qualified reference product” to remove any misunderstanding at the prescriber-level.
While data sharing with foreign regulators can be useful, it does not remove the legal restriction imposed on the FDA concerning the use of the reference product. The FDA could develop a protocol for bridging studies that are limited to abbreviated analytical similarity testing in those instances where the non-US reference product was registered using mainly the same regulatory dossier and met all other requirements of the BPCIA.
Given the high cost and time it takes to bring biosimilars to market, it would be most useful if major regulatory agencies, including the FDA, European Medicines Agency (EMA), Health Canada, Australia’s Therapeutic Goods Administration, and Japan’s Pharmaceuticals and Medical Devices Agency (PMDA) work together to create a global dossier template. One of the most important efforts in the global dossier concept is the biosimilars cluster, which comprises the FDA, EMA, Health Canada, and the PDMA. Through this cluster, regulators from these 4 agencies meet 3 times a year to share information and discuss development challenges (as permitted under the intra-agency confidentiality agreements). However, the focus is on scientific alignment, not harmonization.
5. Establishing a new Office of Therapeutic Biologics and Biosimilars (OTTB) to improve coordination and support of activities under the Biosimilar User Fee Act (BsUFA) program, accelerate responses to stakeholders and support efficient operations and policy development.
This exercise is highly laudable. Whereas the BPD meetings are well organized and structured, the response time to these meetings, ranging from 60 days to 120 days, is too long and needs to be brought to 30 days. There is also a need to restructure the BPD communications by allowing sponsors to raise queries as they arise, instead of waiting for a formal meeting outcome.
6. Building on the FDA’s Biosimilar Education and Outreach Campaign, continue providing critical education to health care professionals, including releasing a series of videos that explain key concepts about biosimilar and interchangeable products.
While the FDA has done a great job in providing information on biosimilars for healthcare providers, prescribers continue to question the safety of biosimilars and even the authority of the FDA. This fact arises partly from the FDA’s practice of identifying biological products with a 4-letter suffix that could be used by originators to imply that their products are safer and even by biosimilar companies to suggest that their product stands apart from other biosimilars because of the company that manufactured it.
The FDA also needs to reprimand the companies that disseminate false information about biosimilars. A good example is the recent citizen petition filed by Pfizer against the unethical practices of multiple reference product sponsors to cast doubt on biosimilars.
7. Publishing final and revised draft guidance on biosimilar product labeling to assist sponsors in determining what data and information to include in the labeling.
The final guidance should include examples of the type of data that can be included using the prescribing information allowed in the licensed biosimilars. The FDA may also state that biosimilar developers may adopt the prescribing information of the first biosimilar with minor changes. Once a biosimilar has been approved, the label of that product could serve as a ready template for fure biosimilars of the same product.
8. Providing additional clarity for product developers on demonstrating interchangeability, including by publishing final or revised draft guidance.
The hurdles in establishing interchangeability include a legislative restriction that requires testing “in a clinical setting”; can the FDA construe this requirement to be conducted in healthy subjects? If so, can this be based on in silico or PD marker study? The best solution is to allow the biosimilar products to be tested in phase 4-type studies with approved protocols to allow a change of status.
The FDA needs to provide sample protocol outlines to demonstrate how to conduct such studies. The FDA should consider suggesting the use of markers instead of efficacy evaluations where possible.
One action that the FDA can take immediately is to issue statements that, for naïve patients, the choice of biosimilars is acceptable to enable the payors to take the reign on decision-making if the prescribers are ready. The FDA’s use of the phrase “no clinically meaningful differences,” has done more damage than good in highlighting differences. The correct statement should be “clinically similar,” to gain the confidence of the public.
9. Providing additional clarity for product developers on analytical approaches to evaluating product structure and function to support a demonstration of biosimilarity, including by publishing revised draft guidance on the use of data analysis methods, including statistical approaches.
The FDA frequently uses the term “fingerprint-like” similarity; this terminology could be replaced with “maximally similar,” and could use a quantitative approach to the degree of similarity and critical points to determine the adequacy of the demonstration of similarity. Additionally, FDA routinely uses “residual uncertainty” in describing inadequacy of similarity at different stages, a term that produces more uncertainty in the minds of developers and prescribers.
The FDA did establish a quantitative approach to the analytical similarity in its now-withdrawn guidance, wherein a 3-tier system resolved suitability of the analytical data; Tier 1 was equivalence interval based, Tier 2 was a range comparison, and Tier 3 was a visual comparison of data.
While failure to meet predefined acceptance criteria on a statistical test did not necessarily preclude determination of a highly similar status, it did create a subjective uncertainty that can be removed in a pass-or-fail system. Developers are eagerly awaiting a new data evaluation tool to allow a more rational and clinically meaningful approach for evaluating the data.
Analytical similarity testing can be made more relevant by dividing the critical quality attributes into 2 groups: release quality attributes (RQA) and structural quality attributes (SQAs).
The RQAs define the quality suitable to deliver a safe and effective product to patients. Some of these attributes are established without referring to the reference product, including protein content (proposed range, 95%-105%), bioassay (proposed range, 85%-115%), inactive components (90%-110%), or experimentally established testing reference product, more particularly surfactants, visible particles (USP <788>), subvisible particles (USP <787>), impurities (not more than 3%, matches reference product impurities), and physical properties including pH and sterility.
In the past, several RQAs statistically evaluated for side-by-side similarity testing, such as protein content and bioactivity, led to several examples where the product met the release requirement but failed analytical similarity, particularly in which a Tier 1 approach is applied based on the mean value that allows failure when the standard deviation for reference is small, and allows passing when the variability is high. All test methods used for release attributes must be validated and all lots tested must be released.
The SQAs are the integral quality elements that are related to expression, and these require testing on a side-by-side comparison with a reference product. The primary structure must match entirely, notwithstanding any terminal amino acid difference allowed. The secondary and tertiary structure is studied using methodologies that produce output that should be superimposable, without any statistical treatment of data.
Proposed testing methods include amino acid sequencing of the primary structure through peptide mapping (liquid chromatography—mass spectrometry), peptide mass fingerprint (matrix-assisted laser desorption deionization/ionization mass spectrometry), matrix-assisted laser deionization/ionization–time-of-flight, and mass spectrometry amino acid sequencing.
Exact matches should be obtained for higher order structure conformation, determined by thermodynamic fluorescence, far- and near- ultraviolet circular dichroism, differential scanning calorimetry, nuclear magnetic resonance, and ELISA.
Binding should be assessed with cell-based assays, surface plasmon resonance, ELISA (with a match within experimental variability), forced degradation profile (with no new impurities and the same relative proportion of impurities).
All test methods should be suitable and not necessarily validated since side-by-side testing of the reference product removes bias. All of the above tests that represent the basic tests used to characterize biological drugs, can be done using a smaller number of lots.
10. Providing additional support for product developers regarding product quality and manufacturing process, including by identifying physical product quality attributes that are most critical to evaluate, and by exploring ways to reduce the number of lots of the reference product required for testing.
“Additional support” means the FDA identifying quality attributes that relate to the manufacturing process; this is a significant change in how the FDA advises developers and stands to reduce the burden on developers significantly. The FDA should also allow the use of smaller lots to develop a product and then let the sponsor file for a comparability protocol after the approval of the product. This will significantly reduce the cost and time for the development of biosimilars; given the length of time and cost, creating product performance qualification lots at scale is an expensive exercise that discourages smaller companies from developing biosimilars.
Additionally, protocols related to shipping validation, leachable and extractable testing, and warnings about shaking the product should be specified.
11. Engaging in a public dialogue through a Part 15 hearing and opening a docket to request additional information from the public on what additional policy steps the FDA should consider as we seek to enhance our biosimilar program.
The FDA has always been open and responsive to receiving public advice, and creating a docket specifically for the purpose is helpful, but some stakeholders may not be eager to let their identity be known if they are in disagreement with the agency. I am recommending that the FDA allow and encourage private submission of comments to the new OTTB office.
Additional Issues Related to the BAP
The BAP does not address a significant issue in conducting toxicology studies in animals that are more suitable for new drugs where the goal is to establish the lethal dose for 50% death. There is no scientific basis for using these studies to establish differences between a biosimilar and a reference product since the studies use a toxic dose, where the variability is high, and does not provide any useful information. The only utility of animal studies comes in conducting PK studies that may indicate any differences in the structure of the molecule. It would be appropriate for the FDA to adopt the EMA position on animal toxicology studies. However, should the FDA continue to insist on animal toxicology, then the FDA should provide more details, templates, protocol outlines, and other advice to the industry to make animal testing clinically meaningful while reducing exposure to animals.
The PK/PD profile of a biosimilar is compared with a reference product using the same model as used for generic drug approval, where we conclude that 2 treatments are not different from one another if the 90% confidence interval of the ratio of a log-transformed exposure measure (AUC and/or Cmax) falls entirely within the range of 80%-125%. Since biosimilars are administered by parenteral route, where absorption is not an issue unless the product provides a delayed absorption, the correct parameters to test are the distribution volume and elimination rate constant, both of which can be sensitive to structural differences.
The FDA may also allow the use of a highly selective population that shows less variability without affecting the robustness of the testing. While in some instances the PK/PD profiles are altered in the disease state, testing in patients does not provide any additional value since using a single lot of the same product removes structural differences. Both patients and the healthy subject respond proportionally, albeit at different levels.
I am also suggesting that the FDA encourage the use of scaled-average-bioequivalence testing protocols that may allow collection of immunogenicity profiles in a single study, reducing a significant burden on developers.
In some instances, the FDA should announce that PK/PD studies are not required, such as where the route of administration (eg ocular, otic, and possibly others) does not result in a measurable concentration of the active moiety in blood; a good example is ranibizumab administered through intraocular route.
There are serious ethical issues in testing biosimilars for immunogenicity; for products that are least immunogenic, such as filgrastim, testing is unnecessary, and where the product is highly immunogenic, such as some interferons, the testing is unethical as it sensitizes healthy subjects.
Both the FDA and the EMA are of the opinion that it is reasonable to develop qualified alternate tests.
The EMA provides the following statement regarding the use of alternate methods of testing immunogenicity:
“…ongoing consideration should be given to the use of emerging technologies (novel in silico, in vitro and in vivo models), which might be used as tools during development or for the first estimation of risk for clinical immunogenicity. In vitro assays based on innate and adaptive immune cells could be helpful in revealing cell-mediated responses.”
A significant advantage of in vitro methods is the ability to test multiple batches for immunogenicity, which is not possible in human subjects. A large number of companies are currently working on establishing immunogenicity testing, and I am suggesting that the FDA encourage the development of this technology by stating that “a clinical immunogenicity study can be waived, if the developer demonstrates a relevant comparison using other methods.”
Additional Clinical Studies
According to the FDA guidance, “as a scientific matter, a comparative clinical study will be necessary to support a demonstration of biosimilarity if there are residual uncertainties about whether there are clinically meaningful differences between the proposed and the reference products based on structural and functional characterization, animal testing, human PK and PD data, and clinical immunogenicity assessment.” Since all products licensed by the FDA to date submitted comparative clinical studies for approval, does this mean that all biosimilars approved by the FDA had ended up with “residual uncertainty” that required a comparative clinical study? If so, then were the comparative study or studies conducted sufficient to remove “residual uncertainty” even when the products were tested in only 1 indication, yet received extrapolation of all indications? In designing these comparative clinical studies, how were the equivalence or non-inferiority levels selected without having any data on the relationship between a response and effectiveness? These questions are pivotal to providing credence to clinical studies.
The FDA should identify drugs for which a PD parameter is acceptable instead of any testing in patients. Good examples are filgrastim and pegfilgrastim, in which the PD profile comparison in healthy subjects is sufficient in place of any additional clinical study. Many cytokines may qualify for a waiver of comparative efficacy studies if all else meets the requirements.