By Benjamin Policicchio, PhD and Suzanne M. Sensabaugh, MS, MBA

The Food and Drug Agency (FDA) released a guidance draft to aid sponsors interested in developing biosimilar products for licensure, titled: Statistical Approaches to Evaluate Analytical Similar. It describes the type of information a sponsor should obtain regarding the structural/physiochemical and functional attributes of their biosimilar product(s), how that information is used in the development of an analytical similarity assessment plan, and the statistical approaches recommended for evaluating analytical similarity.

This guidance draft complements a previously released guidance document: Quality Considerations in Demonstrating Biosimilarity of a Therapeutic Protein Product to a Reference Product. As it stands, 15 pages are split into the following sections: general principles for evaluating analytical similarity, which encompasses an analytical similarity assessment plan and statistical methods for evaluation.

A biosimilar product is described by the FDA as “biological product that is highly similar to and has no clinically meaningful differences from an existing FDA-approved reference product notwithstanding minor differences in clinically inactive components.”  Sponsors wishing to submit a biosimilar product must submit a 351(k) application, including information demonstrating biosimilarity based on data derived from “analytical studies that demonstrate that the biological product is highly similar to the reference product notwithstanding minor differences in clinically inactive components.”

The draft details the preferred methods of determining analytical similarity by first understanding the structural/physiochemical and functional attributes of the reference product and then developing an analytical similarity assessment plan. The draft first describes the “quantity and quality of both reference and biosimilar lots that we generally believe are scientifically necessary for evaluating analytical similarity,” and then describes “general principles for the evaluation of analytical similarity.”

The draft lists and describes several factors that should be considered when selecting lots:

  • Number of reference product lots – minimum 10 reference product lots should be sampled
  • Number of biosimilar product lots – minimum 10 biosimilar lots should be included
  • Variability in reference product lots – “Lots with remaining expiry spanning the reference product shelf life should be selected”
  • Accounting for reference product and biosimilar product lots – “Sponsors should account for all of the reference product lots available to them”
  • S.-licensed reference product and other comparators – “The analytical similarity acceptance criteria should be derived using data from an analysis of the U.S.-licensed reference product”
  • Biosimilar lots manufactured with different processes – Data may be combined from “proposed biosimilar product lots manufactured with different processes and/or at different scales”

The draft goes on to describe the most appropriate way to develop an analytical similarity assessment plan, which the FDA notes should be included when submitting a 351(k) biologics license application. The draft describes a list of factors to be considered when designing an analytical similarity assessment plan and should be developed in four stages. The factors include:

  • Differences in age of the lots produced at testing
  • Multiple testing results
  • Assay performance
  • Differences in attributes that will be considered acceptable

The four stages of the development of an analytical similarity assessment plan are:

  1. Development of risk ranking of attributes
  2. Determination of the statistical methods to be used
  3. Development of the statistical analysis plan
  4. Finalization of the analytical similarity assessment plan

Briefly, these stages encourage sponsors to thoroughly identify risks associated with their proposed biosimilar products; decide which statistical methods are best suited for the product based on the identified risks, split into three tiers (Tier 1 [equivalence testing] is for highest risk rankings; Tier 2 [quality ranges] is for lower risk rankings; Tier 3 [visual comparisons] is for lowest risk rankings); compile the work from the first two stages into a detailed statistical analysis plan; and produce the final analytical similarity assessment plan, which should “specify the anticipated availability of both proposed biosimilar and reference product lots for evaluation.” The draft then goes on to describe in detail the FDA’s current thinking on the statistical evaluation of analytical similarity for each tier.

The draft encourages sponsors to contact the FDA as early as possible to discuss the appropriate attributes/assays that should be evaluated in each tier. It should be noted that the final analytical similar assessment plan should be submitted to the FDA prior to initiating the final analytical assessments and that the “FDA’s final assessment as to whether a proposed biosimilar is highly similar to the reference product is made upon the totality of the evidence, rather than the passing or failing of the analytical similarity criteria.”