Sarfaraz K. Niazi, PhD, is an adjunct professor of biopharmaceutical sciences at the University of Illinois and the University of Houston, and founder of biosimilars companies Karyo Biologics and Adello Biologics. He also founded the biosimilar advisory company PharmSci.

Sarfaraz K. Niazi, PhD

Analytical similarity assessment of biosimilar candidates is the backbone for establishing biosimilarity, a primary requirement for the licensing of biosimilars. After 

 its pivotal final guideline on analytical similarity testing, the FDA issued 

 (Development of Therapeutic Protein Biosimilars: Comparative Analytical Assessment and Other Quality-Related Considerations), with many apparent and subtle differences, primarily based on my 

 to the FDA. For example, few people noticed that the FDA replaced “testing” with “assessment.” To assess is to test the value of something using various tools. Testing is when something is put through a series of trials to determine its value. Thus, 

 is one of the ways to assess something. In this paper, I will draw attention to both substantive and subtle changes to help reduce the cost and time for the approval of biosimilars.

 Orthogonal testing uses different testing methods to affirm the same attribute, such as the protein content using an ultraviolet (UV) reading and chromatography reading. The need for such testing comes when one method cannot be suitably validated. However, many reported tests have used multiple validated methods to confirm the same finding. As an example, 25 analytical tests were submitted for Pfizer’s adalimumab biosimilar, while the Boehringer-Ingelheim adalimumab biosimilar came with 

. Which company was more brilliant: the one that chose to do just 25 tests or the one that overkilled the testing? The FDA guidance stresses the use of orthogonal testing to "definitively" identify differences in product attributes.

 Expression systems are the genetically modified bacteria or Chinese hamster ovary cells that produce a therapeutic protein as a byproduct. You cannot change this with any process changes; as a result, you need not use a large number of lots to confirm. Primary, secondary, and tertiary structures fall in this category; you can use just 3 lots, but these must be assessed side by side with the reference product because you cannot rely on any legacy values (such as those given in pharmacopeia or scientific literature); most methods of assessment for structure are also not well validated, so side-by-side testing removes the constraints of the validity of the test. This stage of development can use orthogonal testing; for example, the secondary and tertiary structures are challenging to study, and innovations such as changing the temperature, osmolality, pH, or ionic strength of the solutions tested can provide greater confidence of 

. "Use of different expression systems will be evaluated on a case-by-case basis," the guidance states.

 Although the expression system is unchanging, what happens to molecules after they are expressed mainly depends on process conditions such as bioreactor operation parameters and chromatography technologies. These are often termed posttranslational modifications, such as glycosylation or impurities. Instead of justifying the differences, it is better to adjust the process to match the profile with the reference product. Know that more important is the identity of variables and not their ratio vs that of reference product variables. 

 Comparison of a biosimilar candidate and the reference product begins with choosing the properties of the reference product to match or deciding what is critical to match. When biosimilars were introduced, the selection of these attributes was a complicated exercise of weighing risk factors and figuring out how to avoid variability during manufacturing; now we know well what the CQAs are and what should not be a CQA. For example, based on the earlier FDA guidance, Sandoz, which developed the first US biosimilar (Zarxio, filgrastim), tested the protein content as a CQA and kept failing until 

, even though none of the lots were out of release specification, meaning they met standards for acceptance. Such exercises are no longer necessary; if you are able to establish release specification based on either acceptance variation or based on variation in the reference product, there is no need to test these attributes in an elaborate side-by-side comparison exercise.

 The first final guidance on analytical testing was rife with mistakes relating to the statistical modeling of CQAs. Without any rationale, the FDA had created a 3-tier system. The most rigid was Tier 1, where the 90% confidence interval of the biosimilar product had to be within 1.5 of the standard deviation (SD) of the reference product. It turned out that most CQAs assessed failed. The new guidance removes statistical modeling but advises that enough lots should be used in a side-by-side comparison if SD is used to measure variability. Likely, the FDA will allow pooling analytical assessment data as they did to evaluate the Sandoz filing.

 Both FDA and the European Medicines Agency have experience evaluating submissions of dozens of biosimilar products, and the initial emphasis on “more is better” is fading away. Still, the need for a “stepwise approach” to biosimilar development remains. Generally, if fewer lots are used for comparison, there is a higher chance of a biosimilar failing (remember Sandoz did 71 lots to pass the protein content test). Matching with fewer lots should be preferred unless the CQA is highly variable. Most traditional CQAs are not highly variable. Testing more than necessary not only adds to the cost of development but also extends the timeline for submission. The goal is not to cut the corners but to go around the corners where appropriate.

We have come a long way from the early days of superfluous testing of biosimilars; the new FDA guidance should be read carefully between lines; it is a more rational approach to achieving the demonstration of biosimilarity, but the FDA is like any other agency—if you keep offering to assess ad infinitum, they will have no objection to it. Manufacturers must learn the difference between “testing” and “assessment.”

Articles

Refinements to the analytical testing of biosimilar candidates to verify similarity were worked into 2019 guidance from the FDA. These changes offer what the author considers a more "rational approach." 

Opinion: FDA Changes to Biosimilar Development Guideline Save Money and Time

In the United States, the use of animals to test human pharmaceuticals dates to 1937, when a liquid formulation of a sulfa antibiotic dissolved in diethylene glycol resulted in the deaths of 107 adults and children. The incident led to the 1938 

, mandating animal toxicity testing. In 1946, language was incorporated into what became the 

designed and based on the results of animal experimentation

 [author’s italics] and a knowledge of the natural history of the disease.” 

The previous statement was written by Andrew Ivy, a strong proponent of animal research. Still, it was 

 such a requirement would improve the safety or efficacy of human drug development. Today, the FDA generally 

 preclinical testing of any new drug or biological therapeutic in animals “for pharmacologic activity and acute toxicity” before human clinical trials begin. However, recently, the FDA indicated it was open to considering 

A weight-of-evidence approach is practiced for biotechnology-derived products when a carcinogenicity assessment is warranted. Genotoxicity [damage to DNA or chromosomes] testing is generally inappropriate for biological products and is not part of the assessment. When sufficient information is available, the need to conduct additional nonclinical studies, regardless of whether the biologic is active and testable in a rodent bioassay, is not warranted. If the assessment does not suggest carcinogenic potential, a rodent bioassay is considered unlikely to add value, and no additional nonclinical testing of biologic products is recommended. Prior knowledge of target-based risks figures prominently in developing a strategy to assess the carcinogenic risk of each new biologic because the nonspecific activity of such molecules is generally considered to be low.

 exploring the use of NAMs to improve regulatory efficiency and potentially expedite drug development. All of these discussions apply to new drugs or biological ones. No animal studies are required for generic chemical drugs because their chemical structures are identical to the reference products’ structures. Biosimilars have presented a dilemma; their molecular structure may be highly similar but not identical to the reference products’ makeup, which actually varies from batch to batch. The 

Biologics Price Competition and Innovation Act

 (BPCIA) states that the data required for FDA approval 

include animal studies (including an assess­ment of toxicity). This requirement of the BPCIA has caused significant confusion and resulted in delays in the approval of biosimilars.

In most cases a correlation between animal toxicology and safe use of biological products in humans cannot be made because of the differences in the receptors between humans and animals. This is well established.

Biosimilar developers, with a mindset of new drug development, took the toxicology studies to a different level, submitting studies that were refused by the FDA. The worst examples were (number rejected/number submitted): 

 (pegfilgrastim; 8/13). For example, for Mvasi (bevacizumab), mice and rat studies were reported to demonstrate that the high mannose variants in the biosimilar candidate were not meaningful—an irrelevant exercise because rodents do not have receptors to bind monoclonal antibodies. 

The same is true about testing trastuzumab in mice (Trazimera and Herzuma), where trastuzumab does not identify the neu receptor. For the etanercept biosimilars Eticovo and Erelzi, studies were conducted in mice and rats that were irrelevant for the same reason. Fifteen animal toxicology studies were reported for an epoetin biosimilar (Retacrit) to justify higher antidrug antibodies than the reference product produced—an irrational approach. It was later determined that the differences were due to formulation and route of administration. 

Testing biosimilars in species that do not present receptors is irrelevant, as the European Medicines Agency and the FDA have repeatedly pointed out. Genotoxicity testing is disallowed because it is irrelevant. However, the difference between how a chemical species interacts with the body vs biological molecules remains included in the approval guidelines. A reference biological product that has undergone extensive multispecies testing has established safety in humans; when it comes to biosimilars, analytical similarity alone is enough to demonstrate that there is no risk of unexpected response.

The most misleading advice comes from the WHO; it states that toxicology studies in monkeys should be avoided, animal toxicology studies should be conducted at several different doses, and immunogenicity testing in animals should be evaluated—all of which are irrational suggestions; unfortunately, many countries follow the WHO guidance, such as India, leading to inappropriate approval of biosimilars. The following points are important to keep in mind:

Unlike chemical drugs, biological drugs do not elicit a pharmacologic response across species and there is no extrapolation to human safety.

The new FDA advice discourages animal testing of new biologicals.

Animal immunogenicity testing is irrelevant; the human immune response has a different infrastructure than for any type of animal species.

No animal testing can be used to justify any differences in analytical similarity; for example, an unidentified and unreported impurity in a biosimilar candidate vs a reference product cannot be proven irrelevant by any animal testing.

Multiple species testing, as reported, is of little value for the same arguments.

Differences in antidrug antibodies in humans can never be resolved with any testing protocol in animals.

In summary, a detailed review of over 100 biosimilars approved in the United States and European Union by the author reveals that none of the animal toxicology studies was pivotal in the approval of the product; essentially, there was sufficient evidence without them for a regulatory decision. Now that the agencies have 2 decades of experience in approving biosimilars, it is about time that clear guidelines against animal toxicology testing are adopted. The advantages of this guidance will include:

Developers will be forced to produce a more analytically similar product because they cannot rely on animal testing to justify any significant differences in the structure.

It could prevent the approval of biosimilars that may not be adequately confirmed to be biosimilar based on irrelevant animal toxicology testing.

There would be a significant reduction in the testing time and cost—both relevant to making biosimilars affordable.

The only justified use of animal studies is in animal pharmacology testing in larger species such as monkeys. The pharmacokinetic profile provides an additional element of analytical similarity—how the body views the molecule, which is the same justification for clinical pharmacology testing in humans.

Animal toxicology studies still get included with biosimilar approval applications, although there are strong arguments that most of these studies are unnecessary. Sarfaraz K. Niazi, PhD, explains the pros and cons. 

Opinion: The Debate Over Animal Toxicology Studies 

My analysis of all biosimilars licensed by the FDA and EMA shows a common trend: Except in a few instances, all products were tested for clinical efficacy and safety; none of the tests failed, and where results were borderline, the FDA allowed adjustment of acceptance criteria to make approval possible. No product was tested in more than 1 indication, despite the allowance of all indications by the FDA and European Medicines Agency (EMA).

If you are one of the big pharma companies dabbling in biosimilars, you would prefer to conduct these studies because you are used to meeting with prescribers with published clinical studies in hand. Other companies, not as well grounded in the value of clinical studies, performed them because they kept thinking of biosimilars as new biologics and had great uncertainty in their abilities. This article is not for those who are fixed in their belief that clinical efficacy testing is needed. Here are a few points of advice for biosimilar developers. First, I continue to 

 that big pharma will soon walk out of the biosimilars field; after my first prediction, I got a lot of wrath from big pharma, although several pulled out within a few months.

One should learn from history: The same things occurred when generic laws came into effect. First, big pharma predicted that use of generics would lead to the demise of millions of patients because only big pharma knows how to make safe products. That campaign was not successful, so they joined the party and began offering generic versions of their products (some still do). But when the bottom fell out with 70% to 90% price reductions, most big pharma companies went back to doing what they best, creating new products. The same will happen to biosimilars, once the price bottom falls.

It may be a surprise to some that clinical efficacy testing for biosimilars is not always a requirement for approval. When I reviewed all licensed products, I inquired why the FDA required these studies; to my surprise, the answer was that the developers offered these studies as part of establishing biosimilarity. If you provide the FDA or EMA with an extensive plan to test clinical efficacy and safety, the agency cannot say no. What if they discouraged these studies and there were safety issues?

The most glaring example of redundant efficacy testing was the approval of Retacrit (epoetin alfa) by Hospira, which conducted efficacy testing in 1000 individuals for its European submission. When Hospira applied for FDA approval using the same EU data, the application was rejected and the FDA required a stepwise approach which involves analytical characterization of the proposed biosimilar, then animal studies, possibly, and then comparative clinical studies. Instead of examining the extensive clinical efficacy proof, the FDA did not require any efficacy testing and accepted the EU study only as an ancillary study. The FDA and EMA are also not requiring clinical efficacy testing for products such as teriparatide, insulins, low-molecular-weight heparins, filgrastim, pegfilgrastim, and other cytokines with pharmacodynamic markers. Some products would not even need safety assessments, such as teriparatide, low-molecular-weight heparin, and insulins.

Here is how biosimilar developers should challenge the FDA if asked to do clinical efficacy testing:

Force the FDA to explain, in writing, where they see any “residual uncertainty,” a prerequisite before the agency will ask for “additional clinical testing.”

If the agency insists, counter that this study need not be in patients, as the Biologics Price Competition and Innovation Act does not require patient testing.

Suppose there are safety issues (such as variability in the posttranslation modifications). Developers could offer to conduct a limited safety study in healthy subjects.

If there is still no concurrence, then it could be argued that conducting efficacy testing in 1 indication (which is currently required and practiced) does not assure efficacy and safety assurance in other indications; therefore, the testing would be limited and wasteful.

A final argument should be to agree to a postmarketing study in an open protocol for a limited time.

In my recent interactions with both the FDA and EMA, there is an openness to receive creative approaches to establishing biosimilarity. We have more than 15 years of experience in seeing the approval of over 100 biosimilar products.

The reason I am insistent on helping biosimilar developers simplify the qualification process is that, so far, the approval of biosimilars has been disappointing; first, markets for biologic drugs are still very much controlled by big pharma; and second, this is keeping smaller companies from entering the field. The reason I am so confident is that I know exactly what it costs to manufacture biological products. Biosimilar antibodies should not cost more than $150/g to manufacture in Western countries. But, not surprisingly, the cost to patients is still about 100-fold; this must come down if the promise of biosimilars will be realized.

Clinical efficacy studies are not always required for biosimilars, and it's possible to resist if the FDA does ask for them. Sarfaraz K. Niazi, PhD, suggests a game plan. 

Opinion: How to Get FDA Clinical Efficacy Testing Waivers for Biosimilars

This year marks 12 years since the appearance of the Biologics Price Competition and Innovation Act of 2009 (BPCIA); and, the FDA has approved 30 biosimilar products comprising 10 molecules. A critical analysis of the licensed biosimilars 

 a dire need to review the BPCIA pathway to find ways to go around it to expedite the entry of more biosimilars. First, let me share with you a few facts about the licensed biosimilars. Except for 2 biosimilar products, clinical efficacy testing was done for all biosimilar products although they met all other requirements of biosimilarity. None of the products that were tested failed. So, a question arises: What would it be like if these applications were filed under 351(a) for standalone biological products instead of under 351(k)? For example, Teva famously filed for FDA approval of its filgrastim product (Granix) as a standalone biologic. Did it make a difference? No. The biggest problem with BPCIA is the 12-year exclusivity of the reference product. Now that we have reached the 12th anniversary of the BPCIA, let us examine how we can go around the hurdles inherent in the BPCIA.

Using 351(a) for Copies of Approved Drugs

Although the 351(a) pathway is intended for new molecules, it is also available for copies of approved molecules and to add new indications (uses); in a way, it resembles the 505(b)(2), which is a streamlined product approval filing for new chemical generic drugs. The FDA has stated clearly in its guidance for COVID-19 product candidates that, under 351(a), the studies required for product approval may be reduced if a similar product has been approved, essentially inviting the developers to challenge the concessions available for biosimilars. Although there is no reference product required for the development of a copy biologic under 351(a), a dataset that is created comparing the product candidate with a reference product will be accepted as proof of comparable safety and efficacy.

We could unofficially call such imitator products “biosimilars.” There would be no 12-year exclusivity to worry about because originator products would not be involved; there would be no marginal failures in similarity testing and no discussions of minor differences in posttranslational modifications. There would be no patent dance and necessary litigation because the onus of proving intellectual property infringement would lie with the claimant. Manufacturers would have to choose just 1 indication for the purposes of efficacy testing—the same way it is done for biosimilars now. For example, manufacturers could select psoriasis for an adalimumab (Humira) imitation product, and this would be a much cheaper study to conduct. Once approved, prescribers and payers would be able to use the product off-label and compete with Humira for all of its approved indications. I am not recommending breaking any laws; I am presenting facts. For example, bevacizumab (Avastin) is used for age-related macular degeneration as an intravitreal injection, but no bevacizumab products are specifically approved for this use. Does it make a difference if one uses bevacizumab or biologically equivalent products for this purpose? Notwithstanding any special formulation factors, it should be OK.

The cost and time for developing products in this new pathway I am suggesting—a modified 351(a)—would not be higher than for a 351(k) application, but with benefits of getting around the 12-year exclusivity automatically granted to originator products when the filing is made under 351(k)—a big game-changer. Incidentally, when the BPCIA was drafted, I had the privilege of explaining to then–president-elect Barack Obama all about biosimilars. He was opposed to the 12-year exclusivity for originator products, and preferred 8 years at most, although he yielded under opposition from big pharma. This issue becomes more acute if we talk about the creation of biosimilar mRNA vaccines; 12 years is too long a period of exclusivity for the originator products because the vaccine will likely change during that time. 

In this column, I am using the term “biosimilar,” and this needs explanation. Biosimilar means a biologically similar product, not a similar product that is biologic. So, we can use the term “biosimilar” to characterize a product approved under 351(a). It need not be essentially associated with BPCIA. However, if you get approval for a standalone product, it will not be eligible for interchangeable status, as was recently awarded to Semglee (insulin glargine). This alternative approval should not be considered as a path opening for interchangeability.

To help my understanding of these issues, I communicated with the FDA about its guidance for COVID-19 products, which reads as follows:

COVID-19 vaccine development may be accelerated based on knowledge gained from similar products manufactured with the same well-characterized platform technology, to the extent legally and scientifically permissible. Similarly, with appropriate justification, some aspects of manufacture and control may be based on the vaccine platform, and in some instances, reduce the need for product-specific data. Therefore, FDA recommends that vaccine manufacturers engage in early communications with [Office of Vaccines Research and Review] to discuss the type and extent of chemistry, manufacturing, and control information needed for development and licensure of their COVID-19 vaccine.

My question to the FDA was, would this advice also apply to other biologics? The answer was, “File a formal meeting request Type B.” This is the FDA asking us to challenge them. I am confident that we can convince the FDA not to require many studies, such as toxicology testing, if we can demonstrate analytical similarity. The pharmacokinetic/pharmacodynamic (PK/PD) studies will be inevitable. We can present the data collected in a comparative study as standalone data to calculate the dose, which we already know, and then demonstrate the PK/PD behavior as additional proof of similarity. A PD response mainly for cytokines where PD parameters are available can be used in lieu of any efficacy testing. Choosing the most straightforward protocol will do; it will not be an equivalence study, nor would it be a noninferiority study, but it can be conducted side-by-side as a monitor arm. The statistical model is relatively simple but requires a bit of creativity. The fact that you would not be able to get interchangeable status is not an issue.

The most significant advantage with the modified pathway I suggest under 351(a) is that you could get into the market at least 5 to 8 years earlier than if you were to enter via the 351(k) pathway; for some products, the 351(k) official biosimilars pathway would be futile if the reference product has changed—a tactic widely adopted by the innovators. You will be able to meet with the FDA in a Type B meeting that is free, and there will be no commitment to paying the yearly BPCIA fee to begin talking to the FDA. The high burden of conducting similarity studies, many of which make no sense, will be reduced. As an example, the biosimilar product may meet all release requirements such as content and potency. However, the product may still fail in comparative testing if the innovator product has lesser variability—which is inconsequential anyway. In short, your biosimilar candidate may be good enough for the patients, but not the FDA.

To recap, I suggest biosimilar developers consider using a modified 351(a) approach for products when establishing biosimilarity is difficult and where there is a need to avoid getting bound by the 12-year exclusivity. The fact that an efficacy and safety study was done for every biosimilar that has been licensed by the FDA (with a couple exceptions) makes the advantages of filing a biologics license application under 351(k) less attractive.

Use of the 351(k) biosimilar approval pathway entails legal challenges, high costs, and potentially higher clinical evidence standards. Sarfaraz K. Niazi, PhD, suggests that using the 351(a) pathway for standalone drugs and copy products may be faster and better. This is part 3 of a series.

Sarfaraz K. Niazi, PhD

Opinion: FDA Changes to Biosimilar Development Guideline Save Money and Time

Opinion: The Debate Over Animal Toxicology Studies

Opinion: How to Get FDA Clinical Efficacy Testing Waivers for Biosimilars

Opinion: A Modified 351(a) Licensing Pathway for Biosimilars