FDA Draft Guidance On CES Signals More Reliance On Toxicity And PK

A conversation with Eva Temkin, Arnold & Porter

The FDA recently published draft guidance revising its position on comparative efficacy studies (CES) for biosimilars, shifting its focus to other evidence of biosimilarity — namely, toxicity study results and pharmacokinetics (PK) and pharmacodynamics (PD).

The move comes as technology and reliability improve for these other approaches and has significant implications since CES are one of the most costly and time-consuming parts of getting biosimilars approved. Meanwhile, FDA reserves the ability to request a CES, leaving open the door that biosimilar developers may still need to take on the burden.

The four-page draft was not immediately open for public comments because of the federal government shutdown. The FDA says it will publish notice of availability in the Federal Register when the public comment period opens.

The draft guidance also comes at a time when the world is reimagining how biosimilar drugs get approved and follows a reflection paper published earlier this year.

We had some questions about the origins of the draft and what biosimilar developers should still expect when crafting regulatory strategies. Eva Temkin, a partner at Arnold & Porter and former acting policy director in the FDA’s Office of Therapeutic Biologics and Biosimilars, agreed to help us. Here’s what she said.

Is the CES draft guidance a part of the FDA’s efforts to shorten approval routes for complex drugs or, more simply, an acknowledgement that the CES is a poor yardstick for measuring comparability?

Temkin: More the latter, partly because biosimilars, in my mind — and I think in a way that FDA reviews them — are not particularly complex. All biologics are complicated relative to small molecule drugs, but biosimilars generally demonstrate biosimilarity to molecules that the agency has some familiarity and comfort with.

All the monoclonal antibody or insulin biosimilars that you've seen, those are generally well-characterized molecules that FDA is comfortable and familiar with. And so, to me, it isn't really in the bucket of “Let's super speed highly innovative new technologies to market,” as much as it is “Let's speed up the process for developing and approving biosimilars.” I think this is how the agency has postured it — it’s more about the administration’s priorities around drug pricing than on timeliness of review and approval.

So, by shortening the time window for development, we bring down the cost of bringing a biosimilar to market. They’re thinking about driving down the cost and improving access.

Temkin: That's certainly what the press that FDA put out around the guidance suggested and what FDA Commissioner Marty Makary spoke about in the time around the draft guidance release.

Earlier this year, the EMA published a reflection paper in the same spirit as this draft guidance. Do you think that’s a coincidence?

Temkin: It's a couple of things. One, FDA always gets pressure to keep up with the EU on these products. When I was at FDA, I was the acting policy staff director in the Office of Therapeutic Biologics and Biosimilars, which oversaw biosimilars policy, among other things. We were constantly questioned publicly about why there were more approved biosimilars in the EU than in the U.S.

So there's a little bit of a feet-to-the-fire sensation, but maybe more importantly than that, the draft guidance seems to reflect where the science is going. As the analytical tools that we have available and the technology for developing comparative analytics has evolved— and this is reflected in the guidance as well — the agency starts to view the analytics as more precise than the data from the clinical comparative efficacy studies.

It puts more of the decision-making emphasis on the analytics and less on the comparative clinical efficacy data.

On that note, what data are still required for the 351(k) BLA? You obviously still need to submit a data package.

Temkin: For sure. You're still submitting BLA with everything that a BLA involves.

The Biologics Price Competition and Innovation Act, which is the piece of the statute that added 351(k) to the Public Health Service Act, is quite prescriptive in how it approaches the types of data that a 351(k) BLA must include.

It’s actually spelled out in section 351(k)(2) of the Public Health Service Act. There has to be comparative analytical data, and that generally goes to the demonstration of high similarity.

Backing way up, biosimilarity has essentially two components in the definition. There's high similarity, and then there are no clinically meaningful differences. And those are shorthand for the statutory language.  High similarity and no clinically meaningful differences are, generally speaking, the two prongs of the definition of biosimilarity. Historically, the data in the comparative analytical assessment has gone to support FDA's finding of high similarity and the data from comparative clinical efficacy studies or other efficacy studies like pharmacokinetics have gone to support the finding of no clinically meaningful differences.

And what you're seeing, I think, is this trend, a shift away from that bifurcation to more of the standard in which the demonstration of biosimilarity is carried by the analytical data.

It would be very surprising to me to see a move away from comparative analytical assessments, which are really foundational. The other requirement is an assessment of toxicity. You're always going to see a toxicity assessment and then also data from a clinical study or studies, which can be the assessment of immunogenicity and PK, or in some circumstances PD. FDA has the authority to waive these requirements. But I think, generally, these pieces are expected.

Is there any precedent of FDA waiving them?

Temkin: They used to waive animal studies with some regularity. The statute used to look a little bit different. That prong — 351(k)(2)(A)(i)(I)(bb) — required an animal study, and FDA had a practice of waiving those animal studies. I'm not sure whether it was a formal waiver or whether it was just a practice. But then the statute was actually changed to flip it so that the animal study requirement is no longer a direct requirement. The assessment of toxicity is a direct requirement.

You have described FDA's move away from CES as a “calculated risk.” What exactly is being risked? How does the agency justify the trade-off scientifically, legally?

Temkin: If you think about the “no clinically meaningful differences” piece of the statute, there is a cohort of people for whom that is a very important concept.

Most of the people who rely on these drugs as prescribers are used to looking at clinical data. You have a broad swath of therapeutic-area physicians, but they're generally used to looking at clinical data to support prescribing decisions.

I think it is a risk that the streamlining and the absence of that clinical data will change the perception of the products or the utility of the information, from the prescribing physician's perspective, in terms of reliability and availability of the information that they want in order to prescribe biosimilars.

That said, this is not new. FDA has, for a while, been considering approving biosimilars without this kind of data. They put out guidance, I think in 2019, on insulins that said a similar thing for that group of drugs. And so, it might not be that much of a risk after all.

FDA retains the authority to request a CES. How would you counsel biosimilar developers who are currently mapping out their development timelines on whether they should make a contingency plan? What kinds of factors may contribute to the FDA requesting one?

Temkin: This has always been the hardest question.

Even the old scientific considerations guidance had this stepwise approach in it, right? Where FDA said, “Come talk to us about your analytical data and we'll use that to inform the size and scope and duration of whatever clinical program is necessary.”

That's a great idea conceptually, but in practice, the timing is such that the biosimilar developer typically needed to start the clinical study at the same time they were generating the analytics. And so, the safe way to do it was to go ahead and do the study because you didn't know whether your analytics were going to prove out such that FDA would permit streamlined clinical trials.

I think what the FDA is saying now is that there is a greater presumption — this is how I read the draft guidance — that you don't need a comparative clinical efficacy study. The guidance sets out a couple of places where it expects that a comparative clinical efficacy study might still be necessary. For example, this applies to locally acting products such as intravitreally-administered products where comparative PK isn't feasible or clinically relevant.

I think you can expand that to basically any product where PK isn't doable because, for example, the mechanism doesn't involve the product getting into the bloodstream, so you can't rely on PK.

But largely speaking, FDA is viewing analytics and PK as the new belt and suspenders under this draft guidance.

About The Expert:

Eva Temkin is a partner in Arnold & Porter’s life sciences and healthcare regulatory practice. Previously, she was counsel and acting director for policy in the FDA’s Office of Therapeutic Biologics and Biosimilars. She received her J.D. from New York University School of Law.