By Anna Rose Welch, Director, Cell & Gene Collaborative
There is no shortage of discussion in the biosimilar space about the regulatory landscape — in particular, where we are today, where we would like to be, and where we should be, based on the science.
As a paper recently published in BioDrugs by BioApproval’s Christopher Webster and Avalere’s Gillian Woollett and Anny Wong argued, we have had more than enough experience (and data from) approving biosimilars in the U.S., Europe, Canada, and Australia for all leading regulators to make several important executive decisions. As their paper entitled “An Efficient Development Paradigm for Biosimilars” explains, data from biosimilar approvals in the aforementioned countries show that, where a biosimilar’s analytical and pharmacokinetics (PK) profiles have been found highly similar to that of their reference product, they have all gone on to be approved following an efficacy trial. In turn, the authors propose an alternative approval paradigm called the “confirmation of sufficient likeness” (CSL), which places greater emphasis on analytical and PK data, eliminating the need for in vivo and bridging studies and large efficacy trials in most cases.
This paper ties closely into the discussion that occurred during the Biosimilar Medicines Conference in Amsterdam earlier this year. Throughout the panel — and as I wrote in the first part of this article for Life Science Leader — one pharmaceutical executive shared his thoughts on why large comparative efficacy trials are, in most cases, ill-equipped to eliminate any residual uncertainty surrounding a biosimilars’ clinical profile.
There are several data-supported arguments to push away from the status quo in biosimilar development. However, there remains hesitance to implement this practice on a large scale. Though analytical comparability has been embraced by regulators for routine manufacturing changes since 1996, there remain what regulators consider to be analytical limitations giving them pause in determining biosimilarity strictly based on analytics and PK.
WHY REGULATORS ARE HESITANT TO EMBRACE TAILORED BIOSIMILAR DEVELOPMENT
As the first article in this series showed, there were several strong opinions expressed about why the clinical trial may not be the best tool for eliminating a biosimilar’s residual uncertainty in some situations. However, for one regulator — René Anour, senior medical assessor for the Austrian Agency for Health and Food Safety and EMA Biosimilar Medicinal Products Working Party (BMWP) member — this doesn’t mean we can or should stop asking questions. There certainly are less sensitive clinical models available for certain biosimilars. But, in those cases, our long-term goal should be to ask better questions and find more appropriate methods (clinical or otherwise) of getting the necessary answers.
While in a vast majority of cases, biosimilar clinical trials and postmarketing real-world evidence studies have illustrated safety and efficacy, in turn, reinforcing the concept of “structure as function,” there were a few situations in the past year that threw the biosimilar and biologics industry for a loop. During the development of biosimilars for Herceptin, one biosimilar manufacturer discovered a shift of the critical quality attribute (CQA) antibody-dependent cell-mediated cytotoxicity (ADCC) in several lots of the originator product. This shift was caused by a routine change in the reference product’s manufacturing process. Now, on the analytical level, the biosimilars and the reference product boasted highly similar quality, preclinical, and PK data. However, in comparative trials, the two biosimilars for trastuzumab actually demonstrated superiority in treatment for breast cancer patients receiving the biosimilars over treatment for patients who received batches of the originator following the manufacturing change. (And, just as a reminder, a biosimilar can only be as good — or as bad — as its originator.)
As Anour explained, though ADCC has long been recognized as a CQA for the product with some impact on the clinical impact of the drug, regulators didn’t quite realize the extent at which it would alter clinical responses in patients. Similarly, the fact that these differences were not revealed during analytical comparability testing was a surprise, one which Anour believes is a prime example of how uncertainty can remain following analytical testing.
Niklas Ekman, a regulator from the Finnish Medicines Agency FIMEA and vice chair of the BMWP, reinforced Anour’s point on how difficult it currently is to determine the impact of a CQA on the clinical profile of a product — and, especially, to which extent a CQA can shift before it alters the clinical performance. It’s important to repeat here that ADCC was not a CQA that regulators were unfamiliar with, and it’s one that is thoroughly examined using assays during the development process. While this clinical impact was the result of an approved manufacturing change as opposed to the assays used during development, Ekman did point out that assay selection for the CQAs overall remains critical. For example, in measuring ADCC, there are a number of assays that can be used, but their sensitivity can vary drastically. Some may reveal everything, while others will reveal nothing. Hence, regulators’ ongoing push for more analytical data to help increase knowledge surrounding how molecular differences can be reflected in the clinic.
WHAT REMAINS TO BE LEARNED FROM ANALYTICAL CHARACTERIZATION
One of the areas the industry continues to learn more about is how subtle differences are reflected in the human body. For example, as the moderator of the panel, Paul Chamberlain, an advisory board member of the drug development consulting firm NDA, pointed out, we did see another instance in which the biosimilar for Enbrel (etanercept) was found to have lower levels of biomolecular weight aggregates. The patients taking the biosimilar also reported fewer injection-site reactions. Now it’s a regulator’s job to look at the differences and, based on evidence, correlate subtle molecular variations with certain outcomes. But as Ekman acknowledged, it’s a rare day when regulators can solidly link a biosimilar’s clinical difference (superiority or inferiority) to a specific molecular variability. For one, a site injection difference, as seen in Enbrel biosimilar development, may not necessarily be the work of a single CQA, but multiple. Similarly, as a biologic is a combination of quality attributes, it can be very difficult to determine how the combination of small differences in multiple CQAs will be reflected clinically.
"Though analytical comparability has been embraced by regulators for routine manufacturing changes since 1996, there remain what regulators consider to be analytical limitations."
In other cases, the ability to earn that long-sought-after tailored development also will depend on the amount of data that is submitted. There are certainly situations where more than enough data is recommended and supplied — which, many of you will likely argue, may be too much data given what we already know about certain molecules. This has been one of the bigger responses from the industry following the FDA’s recently released Comparative Analytical Assessment and Other Quality-Related Considerations guidance. But, as Anour explained, tailoring a development program should not be done just because there’s a desire to see the development process accelerated; it must be based on the evidence from molecule to molecule. Certainly, regulators are obtaining great amounts of data on many of the first biosimilar molecules — enough so that more is known about how or whether certain molecular attributes could impact clinical efficacy. But it will ultimately come down to supplying agencies with mature quality data.
“There are just some cases of biosimilar development that spread uncertainty about whether we know enough in every case to allow for reducing the amount of clinical data,” Anour said. “It might seem like we, as the regulators, are slowing the process down and not wanting to allow for tailoring. But we should not do it just because we dare to or want to, but because the decision is based on evidence.”