FDA's Approach to Regulating Biosimilars

The biosimilar pathway, enacted under the Biologics Price Competition and Innovation (BPCI) Act on March 23, 2010, granted the FDA the authority to approve a biological product under section 351(k) of the Public Health Service (PHS) Act that has been demonstrated to be biosimilar to or interchangeable with a reference product. A reference product is a biological product licensed in the United States under section 351(a) of the Public Health Service Act (1).

Biosimilarity is defined to mean that “the biological product is highly similar to reference product notwithstanding minor differences in clinically inactive components,” and “there are no clinically meaningful differences between the biological product and the reference product in terms of the safety, purity, and potency of the product” (1, 2).

For the FDA to designate a biological product as interchangeable, a provision that allows for substitution of the reference product with the interchangeable product without the intervention of a health care provider, a product must be biosimilar; a product must be expected to produce the same clinical result as the reference product in any given patient; and for a product that is given to a patient more than once, the risk in terms of safety or diminished efficacy of alternating or switching between the interchangeable product and the reference product is not greater than the risk of using the reference product without alternating or switching (1). This brief review will focus on the scientific considerations regarding the demonstration of biosimilarity. A discussion regarding interchangeability is beyond the scope of this review.

To support approval of a reference product, an applicant (or sponsor) must submit a full complement of data in the marketing application [i.e., a Biologics License Application (BLA)] to support the safety and effectiveness of the product. In contrast, in addition to submitting data and information to support a demonstration of biosimilarity to the reference product, the BPCI Act requires biosimilar sponsors to include publicly available information in their BLA regarding the FDA's prior determination that the reference product is safe, pure, and potent (i.e., safe and effective; refs. 1, 2). In addition, biosimilar sponsors, with adequate scientific justification, can extrapolate data supporting biosimilarity in one indication to other indications and seek licensure of the proposed biosimilar product for conditions of use (e.g., indications) previously approved for the reference product that were not directly studied in the biosimilar development program (1–3). As such, the biosimilar pathway is an abbreviated licensure pathway, where the goal of development is to demonstrate biosimilarity between the biosimilar and reference product, not to independently establish the safety and effectiveness of the biosimilar product (2). The goals of the BPCI Act are similar, in concept, to goals of the Drug Price Competition and Patent Term Restoration Act of 1984 (“Hatch-Waxman Act”; ref. 4). The Hatch-Waxman Act created abbreviated pathways for the approval of drug products, including nonidentical drug products approved under the 505(b)(2) pathway, under the Federal Food, Drug, and Cosmetic Act (4). The BPCI Act is consistent with the FDA's longstanding policy of permitting, when appropriate, reliance on known information about a drug and avoiding unnecessary duplication of human or animal testing (4).

The FDA uses a totality-of-the-evidence approach to make a determination of whether a product is biosimilar to a reference product (2). The BPCI Act requires that a biosimilar application contain information from (i) analytic studies demonstrating that the product is highly similar; (ii) animal studies, including an assessment of toxicity; and (iii) a clinical study or studies, including an assessment of immunogenicity and pharmacokinetics or pharmacodynamics, in one or more appropriate conditions of use unless the FDA determines that one or more of these elements is unnecessary (1). The FDA recommends that sponsors use a stepwise approach to develop the evidence needed to demonstrate biosimilarity (2). For example, after beginning development by conducting extensive comparative structural and functional characterization of the proposed biosimilar product and the reference product, sponsors should assess where, if any, residual uncertainty exists regarding demonstrating biosimilarity of the proposed product and determine what additional data (e.g., functional, clinical) are necessary to address these uncertainties and to establish that there are “no clinically meaningful differences” (2).

The foundation of a biosimilar development program is the analytic similarity assessment conducted by a sponsor to support a demonstration that the proposed biosimilar product is highly similar to the reference product (2). This assessment will include data from analytic studies that directly compare the structural/physiochemical and functional properties of the proposed biosimilar to the U.S.-licensed reference product (5).

Unlike small-molecule drugs, biological products are complex mixtures, where the individual proteins display inherent variability as a result of being made in living systems; therefore, biological products contain minor differences within and between manufactured lots (5, 6). For this reason, the scientific and regulatory standard for establishing analytic similarity for a biosimilar licensing application is “highly similar” and not “identical.” Biological products may also minimally change over time due to manufacturing process changes postlicensing that do not alter their safety, purity, or potency (7, 8). For example, since 1996, the FDA has approved many manufacturing process changes for licensed biological products based on product quality and analytic comparability data obtained before and after the process changes without the need for additional nonclinical or clinical studies (5). Although microheterogeneity exists for biological products, product quality is controlled within predefined acceptance parameters (i.e., release specifications), and changes to manufacturing processes are reviewed by the FDA to ensure that the safety, purity, and potency of the biological product are maintained (5, 8). As such, it would not be scientifically appropriate to hold a biosimilar product to an “identical” requirement, given that sponsors of “stand-alone” biological products, such as reference products, are not held to such a requirement. Nevertheless, it is the responsibility of sponsors of biosimilar products to demonstrate that differences in product characteristics between the proposed biosimilar and reference product do not impact clinical performance, for example, the product's clinical activity, pharmacokinetics/pharmacodynamics, safety, or immunogenicity.

During development, sponsors will assess for differences in structural/physicochemical and functional product quality attributes. Examples of structural or physicochemical assessments include primary structure, higher order structures, enzymatic posttranslational modifications (e.g., glycosylation patterns), other structural variants (e.g., deamidation or charge variants), and any intentional modifications (e.g., PEGylation; ref. 2). Functional assays are essential to confirm the higher order structure and specific activity(ies) of the product (e.g., potency). The type of functional assays used will depend on the type of product and could include in vitro cell-based bioactivity assays, such as antibody-dependent cell-mediated cytotoxicity, binding assays, or assays of enzyme kinetics (2, 5). One way to reduce uncertainty in the characterization of attributes is to use more than one state-of-the-art orthogonal method to assess the same attribute (5). The FDA also recommends that sponsors, where applicable, develop qualitative methods to assess product quality attributes.

Sponsors should address the potential clinical impact of observed differences between products and demonstrate or justify that the differences will not impact activity, pharmacokinetics/pharmacodynamics, safety, or immunogenicity (5). For example, differences could be in clinically inactive components of the product, such as excipients used in the drug product formulation. However, if there are differences that contribute to residual uncertainty about whether the products are highly similar, then the sponsor should consider addressing the uncertainty through additional studies (e.g., analytic, animal, clinical). It is also possible that differences could lead to a determination that products are not highly similar and preclude a demonstration of biosimilarity.

The BPCI Act states that an application for a biosimilar product must include information from “a clinical study or studies (including an assessment of immunogenicity and pharmacokinetics or pharmacodynamics) that are sufficient to demonstrate safety, purity, and potency in one or more conditions of use for which the reference product is licensed and for which licensure is sought for the biosimilar product” (1). Because the biosimilar pathway is an abbreviated licensure pathway where the goal of development is to establish biosimilarity between the proposed biosimilar and the reference product and not to independently establish the safety and effectiveness of the proposed product, sponsors should design clinical studies to assess for clinically meaningful differences, should they exist (2). These studies should take into account residual uncertainties from the analytic similarity assessment and, where applicable, data from animal studies (2).

Prior to initiating clinical studies with a proposed biosimilar, sponsors must provide sufficient analytic similarity data to support safe use of the product in human subjects. The FDA recommends a risk-based approach to determining the necessity of animal studies to support initiating a clinical study for a biosimilar development program. If there is insufficient analytic data in the investigational new drug (IND) application to support reliance on the known safety profile of the reference product, thereby leading to uncertainty regarding the risk of the product, then toxicity data from an animal study might provide supportive evidence that the proposed clinical study is considered “safe to proceed” (2).

In general, the FDA expects a BLA submitted under 351(k) of the PHS Act for a biosimilar product to include data from a clinical pharmacology study demonstrating pharmacokinetic similarity and, where applicable, pharmacodynamic similarity between the proposed biosimilar product and the reference product (9). Such studies may represent the most sensitive setting to assess for clinically meaningful differences. Sponsors should conduct the study in an adequately sensitive population to detect differences unless justification can be provided for the selection of a different population, for example, safety concerns in a healthy volunteer population (2, 9). Sponsors should also consider the following elements when designing the pharmacokinetic/pharmacodynamic study: cross-over versus parallel design, study population (e.g., healthy volunteers vs. patients), dose (e.g., a dose lower than the approved dose may be more sensitive to detect potential differences in pharmacokinetics), route of administration, and specific pharmacokinetic/pharmacodynamic assessments (9).

If necessary, a comparative clinical study in a patient population can further reduce residual uncertainty in regards to whether clinically meaningful differences exist between a proposed biosimilar product and the reference product (2). Sponsors will also be expected to conduct an adequate assessment of immunogenicity in their biosimilar development program as part of the clinical evaluation (2). Although a detailed discussion of immunogenicity is beyond the scope of this review, sponsors should consider the incidence rate of anti-product antibodies and neutralizing antibodies of the reference product, observed or anticipated clinical effects of immunogenicity, and the performance characteristics of the assay to determine the scope of the assessment of immunogenicity (2).

Because comparative clinical studies should be designed to assess for no clinically meaningful differences and not to independently establish safety and efficacy of the proposed product, endpoints for comparative clinical studies may differ from the endpoints used to establish the effectiveness of the reference product (2). Certain endpoints, such as pharmacodynamic measures, that are more sensitive to detect potential differences between the proposed biosimilar and the reference product than the primary endpoints used to demonstrate efficacy in the reference product's clinical trials, may enable more precise comparisons of relevant therapeutic effects between products (2). A comparison of multiple endpoints in a comparative clinical study may also reduce residual uncertainty in regards to whether clinically meaningful differences exist. For example, comparative results of response rates (or other appropriate endpoint), cardinal safety findings (e.g., hypertension with an anti-VEGF antibody), and other pharmacodynamic markers could increase confidence in a finding of no clinically meaningful differences (10). Sponsors should provide a justification for which endpoints are formally tested statistically, acknowledging concerns regarding multiplicity when testing multiple endpoints (2). Identification of cardinal safety findings will generally be more important to the assessment of no clinically meaningful differences than an overall assessment of each adverse event in a clinical study, especially in a clinical setting such as advanced cancer with a high background rate of adverse events due to the disease or concomitant therapies (10).

In general, comparative clinical studies in a biosimilar development program will use a two-sided statistical test based on a prespecified similarity margin, because the biosimilar product should have no clinically meaningful differences, that is, it is neither better nor worse than the reference product (2). FDA guidance describes scenarios where a sponsor could propose a study with asymmetric two-sided tests or noninferiority designs (2). Sponsors should provide adequate justification and discuss such designs with the FDA prior to conducting the comparative clinical study (2).

Prior to conducting a comparative clinical study, sponsors should also consider whether the studied condition of use is adequately sensitive to detect differences to support extrapolation of data to other conditions of use (see below; ref. 2). For some products, there may be more than one condition of use that is adequately sensitive to support extrapolation.

The biosimilar pathway is an abbreviated licensure pathway, and the concept of extrapolation is fundamental to facilitating the intent and goals of the legislation. Without extrapolation, development costs would be prohibitive for some products, and multiple large comparative studies could divert thousands of patients from clinical trials of newer drugs intended to prolong the lives of or cure patients with cancer (10).

If a product meets the statutory requirements for licensure (i.e., approval) as a biosimilar product based on, among other things, data derived from a clinical study in an appropriate condition of use (i.e., indication), the sponsor may seek licensure for one or more additional conditions of use for which the reference product is licensed (2). For example, if a reference product is approved to treat different tumor types, or in different lines of therapy (e.g., metastatic or adjuvant), and the biosimilar sponsor conducted a comparative clinical study in one of these settings, then it is possible for a sponsor to extrapolate the demonstration of biosimilarity from the studied indications to other indications for which licensure is sought and for which the reference product is licensed. To support extrapolation of data for additional conditions of use for a biosimilar product, sponsors should provide scientific justification that addresses, for example, how the mechanism of action of the drug relates to each condition of use, whether pharmacokinetics and/or pharmacodynamics of the drug may vary in different patient populations, differences in expected adverse reaction profiles in different patient populations, and any other factor that may affect safety or efficacy in the different patient populations (2, 3). One example of information that sponsors could include when providing justification pertaining to the mechanism of action is a comparison of whether the known functional activity or activities of the product are the same or different in each condition of use, and supportive data and information as to why any differences would not impact the safety or efficacy of the product in the extrapolated condition(s) of use (for additional information on extrapolation see Q&A I.11 in the FDA guidance for industry, “Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation Act of 2009”; ref. 3).

The FDA will make decisions regarding extrapolation based on the totality of the evidence submitted in a 351(k) BLA for a biosimilar (2). Analytic and clinical pharmacokinetic/pharmacodynamic considerations will be especially important in the assessment regarding extrapolation.

The FDA's approach to the review of biosimilars involves an assessment of the totality of the evidence provided by a sponsor to determine that a biosimilar product satisfies the requirements that the biological product is highly similar to the reference product, and there are no clinically meaningful differences. The foundation for the evaluation of biosimilarity, including extrapolation, rests on the analytic similarity assessment. Improvements in analytic science may further enable a sponsor to characterize similarities and detect differences, should they exist, between products and allow for a more targeted approach to clinical development (2, 5).

The concepts underlying the FDA's determination of biosimilarity are the same regardless of whether a product is a supportive care product or a therapeutic product. Nevertheless, the larger size and increased complexity of many therapeutic products (e.g., mAbs), as compared with some supportive care drugs in oncology (e.g., growth factors), may necessitate more extensive analytic testing to analyze quality attributes. Furthermore, the FDA's assessment of residual uncertainty based on product complexity and any associated limits of analytic testing that would help to discern some of this complexity could influence the need for additional clinical data. Conversely, the risk of serious health consequences for anti-product antibodies may be increased for some supportive care drugs (e.g., erythropoietin). The assessment of such risks will need to be addressed in the sponsor's biosimilar development program.

Ultimately, for marketing approval, the FDA will generally expect an extensive side-by-side analytic characterization of the biosimilar and reference product, as well as comparative clinical data. The comparative clinical data will generally include, at minimum, comparative assessments of pharmacokinetics and/or pharmacodynamics and an assessment of immunogenicity. Other clinical data (e.g., safety, activity, efficacy) may also be necessary to support a demonstration of no clinically meaningful differences between products.

Development of a biosimilar product is a complex endeavor necessitating thoughtful and careful design and execution of an extensive analytic similarity assessment; one or more focused clinical studies in an adequately sensitive population; and, if needed, animal studies. As more biosimilar products gain FDA approval, health care providers and patients will gain greater access to treatment options without compromising safety or effectiveness, and that may lead to less expensive alternatives to comparable products, thus decreasing costs for patients and the health care system in the United States.

No potential conflicts of interest were disclosed.

Conception and design: S.J. Lemery, R. Pazdur

Development of methodology: R. Pazdur

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Pazdur

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Keegan, R. Pazdur

Writing, review, and/or revision of the manuscript: S.J. Lemery, M.S. Ricci, P. Keegan, A.E. McKee, R. Pazdur

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Pazdur

Study supervision: R. Pazdur