Resurrecting Antibody Genus Protection: Allowed Claims Through CDR-Scanning

Welcome & Introduction

0:00:00 Gene Quinn
Welcome, everybody. My name is Gene Quinn. Thanks for joining us here today for this conversation about antibody genus claims — resurrecting that type of claim group and seeing what needs to be done, how we can do it, and whether it’s even possible in the first place, which is perhaps the first question. Not surprisingly, I think we’re going to say yes, it is possible, but we’re going to talk about how you can actually do that.

This is obviously in response to the Supreme Court’s decision in Amgen v. Sanofi several years ago, which — my characterization — took this industry out behind the woodshed and tried to shoot it. I just never understood that decision, but we have to live with it, and now strategies are developing so that we can. That’s what we’re going to talk about here today. We’re not going to complain about the Supreme Court or Amgen. We’re going to talk about what you can practically do in your practice today to address the new reality we’ve been facing over the last couple of years. So with that brief introduction to the topic, let me kick it over to Renee to introduce our speakers.

0:01:07 Renee Quinn
Thanks, Gene. Today on our panel we have Rich McCormick. He’s the founder and principal of Immuno Innova IP Consulting, with more than 25 years of experience in pharmaceutical and biotech patent law. He previously served as a partner at Mayer Brown, where he represented major companies including Bristol Myers Squibb, Sanofi, and Regeneron in complex patent litigation and portfolio strategy. Today he advises the life sciences industry on intellectual property issues related to biologics, antibodies, gene therapy, and other advanced immunotherapies.

Next we have Zach Silbersher. He is a partner at Silbersher Law and an experienced patent attorney with a practice spanning patent litigation, inter partes review proceedings, prosecution, and licensing. He also frequently advises financial analysts on patent disputes in the pharmaceutical and high-tech sectors. Zach has been recognized in IAM Strategy 300, the world’s leading IP strategists, and is regularly quoted in major media outlets on patent law and innovation issues.

Next we have Ben Doranz. He’s president and CEO of Integral Molecular, a company focused on developing technologies that enable the discovery of therapeutics for difficult protein targets. He co-founded the company in 2001 and has led its growth while bringing multiple scientific technologies from research to market. Dr. Doranz is an inventor on 50 patents and the author of more than 100 scientific publications, with work appearing in leading journals such as Cell, Science, and Nature.

And last but not least, we have Gene Quinn. He’s a patent attorney and a longtime thought leader in the IP world. He launched IP Watchdog in 1999, which has grown into one of the most trusted voices in patent law and innovation policy. He is widely recognized for his expertise in software patentability and U.S. patent procedure, and has been repeatedly honored as one of the top IP influencers and strategists in the global IP community. So you’re in great hands with this panel. Back to you, Gene.

Opening Remarks

0:03:07 Gene
Great, thanks a lot, Renee. And again, thank you all for being here. It’s an important conversation. I know for many of you it’s probably an existential question about whether you can get the protection that you need. We’ll have that conversation here today. If you have any questions during our time, please put them in the Q&A box — that’s where I’ll be monitoring for questions to bring to the panelists. If you want to chat amongst yourselves, please use the chat box, though it can get rather active and makes it a little hard for me and the speakers to follow along. So if you have something you’d like addressed, please put it in the Q&A box.

We do have Aayush here monitoring both the chat and the Q&A. So if you have a highly specific question, don’t be afraid to ask. Even if we’re not able to get to it during the webinar, we’ll try to get everyone answers to all of your questions either online or offline afterward.

0:04:13 Gene
So, gentlemen, let’s talk big picture to get us going. Ben, maybe I’ll ask you first — give you the opportunity to lead us off. What are your big-picture thoughts on this topic: antibody genus claims, or any aspect of that?

0:04:31 Ben Doranz
Yeah, thanks, Gene. And thanks to everyone for being here. We’ve been thinking about this topic for a long time. As a biotech company, we approach it from a data perspective. Can we discover and create the data that supports the kind of genus claims our customers have been asking for? Actually, our own lawyers too — as a company we file our own patents on antibodies and have for 20 years, and our lawyers have been asking us: can you get us more data? Can you make mutants of this antibody? And we can — that’s what we do as a company.

As Amgen progressed through the courts and the Supreme Court decision came out, we said, well, why don’t we just give you all of the mutations of the antibody? We don’t have to make just a few — here are all of them. And that kind of progressed into the genus claims we’re getting now. So we’re really happy to say there is a pathway forward. From our perspective, there’s a lot of data that can be used to create better claim sets, and that’s what I’d like people to take away: there’s a pathway forward through good data.

0:05:50 Gene
Great, excellent. Thanks, Ben. Richard, let’s go to you next. What are your preliminary thoughts?

0:06:12 Richard McCormick
I don’t want to be long-winded — there’s a lot to say — but I think the future is big data and AI. Surprisingly, right? That’s the answer to everything these days. I think the CDR scanning that Ben is going to talk about goes a really long way toward contributing to the pool of data we can draw from.

I tell my clients: don’t swing for the bleachers anymore. You’re not going to get a functional claim to an antibody without structure, binding anywhere on an antigen, with some sort of clinical or physiological response. Those days are over. But focus a little — look at these limited epitopic spaces. Maybe look at a single epitope that your lead candidate antibody binds to, and think about the structure you need for a common structural features analysis in a way that goes beyond just the amino acid sequence. I think what Ben’s going to talk about — that technology — goes a long way toward getting us there.

When you look at these enormous databases, the Oxford Protein Informatics Group maintains huge antibody-antigen structural databases. That’s the kind of data that AI can learn from to further identify common structural features for antibodies, as long as we’re talking about claims focused on a very narrow epitopic space — along the lines of what we saw in Amgen, maybe even narrower.

0:07:42 Gene
Great, thanks a lot, Richard. And Zach, your initial thoughts?

0:07:46 Zachary Silbersher
Thanks, Gene. I was following the Amgen v. Sanofi case for several years, including attending many of the district court proceedings. And unlike some people, I actually thought the Supreme Court came out on the right side of this issue. The big question I’d like to address today is whether this new era — where we can no longer get broad functional genus claims — is going to encourage biotech innovation or discourage biotech investment.

I think on balance, biotech innovation will actually increase in this new era, in large part because it will reduce investment risk for companies that get sandbagged by broad functional genus claims because a competitor beat them to the patent office. I think this is actually a good thing for the industry. What we’re going to hear about today is a number of different options for pursuing genus claims. We can no longer do what we did in the pre-Amgen era. The paratope mapping strategy that Ben is going to talk about is never going to capture the same patent breadth as those broad pre-Amgen functional genus claims, but it provides an interesting middle ground that will need to be developed and blessed by both the district courts and the Federal Circuit. It’s a strategy worth considering right now for all the reasons we’re going to hear about.

0:09:14 Gene
Great. Let me share the results of our poll question. Not a shock to me — most people picked “major challenge,” followed by “moderate challenge.” Does this surprise any of you? Richard, we’ll start with you this time.

0:09:37 Richard
It all depends on where you’re sitting in terms of the level of technology. At the time of Amgen, yes — major challenge. But I think things have advanced exponentially since then. So what I’m going to be talking about is how maybe it’s less of a challenge now. But if you’re thinking about this from a 35,000-foot view, “major challenge” does not surprise me.

0:10:03 Gene
Ben, thoughts?

0:10:05 Ben
It’s something we think about from two perspectives. We can help solve these problems, but actually we’re solving our own problems too, because we patent and license antibodies. We do a lot of discovery — we don’t do any clinical work, but we license our antibodies to companies like AstraZeneca and Context Therapeutics and others. When we license those patents, the value is the patents. We’re receiving significant licensing fees from major companies, and it’s all based on the strength of those patents. So the stronger we can make those patents, the more valuable they are.

0:10:43 Gene
And Zach?

0:10:45 Zach
I think these results don’t surprise me. It’s definitely going to be a challenge to obtain broad, defensible antibody claims, but we live in a new era. Pursuing a panoply of strategies to protect your antibodies — including strategies beyond just claiming the antibody itself — is important. What I’ve learned consulting with clients on how to delay biosimilar or generic competition is: attack the problem in as many ways as possible, with as many patents as possible. We’re never going to go back to the pre-Amgen era without some legislative change, but that doesn’t mean we can’t protect our antibodies, even if the protection is more narrow.

0:11:29 Gene
Great. So I’ll stop sharing that question now, and I think we’re going to kick it over to Richard. We have some slides to push our conversation forward, and Richard is going to lead off. Richard, take it away.

Presentation: Richard McCormick

0:11:46 Richard
Hello, everybody. I’m happy to be here. My goal is to tee up the discussion that Ben and Zach are going to have. I’m going to go over at a high level the §112(a) landscape post-Amgen, post-Juno, and post-AbbVie, and then talk about some patenting strategies — legacy ones and some old ones that have become new again, like means-plus-function — and see how those are going to fare in the world we’re living in now.

0:12:20
Last year was the 50th anniversary of the Köhler and Milstein paper, published in Nature in 1975, describing the creation of the monoclonal hybridoma — without which we perhaps wouldn’t be having this webinar. They stated with what I think was Nobel Prize-worthy understatement that hybridomas could be of valuable use in medicine. And it turns out, of course, they were absolutely right. That promise has been fulfilled. Paul Carter wrote a paper in Nature Reviews Immunology last year looking back at the last 50 years of monoclonal antibody development.

0:13:10
So where do we stand now? We have well over 200 therapeutic monoclonal antibodies approved by regulatory agencies across the world. Tens of millions of patients treated. There’s a global market value forecast of about half a trillion dollars by 2028 for the monoclonal therapeutic antibody market. And upwards of about 16,000 issued U.S. patents have claims directed to monoclonal antibodies. Antibody therapeutics are clearly on the rise. What about the functional claims that make up a lot of those 16,000 patents?

0:13:55
What we have seen is a rise, decline, and fall — like the Roman Empire for functional genus claims to antibodies. I’ll go through this quickly. We start with Regents of the University of California v. Eli Lilly (1997), which sets out the two different tests you can look at, which we’ll talk about in a moment. Then we move on to Noelle v. Lederman (2004) — the antibody exception. If you have a fully characterized antigen, you can have a functional claim without having to satisfy the Regents test. AbbVie v. Janssen, I think, effectively abrogates that sub silentio — they invalidate the IL-12 functional claims and don’t even give lip service to Noelle. After that we have Amgen v. Sanofi at the Federal Circuit, which expressly abrogates the fully characterized antigen test as inconsistent with the statutory mandate for written description. Then Juno v. Kite, which is a further retrenchment and a rigid application of the Regents v. Lilly standard. And then obviously we end up with Amgen v. Sanofi at the Supreme Court, requiring full-scope enablement. Now we have USPTO guidelines implementing Amgen — basically saying they’re applying the Wands factors but not really providing patentees much further guidance on how to meet enablement.

0:15:10
Let me walk through the Regents v. Lilly test. There are two prongs: representative number of species, and common structural features. Both have in common that you have to look at structure. For the representative number of species prong, you need enough antibody species to reflect the structural diversity of the claimed genus. For the common structural features prong, you need a common structure that runs through all of the species within the genus, such that when you see the claimed function you immediately call to mind this common structure.

0:15:50
What I tell my clients when advising them on either of these tests is: think beyond the primary structure — beyond the simple amino acid sequence. When I think about structure in today’s technological context, I think about tertiary structure — the three-dimensional topography of the variable region. I think about all of the non-covalent interactions that go on with antigen. Other structural considerations could be the distribution of charges on the variable region, hydrophobicity mapping, or even just the spatial arrangement of polar R groups. That’s all structure we should be thinking about, beyond just the simple amino acid sequence.

The goal that the Federal Circuit has stated for both tests is to allow a skilled artisan to readily distinguish species that fall within the genus — those that have the function — from those that don’t, so that you can visualize or recognize the members of the genus.

0:16:50
Here are the two leading cases applying the Regents test in the antibody space: AbbVie and Juno. I’ll assume a certain familiarity with them, as we could discuss these cases forever. AbbVie is basically a functional antibody claim binding IL-12 anywhere on the antigen, requiring neutralization of IL-12 and a certain k-off rate. The species disclosed in that patent is one parent antibody — J695 (also called Joe-9) — and about 200 variants of Joe-9, all at 90%-plus sequence identity to Joe-9. What the court finds — and first I should say, for common structural features, AbbVie concedes there are none. They put all their eggs in the representative number of species basket. They argued the 200 variants are representative of the structural diversity of the genus, and the Federal Circuit essentially says, not so fast. You’ve accused Stelara® — which is Janssen’s antibody — of infringing. Stelara has a totally different amino acid sequence; it’s about 50% identity to Joe-9 and its variants. So Stelara is like that Karl Popper black swan that falsifies AbbVie’s argument. You only need one example to show the species aren’t representative, and the claim falls.

0:18:20
Juno v. Kite is this further retrenchment and a very rigid application of Regents v. Lilly. We could talk about this case forever. Gene has submitted critiques of it. I would point out that at trial, the jury found that Kite had not proven invalidity under written description. The district court denied JMOL on that, finding the jury could have found the claims met either representative number of species or common structural features — and could even have found the so-called Capon well-known genus principle applied, where you don’t need to describe all the binding elements because the genus was well known. The Federal Circuit didn’t buy that and invalidated the claim. We know it went up to the Supreme Court, which denied certiorari around the same time as Amgen v. Sanofi. At the end of the day, my feeling is that if this claim was going to fall, it should have gone down on the point of novelty — specifically, whether you can activate T cells once the single-chain variable fragment binds to antigen. scFvs were really well known by 2002 when the patent was filed, and we knew how to clone variable regions. But in any event, Juno represents, for me, the true fall for written description for functional antibody claims.

0:19:55
Moving on to enablement: Amgen v. Sanofi and the epitope claim. The exemplary claim is directed to an antibody — no structure mentioned — with the function of binding two or more of 15 amino acid residues on PCSK9. In terms of supporting the claim, Amgen had 10 antibodies in their patent. Most people note that three amino acids had no antibodies binding to them, and none of the 10 antibodies bound more than 9 amino acids. When I was thinking about this case — full disclosure, when I was at Mayer Brown, our client Unified Patents put in an amicus brief on the side of Sanofi — I was struck by some quick combinatorial math. Two or more of 15 residues comes out to about 32,000 different permutations, meaning roughly 32,000 different ways an antibody could, in theory, bind to PCSK9 within the scope of the claim. There was testimony at trial that many of those permutations were unlikely to have antibodies made to them. So that’s how I always thought about this case.

0:21:15
The key question in the patent was: how do you identify undisclosed antibodies aside from those 10 that have the claimed function? Amgen set forth two methods: the roadmap and conservative substitution. The roadmap is basically: do what we did — screen, screen, screen, and you’ll find antibodies that bind to two or more of these residues. Conservative substitution: start with an antibody that actually meets the function, and then make minor modifications — maybe enhancing binding in some way, reducing immunogenicity. The Supreme Court said no — Amgen’s roadmap and conservative substitution methodologies are just giving the person of ordinary skill two research assignments: essentially saying, do what the inventors did, go through that iterative trial-and-error process to find antibodies that bind. That is painstaking experimentation, not the reasonable experimentation the statute allows for enablement.

0:22:30
Importantly, what the Supreme Court did say is that if the inventors had identified a quality common to every functional embodiment — a common structural feature — that would have reliably enabled the skilled artisan to make and use the full scope of the genus. My takeaway is that common structural features can satisfy both description and enablement. That’s where we are now.

0:23:05
Some of the claiming methodologies to consider after Amgen and Juno: We still obviously have claims to the actual amino acid sequence of the variable region or CDR sequences of your lead candidate antibody — usually claimed by SEQ ID numbers, with functional limitations as well but with a fixed structure. Claims like that are the strongest on validity, and when thinking about loss of exclusivity, you can pretty much assume infringement by biosimilars won’t occur before expiration of that patent. The problem is they’re not going to capture a competitor’s different antibody that may bind the same antigen, even around the same epitope, using the same mechanism of action. So they’re going to be of limited use as a substitute for functional genus claims.

0:24:10
We also have sequence identity claims — you claim 80%, 85%, up to 99% sequence identity compared to your lead candidate amino acid sequence, potentially with functional limitations as well. First, especially after [Duke v. Sandoz — to be verified from video], you’ve got serious written description concerns and risk of invalidity for a claim like that, especially if you haven’t mapped out exactly what that field of 80–99% identity looks like in terms of actual examples. Once you add a functional limitation into a claim like this, you’re also bringing in the Amgen analysis and risking invalidity on enablement. And at the end of the day, this type of claim is easy to design around — make a couple of changes to the CDRs and you’re below 80%. I’ve been doing FTOs and due diligence for 25 years and I’ve never come across a sequence identity claim where I could match it up against a competitor antibody and say that it was within the 80%. So a claim like this is not going to be a substitute for a functional genus claim.

0:26:00
That brings us to means-plus-function, which has generated a lot of discourse as a possible Amgen and Juno loophole or workaround. We all know Ex parte Chamberlain, which went up on appeal to the Federal Circuit as In re Xencor. The claim at issue — Claim 9 — was a method of treating a patient by administering an anti-C5 antibody comprising a “means for binding human C5 protein.” What the Appeals Review Panel did was find that yes, this properly implements the means-plus-function statute using the right language — there’s no structure in the claim — and that the antibody species in the patent spec, 5G1.1, is the corresponding structure. What you’re entitled to in terms of claim scope is then equivalents of 5G1.1. And importantly, what the ARP said is that you don’t need to provide a written description of those equivalents in the specification — so you don’t need to disclose a representative number of species or common structural features for the equivalents to 5G1.1.

0:27:15
This was talked up a lot. Mark Lemley and Jacob Sherkow mentioned it in their Yale Law Journal article “The Antibody Patent Paradox” as a way to potentially get around the §112(a) jurisprudence. I have some reservations. First, this does not have the imprimatur of the Federal Circuit. The claim was ultimately rejected, and the ARP decision was affirmed on other grounds — the Federal Circuit never passed on this particular holding, so we don’t know what they would say. Another question is whether you still have to have enablement of the undisclosed equivalents — and I think there’s some Federal Circuit case law that even Lemley and Sherkow acknowledge suggests you do. How do you enable that in view of Amgen? And even if your patent survives a validity challenge, the scope of equivalence is uncertain. Is the equivalent of 5G1.1 going to capture a completely different competitor antibody? We just don’t have case law on that. My feeling is it probably won’t be able to capture different competitor antibodies, but we really don’t know yet. Those are the three asterisks I’d put next to means-plus-function.

0:29:00
So where are we on the future of functional genus claims to antibodies? We have PTO guidelines following Wands and Amgen, but not really telling patentees how to get claims enabled. What we do know is that now common structural features can satisfy enablement and written description. The question in 2025 is: how can we identify those common structural features? And I think, given big data and AI, we are well beyond where the technology was when the Amgen patent was filed. We’re light years ahead of that now.

0:29:45
To circle back to my big picture: focus on narrower claims. Don’t try to hit it out of the park. You’re not going to get a genus claim binding antigen anywhere — you might not even get a claim to the Amgen epitope scope, but maybe something narrower, or even a fixed epitope with a single antibody. And that’s where the CDR scanning that Ben is going to talk about comes in. A methodical, comprehensive, systematic way of generating antibodies in a high-throughput fashion, populating your specification with a lot of examples, especially as claims get narrower — I think that will allow you to find, at least from the amino acid sequence, a common structural feature. And thinking even more broadly, looking at big data, applying AI, doing what AI does best in terms of pattern recognition — finding common structural features across a huge scope of antibody-antigen interactions. I’ll leave it at that and turn it over to Ben.

Presentation: Ben Doranz

0:31:25 Ben
Great, thanks, Gene. Thanks, Rich, for the introduction and for covering a lot of background — that covered a lot of ground. I just want to back up for a second to give a bit of context about Integral Molecular, because it’s relevant to the conversation. As a company, we’ve been around for 20 years in antibody discovery, engineering, and mapping. We have a number of products and services in the antibody space, and a little in the virology space. A lot of people in the industry know us for our Membrane Proteome Array for specificity screening. But what I’m going to talk about today is around paratope mapping, which feeds into the IP discussion.

0:32:05
This started many years ago for us. For much of our 20 years, we’ve been scientifically engaged in understanding how antibodies bind their targets — epitope mapping and related work. We’ve epitope-mapped over a thousand antibodies for customers, partners, and ourselves, and published many of those in over 200 papers. Our customers and clients have also published and patented from that work to generate IP. So there have been over 200 papers and patents in this space, mostly on the epitope side. We published a paper in mAbs a few years back that talked specifically about the epitope mapping contribution to patent protection. At the same time, we’ve been doing antibody engineering for 20 years, so when the Amgen v. Sanofi case evolved and the decisions came out, we realized there was a fundamental shift in the landscape for patenting antibodies — and that we might have some of the tools to answer the questions the court cases left open.

0:33:25
In particular, as everyone knows, the court struck down functional claims and said: you need a common structure, you need reasonable experiments, and so on. But it didn’t provide a clear pathway forward. So we published a paper in Nature Biotechnology about a year ago describing CDR scanning for intellectual property protection around antibodies — there’s a link in the chat if anyone wants to download it.

The problem, obviously, was that yes, you want to claim more than one species. You can claim your antibody’s exact sequence, but that’s easy to get around and isn’t really claiming any kind of genus. You can’t claim a genus anymore, at least as defined pre-Amgen — claiming every antibody that binds to that target. As Zach has pointed out, that’s not the reality anymore. So our challenge became: how do you claim a genus that meets the USPTO guidance for enablement and written description, in a way that’s reasonable, meets the premise of the law, the quid pro quo, and the guidance of the courts?

0:35:00
The paper describes a strategy — more theoretical at the time, though I’ll show you the reality now — where we take an antibody and, experimentally in the lab, define an antibody’s genus and its variants using a common structural feature, which is essentially the paratope.

This general strategy is borrowed from what’s been around for a long time in small molecule chemistry. It’s well documented, and many people have patents derived around small molecule scaffolds — you’ve got the scaffold, you can put R groups at various positions, with limitations around which substitutions are permitted around that scaffold. What we realized is that for antibodies, the equivalent of a scaffold is the paratope: the critical amino acids of the antibody that you can’t change. That’s the common structural feature — the paratope. These are the residues that bind the epitope. If you accept the premise that the paratope is a common structural feature — which is what the courts want — we can define that paratope experimentally. That’s feasible.

0:36:15
So here’s the approach: we take an antibody, we mutate all of the CDR residues to every other possible amino acid. For antibodies, there are only 20 amino acids per residue — only 20 possible substitutions — versus a small molecule where an R group can theoretically be anything. For antibodies, you’ve got 20 residues you can substitute at roughly 50 different positions in the CDRs. You make those variants, and you can define which substitutions are permissive and which are not — what you can’t change. That gets you to an antibody genus. In these cases, for instance, you can change arginine to cysteine, glycine, or histidine, but you can’t change a particular proline. That’s one of the common structural features. And that translates directly to patent claims. This gives you enablement and written description — there’s the data. You’ve got the antibodies and the variants, and they translate directly to the claims.

0:37:30
That was all in theory. So we went into the lab and did it. We took an antibody against a particular G protein-coupled receptor we work on. We have a lot of robotics in the lab that run high-throughput screens. We took every CDR residue in the heavy chain and changed it to every possible substitution. We could see which residues you could substitute — those acceptable substitutions — and which you couldn’t. Take residue 155: you can’t change it to anything else. It has to be a glycine. That’s the common structural feature. What’s also really interesting from both a scientific and IP perspective is that you end up identifying mutations — shown here in orange — that bind better than the wild type. That’s exciting from a novelty standpoint.

0:38:45
We took that data and those proof-of-concept results and launched a service where clients can submit sequences; we mutate the antibody, express every variant, and quantify them by ELISA — by expression and by binding, all normalized to expression. That ends up in a report describing all the data: the paratope, the common structural features, the permissive substitutions, enhanced substitutions with improved binding, and we even write the patent claims for you in that report. When we started working with clients, we found it’s a lot of data processing, and rather than having clients sift through giant Excel sheets, we’ve automated it so you can push a button and turn that data into claim sets immediately.

0:39:40
The project takes about 12 weeks and generates about 1,000 variants. This is not a new concept — mutating antibodies and generating lots of variants is not new — but the novelty we’ve introduced is cost. Even 5–10 years ago, this kind of project — a thousand variants across an antibody — would probably have cost half a million to a million dollars, and it has been done by major pharmaceutical companies. The novelty is that we can do it for a tenth of the price: typically $50,000 to $100,000, and it gives you all this data and all the IP around that antibody.

0:40:30
We launched this service, and the immediate question from lawyers was: well, can you actually get a patent out of this? And the answer is yes. We actually just had one issue about a month ago. We filed the patent based on that initial antibody, and there’s a link we’ll send out with the follow-up materials — you can see the patent, the annotations, the claims, the USPTO office action, and all the file wrapper information. We got our genus claims allowed, and they basically cover the permissible substitutions as well as the theoretical combinations of those substitutions.

0:41:20
Briefly: for the first set of permissible substitutions, we said you can substitute this serine for all these amino acids, but you can’t substitute, say, this glycine with anything. That’s the data. We’ve reduced it to practice. It’s a common structure because we’ve identified residues you can’t substitute. It’s all based on structure, not function. And the claim we got allowed is: this sequence, or a variant thereof, where that particular first residue — serine — can be changed to any of these other residues. That’s very consistent with the Supreme Court and USPTO guidance: provide data and a claim set that enables a person skilled in the art to make and use all of what’s claimed. We’re claiming these substitutions, and there’s no real controversy around enablement and written description for these variants because we have the data and they’re all in the patent.

0:42:35
The second part of what we got is: same claims, same data, but combinations of those substitutions. We got not just that you can change each residue to these alternatives, but that you can change multiple residues to different combinations — except for the common structural feature residues, like the glycine where there are no substitutions. That glycine is the common structural feature running through all the CDRs — the “general quality running through the class” the Supreme Court and USPTO are looking for. This allows you to visualize and recognize the members of the genus: they all have to have that glycine. So we got that claim allowed: the antibody sequence selected from these residues and/or these other residues.

0:43:45
A question we sometimes get: you’re predicting these theoretical combinations, but is this reasonable? Are the common structural features actually allowing a person skilled in the art to build the genus within a reasonable amount of experimentation? We didn’t know at first. So we tested it — we took those common structural features and held them constant, then built about 400 other mutations, changing the other permissive residues around the common structural feature. Ninety-five percent of those variants were as good as wild type or better. In many cases we identified variants 30 times better than the parental — new IP at the same time. So we think this answers the question the courts and patent office have been guiding everyone toward: you don’t need to make every single embodiment within the claims, but you need to identify the common structural features that enable you to make and recognize the genus within a reasonable amount of experimentation. If you can predict the genus 95% of the time, or even 50% of the time, that’s pretty reasonable by most measures.

0:45:10
Backing up a step, when we look at our strategy, we think it’s a nice balance of what the courts intend — the quid pro quo — where you’ve got claim breadth, not everything under the universe but around your antibody and anything sharing those common structural features, and you’ve also got §112 compliance. This has held up really well. We launched this service last year — it’s all fee-for-service — and we’ve had really good engagement and really good feedback. We’re making some examiners happy and checking off the boxes for enablement and written description. The CEOs we talk to understand the value: they’re getting stronger patent protection and new IP. Investors really like it too — some of our customers are VC firms building better patent portfolios for their companies and reducing FTO risks. And scientists love it, because they get better information for antibody engineering, efficacy, safety, and developability. So far, this has been really well received, and we’re excited to work with people. Thanks, and I’ll kick it back to Gene.

0:46:55 Gene
Great. There’s one question you might want to answer that people may have in their minds. What separates the projects that you file on versus the ones your clients file on?

0:47:10 Ben
There are only a few targets we go after as a company. So if someone comes to us with a target we can’t work on, we flag that immediately and don’t do the project. We have some licenses — with AstraZeneca, Context Therapeutics, Conjugate, and a few others — where we’ve licensed our IP to those companies, and those licenses prohibit us from working on those targets. If someone brought us that target for fee-for-service work, we’d flag it. It’s a pretty narrow list.

0:47:45 Gene
So it’s a carefully choreographed conflict-checking process. You’re not in competition with the people you’re offering fee-for-service to.

0:47:54 Ben
No, we’ve been in business for 20 years and we’re pretty conscious not to be in competition with our customers.

0:48:00 Gene
I figured that was the answer, but it was worth nailing down. Thanks, Ben. Zach, let’s kick it over to you.

Presentation: Zachary Silbersher

0:48:19 Zach
Sure. I only have a few minutes, so I want to cover the most important points. Richard did a great job providing background on how we got to Amgen v. Sanofi and identifying patenting strategies in its wake. One of the most important of those strategies is the paratope mapping approach that Ben talked about.

There are two questions I’d like to address. Part of where I’m coming from is that many of my clients are investors in the pharmaceutical industry, including analysts who talk to companies on both the buy side and the sell side. First: will the inability to obtain broad functional genus claims discourage biotech innovation? Second: given the current array of strategies, which ones are going to encourage the most biotech investment and move the industry forward?

0:49:30
With respect to biotech innovation, as I mentioned earlier, I do believe the Amgen v. Sanofi decision will actually encourage more innovation. During the Supreme Court case, a number of amicus briefs were filed on both sides. Interestingly, the amicus briefs filed on behalf of Amgen — mostly big pharma companies — argued that without the ability to file broad functional genus claims that could patent an entire class of new inhibitors, that would kill the incentives to do research and development into life-saving therapies. As expected, the amicus briefs filed on behalf of Regeneron and Sanofi argued the opposite. I think the most interesting takeaway was that broad functional genus claims, even though they seemed to be a boon for the companies that held those patents, created major investment risk for their competitors. Regeneron was an example of that — Regeneron faced an injunction against its PCSK9 drug after investing hundreds of millions of dollars. Not because it had inferior efficacy or toxicity issues, but really because Amgen beat it to the patent office with broad functional genus claims. When I was at the district court hearing on the injunction proceeding, the judge actually remarked: “I don’t really see that Regeneron did anything wrong here — I just see that Amgen made it to the patent office first.” Having a single patent potentially sink a hundred-million-dollar investment spooks a lot of investors in any industry.

0:51:10
In fact, even though Amgen was not able to enforce its broad genus patents against Regeneron — even though it lost those patents at the Supreme Court and could never enjoin Regeneron’s Praluent drug — Amgen’s business case for Repatha was not undermined. There’s actually some data that Amgen is beating Regeneron in the market today. And I think that’s important as we go forward and realize that even though we can’t get those enormous breadth claims anymore, that doesn’t necessarily mean it’s going to kill the business case for our drugs going forward. We don’t have to hit it out of the park — get narrower claims that are just as useful and use them in a smart way.

0:52:05
Richard talked a lot about the cases leading up to Amgen. I want to identify a couple of cases in the wake of Amgen that illustrate this landscape is not going anywhere. The Baxalta case is really just an instance of the Federal Circuit applying Amgen‘s analysis against broad functional genus claims — materially indistinguishable facts from Amgen itself. Richard also mentioned [Duke v. Sandoz — to be verified from video]. What’s interesting about that case is that it applies the Amgen reasoning within the context of a small molecule patent, showing some read-through outside of the antibody scenario toward narrowing the requirements for both written description and enablement.

0:53:00
Another interesting case, still pending, that I’ve written about: a series of post-grant review proceedings filed by Merck against patents that may soon be asserted against Merck’s oral formulation for its cancer drug, Keytruda. What’s interesting about these PGRs — which have now been instituted — is that they include patents involving the sequence identity claims Richard discussed. Even though the structure of those claims required 95% identity to reference proteins, Merck made the point that the claims still encompassed billions and billions of antibodies — in fact, they argued that if you manufactured every single antibody encompassed in one of those claims, the aggregate weight of all those antibodies would be more than the weight of the planet Earth. That was their way of illustrating the magnitude of antibodies that could be included even in sequence identity claims, currently entertained as an alternative to genus claims in the post-Amgen era.

0:54:20
Another interesting point about that case: the patent owner tried to circumvent the written description and enablement analysis by arguing that the claims are directed only to structure, with no function — so you don’t have to prove any activity and don’t have to test them. Without a functionality, though, you’re probably going to run into a utility problem. The argument was rejected by the PTAB under claim construction analysis, but it will probably go up to the Federal Circuit at some point. It’s probably not a winning strategy.

0:55:10
Richard and Ben talked about a series of new patenting strategies in the wake of Amgen, and it’s worth looking at them all at once. Each represents a trade-off between breadth of patent coverage on the one hand and §112 issues on the other. Functional claims like those in Amgen are no longer viable, barring legislative change. Parental sequence claims are always worth getting for biosimilar purposes, but they won’t give you broad genus coverage. Sequence identity claims, as illustrated by the Merck case, present many of the same written description and enablement problems given the enormous number of antibodies they may encompass. Means-plus-function, I’d note, may seem to have broad coverage because it includes structural equivalents, but those equivalents often have to be known and identified before the patent issues — which means that if you assert them in litigation, your competitor’s antibody might actually be prior art to your own patent.

0:56:20
That’s why I think Integral Molecular’s approach provides an interesting middle ground: patent coverage on the order of thousands of potential antibodies, as opposed to just a handful, because it identifies the common structural feature — showing what is essential in the paratope — that could potentially satisfy both the written description and the enablement requirements.

There are still open questions. One Ben identified is whether the combinatorial claims will hold up on enablement or written description — that’s a question that will have to be decided by the courts. But one thing I’ve learned working with Ben is that the technology is developing to include those combinations quickly, and the technology may be moving faster than the court decisions. We may not get certainty from the Federal Circuit for another 10 years. Big picture: pursue as many strategies as possible. We’re in a landscape without any blessing from the Federal Circuit on anything, so an all-of-the-above strategy is the right one.

Question & Closing Remarks

0:57:28 Gene
Okay, great. There’s one question, Ben, from Sarah about expanding from proteins to nucleic acids. Do you have any thoughts on that?

0:57:48 Ben
Our focus has always been antibodies, but the approach and strategy we’re talking about actually applies beyond antibodies. You could do any protein, and anything encoded by a nucleic acid — you can start making substitutions and get the same kind of protection. In the Nature Biotechnology article, we allude to that: you can do viruses, genes, RNA. There are a lot of other modalities these days. We focus on antibodies, but the same strategy applies.

0:58:18 Gene
So if anyone has questions about applying this beyond antibodies, they can still get in touch. Okay, I want to give everybody an opportunity to wrap up, since we’re just about at time. We’ll get all of you the answers to your remaining questions as we move into the post-webinar phase with materials, the recording, and the PowerPoints. Ben, we’ll give you the final word — but Zach, how about we start with you? What’s your takeaway message today?

0:59:11 Zach
The big picture is that we’re in a new era. We can no longer get these broad functional genus claims, but that doesn’t mean it’s going to kill the industry, and it doesn’t mean we can’t get strong patents that will protect our IP not only against biosimilars but also against competitors. The paratope mapping strategy and the technology developing around it is happening so quickly that it’s worth pursuing, and you really have to pursue all strategies. At the end of the day, we don’t really know where the Federal Circuit is going to come out in 10 years.

0:59:44 Gene
And that’s not just this area — it’s across the board. It really is an all-of-the-above patent strategy, because there’s not a lot of certainty. Thanks a lot, Zach. Richard, your message — if there’s one thing you want people to leave today remembering, what is it?

1:00:01 Richard
I agree with what Zach just said. I want to add a little extra value to the CDR scanning and the product you’re going to get from Integral Molecular. You’re going to get sequences, and an amino acid sequence is a structure — that’s a common structural feature. I think we’re at the point now where you have AI programs like AlphaFold and even better that can take those sequences and expand the common structural feature beyond the primary amino acid sequence. You can plug those sequences into AlphaFold and get really, really good results — probably almost as good as, if not equal to, X-ray crystallography or electron microscopy. You can see the three-dimensional structure, and then work off of that to look at charge distribution and so on. I think you could have even broader claims to different kinds of common structures beyond just the amino acid sequence. That’s an added value to what Integral Molecular is providing you as a product.

1:01:07 Gene
Excellent. Thanks a lot, Rich. And Ben, the final word goes to you.

1:01:14 Ben
Thanks. First I just want to say thanks to Gene, Zach, and Rich — and by the way, they’re not being paid to say these things. We do share common views, but I appreciate being part of the webinar.

What they said is exactly right. The past is the past — it’s not coming back. The future, to us, looks like data. Any scientist or IP attorney who works in this space knows that data drives not just good clinical results but also good IP, and that’s the direction the courts are telling everyone to go. As a company, we’re somewhat agnostic to what the courts and patent office and Supreme Court say — those are just the rules we’re following. But within those rules, the guidance they’re giving us is exactly what’s been done, more or less, for small molecules. We’re just applying it in a different field. Whichever way you go, there are limited strategies going forward, and relying on data is never a bad strategy. We hope we can help people.

1:02:57 Gene
Certainly in this area, relying on data is an absolute necessity. Thanks a lot, Ben — really appreciate you being here. I appreciate everybody in the audience being here and sticking with us a few extra minutes to give us a chance to wrap up our thoughts. I thought this was fascinating. If you asked a question, we’ll get you an answer. That’s all we have for today, everyone — bye for now, and thanks for coming. We’ll see you soon.