Thoughts On The Celgene Revlimid Markman Hearing
A Markman hearing in Celgene's Revlimid patents case took place Thursday.
Markman process defines what Celgene's Revlimid patents cover.
This article reviews the issues addressed at the hearing and provides my expected outcome.
On Thursday, in federal district court in Newark, New Jersey, there was a very important hearing in the patent infringement case brought by Celgene (CELG) against Actavis' (ACT) affiliated companies, Natco, Arrow and Watson, stemming from their filing an application with the FDA to market a generic version of Celgene's top selling drug, Revlimid. I provide here an overview of Celgene's Revlimid patents, a background on the Markman process, and a summary of the issues discussed during Thursday's hearing. I also provide my expectation for how I believe the judge will rule on the various issues and what that means for the ultimate outcome in the case.
Celgene's Revlimid Patents
While there are 23 patents listed in the Orange Book for Revlimid, Celgene actually has at least 56 issued patents that relate to the drug, as well as several more patent applications currently pending at the Patent Office that may result in even more patents. In the current case against Actavis, Celgene is asserting 18 of its 56 Revlimid patents.
Of the 18 asserted patents, 15 are listed in the Orange Book and 3 are not. The significance of listing a patent in the Orange Book sometimes confuses people, so to explain briefly, patents listed in the Orange Book for a drug are not necessarily the only patents a potential generic competitor may infringe. Thus, it is incorrect to say Celgene only has the 23 patents listed in the Orange Book for Revlimid. Celgene lists in the Orange Book those patents that it believes would be infringed by an exact copy of its version(s) of the drug approved by the FDA. If Celgene's drug has a particular form, for example, then Celgene's patents on other forms that are not approved would not be listed in the Orange Book, because those patents on the other forms would not be infringed by an exact copy of Celgene's form of the drug.
However, if a potential generic wants to offer a form of the drug different than the exact form Celgene sells, Celgene may have other patents that cover that different form that are not listed in the Orange Book. Those non-OB patents can absolutely be asserted by Celgene against the potential generic. The fact that they are not listed in the Orange Book or do not cover Celgene's exact form of Revlimid is of no consequence. This is why 3 of the 18 patents Celgene accuses Actavis of infringing are not listed in the Orange Book for Revlimid, because Actavis is presumably proposing to produce a generic form of the drug that is not exactly the same as Celgene's, even though it may still nonetheless infringe those other non-OB patents. To be sure, whether a patent is listed in the Orange Book or not, if it is infringed by Actavis' proposed generic version of Revlimid, Celgene has the right to keep such infringing generic off the market until any and all infringed patents expire. Further, Celgene can (and did) sue Actavis for infringing patents that are listed in the Orange Book, because patents on the compound, composition, and method of use can be infringed by Actavis' proposed generic version even if it is not exactly the same form as Celgene's. If that's a bit confusing, hopefully it will make more sense as I discuss the specific patents being asserted in the current suit below.
When reviewing a patent portfolio, it is best to think of patents in terms of families, not individuals. A patent family is a group of patents and patent applications that claim a relationship to one another because they have common subject matter. I won't go in to all the reasons why patents and applications may claim relationships to one another, but, as an example, patent applications will often claim a relationship to an earlier filed application in order to get the benefit of the filing date of the earlier application for prior art purposes. As another example, sometimes the Patent Office will require an application be split into several separate applications in order to make its review easier (and increase the amount of fees it gets from the applicant). All those applications will relate to each other and be considered part of the same family.
Celgene began applying for patents relating to Revlimid back in July 1996 and has continuously filed more and more patent applications ever since then, including as recently as last month. This is very typical for a company that continues to develop and improve its technology, as it should constantly be applying for new patents on each incremental improvement and innovation. There is never an end to technological development. A company should always be working to make their technology better, faster, cheaper, safer, etc., and each milestone towards those goals can be sufficiently inventive to be worthy of patent protection. I'll come back to this point later.
To date, the 56 patents in Celgene's Revlimid patent portfolio divide up in to the following six families:
The Muller family of Revlimid patents and applications all relate to the application filed by Celgene in July 1996. There are 13 patents issued in the Muller family and 2 more applications still pending. Of the 13 Muller patents, 5 are listed in the Orange Book, and all 5 of them are being asserted against Actavis in the current suit. The asserted Muller patents are U.S. Patents Nos.5,635,517, 6,281,230, 6,555,554, 7,119,106, and 8,288,415. The '517 patent expires Oct. 4, 2019, the '230, '554 and '106 patents expire Jul. 24, 2016, and the '415 patent expires Apr. 7, 2017.
The Elsayed family of Revlimid patents and applications all relate to an application filed by Celgene in August 1998. There are 7 patents issued in the Elsayed family. Of the 7 Elsayed patents, 5 are listed in the Orange Book, and 2 of them are being asserted against Actavis in the current suit. The asserted Elsayed patents are U.S. Patents Nos. 6,045,501 and 6,561,976. The '501 and '976 patents expire Aug. 28, 2018.
The Williams family of Revlimid patents and applications all relate to an application filed by Celgene in October 2000. There are 8 patents issued in the Williams family and an application still pending. Of the 8 Williams patents, 5 are listed in the Orange Book, and 4 of them are being asserted against Actavis in the current suit. The asserted Williams patents are U.S. Patents Nos. 6,315,720, 6,561,977, 6,755,784 and 8,315,886. The '720, '977, '784 and '886 patents expire Oct. 23, 2020.
The Zeldis-1 family of Revlimid patents and applications all relate to an application filed by Celgene in May 2002. There are 16 patents issued in the Zeldis-1 family and 9 more applications still pending. Of the 16 Zeldis-1 patents, 4 are listed in the Orange Book, and one of them is being asserted against Actavis in the current suit. The asserted Zeldis-1 patent is U.S. Patent No. 7,968,569. The '569 patent expires Oct. 7, 2023.
The Zeldis-2 family of Revlimid patents and applications all relate to an application filed by Celgene in October 2002. There are 5 patents issued in the Zeldis-2 family and an application still pending. Of the 5 Zeldis-2 patents, 2 are listed in the Orange Book, and both are being asserted against Actavis in the current suit. Those 2 Zeldis-2 patents are U.S. Patents Nos.7,189,740 and 8,404,717. The '740 and '717 patents expire Apr. 11, 2023.
The Jaworsky family of Revlimid patents and applications all relate to an application filed by Celgene in September 2004. There are 7 patents issued in the Jaworsky family and 4 more applications still pending. Of the 7 Jaworsky patents, 2 are listed in the Orange Book, and one of them is being asserted against Actavis in the current suit. Celgene is also asserting 3 other Jaworsky patents not listed in the Orange Book against Actavis. The 4 Jaworsky patents being asserted against Actavis are U.S. Patents Nos. 7,465,800, 7,977,357, 8,193,219 and 8,431,598. The '800 patent expires Apr. 27, 2027, the '357 patent expires Jan. 8, 2025, and the '219 and '598 patents expire Sep. 3, 2024.
So, to review, Celgene currently has 56 issued patents that relate to Revlimid, and more applications pending. Of the 56 already issued patents, 23 are listed in the Orange Book for Celgene's approved versions of Revlimid. Celgene is asserting 18 of its 56 Revlimid patents against Actavis in the current patent infringement suit, 15 of which are listed in the Orange Book and 3 of which are not. As Celgene is granted more patents relating to Revlimid, it may file additional patent infringement suits against Actavis for infringing them as well. (At the initial stages of the current suit, Celgene added new patents as they were issued, but at this stage of the suit I expect the judge would not allow Celgene to simply add more new patents, but instead have them file new suits that would likely be presided over by her as well.)
Now that we have an overview of Celgene's Revlimid patent portfolio, the next logical question to ask is what those patents cover. That may seem like a simple question, but it is actually perhaps the most difficult question to answer in all of patent law. Determining what a patent covers is indeed so difficult that the Supreme Court decided it is a job that must be done by the judge assigned to the case, and not left to a jury. Named for the Supreme Court case that spawned it, the "Markman" process that federal judges undertake in order to determine what a patent involved in litigation covers is something they have to do before the case proceeds to determining whether the patent is infringed, because you can't possibly know if a patent is infringed until you define what it covers.
The "Markman" process is generically called the "claim construction" process, because the boundaries of a patent are determined by single sentences at the very end of the patent called "claims". These (often very lengthy) sentences literally start with, "I claim" or "What is claimed is ..." and then proceed to set forth, in words, what the patent covers. There can be several claims within a patent, and each claim stands on its own for purposes of infringement. So long as one claim is infringed, the patent is infringed, and the patent holder is entitled to appropriate remedies (injunction, damages, etc). Under Markman, the job of the judge is to interpret the words and phrases contained within the claims to determine what they mean. For example, my patent claim may say the word "yellow" but we can disagree about whether that means all versions of yellow or only bright stoplight yellow. If you sell something that is canary yellow, the decision about how the word "yellow" is interpreted will be very important to the case. In essence, because language is inherently arguable (especially the American dialect of the English language), there are almost always some differences of opinion on what various words and phrases within a patent claim mean. This is why we have a Markman process, to figure all that out.
As you can probably predict by now, claim construction can be an extremely painstaking and laborious process, rife with areas for disagreement. This is a large reason why patent litigation can be so expensive and lengthy. In fact, Markman decisions made by district court judges, if appealed, are reviewed by the Court of Appeals without any deference. Because of that, the Court of Appeals frequently reverses district court Markman claim construction decisions.
Typically, the Markman claim construction process involves the parties in a patent lawsuit first identifying the words and phrases in the asserted claims on which they agree and those on which they disagree. The parties then file a joint statement with the judge setting forth their agreements and disagreements. Celgene and Actavis did that in the current Revlimid case in October. Then the parties proceed to file briefs addressing the issues of disagreement explaining why their proposed interpretation of the disputed claim terms is correct and the other party's proposal is incorrect. In the current Revlimid case, Celgene and Actavis submitted their initial Markman briefs in late October and then response briefs in April. After briefs are submitted, there is usually a hearing in front of the judge, after which she will issue a written opinion with her conclusions.
Having reviewed the joint statement and the briefs submitted by the parties (Celgene Opening Brief, Actavis Opening Brief, Celgene Response Brief, Actavis Response Brief) and attended the Markman hearing on Thursday, I can gladly report that there was no dispute regarding any claim terms in 12 of the 18 asserted patents. For the other 6 patents, there are a total of 11 different claim terms on which the parties disagree, although those 11 break up in to groups as set forth below. Going in to Thursday's hearing, there had been 16 claim terms in dispute, but late Wednesday night Actavis conceded to Celgene's arguments on 5. Before I discuss the disputed claim terms in more detail below, I want to explain how the Markman hearing on Thursday proceeded.
As is frequently done in patent cases, especially ones involving highly complex science, the assigned judge will ask the parties for a basic science tutorial at the beginning of the first substantive hearing in the case. This is meant to help educate the judge, who often has no science background. In the current Celgene v Actavis Revlimid case, there haven't yet been any major substantive hearings, so Thursday's Markman hearing provided that opportunity. Thus, the parties spent the first part of the morning going over the basic science involved in the patents before getting to the specific issues on which they disagree. Each side took about 20-30 minutes to do so. As is typical, the scientific tutorials given by the parties were imbued with bias supporting the positions they would later argue expressly, which is why the judge allowed both sides to present their version of the "basic" science.
Once the tutorials were over, the parties proceeded to discuss all 11 claim terms in dispute, going one by one, with each side presenting what they think is the correct interpretation of each claim. The Markman hearing Thursday began at 10am and went until 1pm.
With respect to the '230 and '554 patents (which expire in July 2016), referred to by Celgene as "compound" patents, the parties disagree on what the phrases "said compound has the R-configuration" and "said compound has the S-configuration" mean. Specifically, the question is whether these terms exclude racemic mixtures, such that the patents only cover compounds with all or substantially all of the identified isomer (R or S configuration; a "racemic" mixture has some of both).
Celgene pointed out that the claim term is "has" isomer, not "has only or substantially only" isomer. Thus, to Celgene, the patents cover mixtures that are racemic. Actavis argued that expressly saying "R (or S) configuration" ordinarily means that the mixture is not racemic, that it only has the specified isomer present. Further, Actavis argued, the disputed claims are subordinate to other claims on a racemic mixture and therefore they must be different (i.e. NOT racemic).
Celgene has the stronger argument, because the plain meaning of the word "has" does not mean "has only or substantially only." It's inclusive, not exclusive. Actavis tries to point to the body of the patent (called the specification) for discussion that suggests a mentioning of R or S configuration implies non-racemic mixtures, but in the pecking order of claim construction evidence, the plain language takes precedence over the specification unless the specification clearly is defining the term to mean something else. Here, there is no such clear definition in the specification to limit a mention of "has R/S configuration" to mean "has only or substantially only the R/S configuration."
As you will see throughout, Actavis generally argues for narrow constructions of the asserted patents with the hope that they can later show in the case that their proposed generic does not fall within that narrow definition. Celgene, on the other hand, generally wants to maintain broad patent claims so as to make it easier to prove later in the case that Actavis' proposed generic infringes the more broadly interpreted claims. Here, if the court were to adopt Actavis' construction, they could get around these patent claims by simply adding enough of both isomers to their generic so that it is not "all or substantially all" of only one of the isomers. Indeed, in Actavis' opening Markman brief, they say that their proposed generic is racemic, which is why they want the judge to interpret these patents to not cover racemic mixtures.
Polymorph (Crystalline Form) Patents
With respect to the '800 patent (which expires in April 2027), referred to by Celgene as a "polymorph" patent, but which I'll call the "hemihydrate polymorph" patent, the parties dispute two different terms in the asserted claims.
First, the parties disputed whether the term "hemihydrate" requires an exact ratio of water to compound of 1:2. Celgene argued such is not in the claim language itself, and it is ordinary to assume some approximation. Further, Celgene argued, a narrower claim specifically says the ratio is .46-.59:1, and so the broader claim without an express ratio given can not be interpreted to be exactly 1:2. Actavis argued that the common definition of the term "hemihydrate" means exactly 1:2 and cited some cases involving the interpretation of the term in other patents. Celgene has the more convincing argument, because there is almost always assumed to be some imprecision in science, and the narrower claim's specific recitation of .46-.59:1 proves that a precise 1:2 ratio was not intended in the broader claims that cite no ratio at all. The other cases cited by Actavis involving other patents are wholly irrelevant, as the same word can be interpreted differently in different patents. There is no "precedent" set when a term is construed in one case that demands the same word be defined the same way in other cases with other patents. Never happens, never will.
Second, the parties disputed whether the patent is limited to the specific example polymorph identified in the patent as "Form B." Celgene argued the claim language does not say "Form B", but instead the more general term "hemihydrate." Actavis argued that the only hemihydrate disclosed in the patent is the Form B, and thus the claims of the patent should be limited to Form B. However, it is improper to limit a claim to just the example(s) given in the specification unless the claim expressly says it is limiting itself to that example. Thus, Celgene has the more convincing argument on this issue as well.
With respect to the '357, '219 and '598 patents (which expire in January 2025, September 2024, and September 2024, respectively), all also referred to by Celgene as "polymorph" patents, but which I'll call the "anhydrous polymorph" patents, the parties dispute whether the asserted claims are limited to the specific example Form A described in specification. Unlike the hemihydrate polymorph patent, where none of the claims expressly say "Form B", some of the asserted claims in the anhydrous polymorph patents do expressly say "Form A", while others do not. Here, I stand by my opinion above, that a claim should not be limited to a specific example in the specification unless the claim itself expressly cites that example.
Therefore, for the claims in the anhydrous polymorph patents that expressly say "Form A", I agree with Actavis that they should be limited to the specific Form A in the specification. As for the other claims in these patents that do not expressly say "Form A", I agree with Celgene that they should not be limited to that specific example polymorph. For clarity, "Form A" is stated expressly in all of the asserted claims of the '357 patent and claims 1-4 and 14 of the '598 patent. Thus, I would side with Actavis on those claims. "Form A" is not stated expressly in any of the asserted claims of the '219 patent or claims 5-13 or 15-23 of the '598 patent. Thus, I would side with Celgene on those claims.
On this issue, Actavis says in its opening Markman brief that its proposed generic lenalidomide product does not contain a polymorphic form having all of the characteristics of Form A from these patents. Thus, you can see how important this issue is, because if the judge interprets any of the claims to require all of the characteristics of example Form A from the anhydrous polymorph patents, Actavis will argue it does not infringe those claims. Celgene may very well dispute those arguments, or argue that Actavis' proposed generic nonetheless infringes under the Doctrine of Equivalents (which counts as infringing things that are not within the literal claims if the differences between the accused thing and the patent are "insubstantial"), but it is obviously going to be easier for Celgene to prove infringement if the court adopts its construction of those patent claims to not require all of the characteristics of example Form A.
One major open issue, on which I do not have much clarity, is whether Actavis' proposed generic Revlimid is a hemihydrate or anhydrous or if they are proposing both. It seems to me that they must at least at this point be pursuing both forms, as Celgene would otherwise have no basis for suing them for infringing both the hemihydrate polymorph patent and the anhydrous polymorph patents.
The benefit to Actavis of pursuing an anhydrous form is that they then need not worry about the hemihydrate polymorph patent that lasts until 2027. They'd only have to worry about the anhydrous polymorph patents that last until 2024/25. However, the downside of pursuing an anhydrous form is that the FDA may push back on the appropriateness of their relying on Celgene's clinical trials, which were based on a hemihydrate form. On the other hand, the benefit to Actavis of pursuing a hemihydrate form is that they should have an easier time satisfying the FDA that their form is bioequivalent to Celgene's (although that may still be a difficult thing to do, even if not as difficult as getting an anhydrous form approved). However, the downside to pursuing a hemihydrate form is that they then have to contend with the hemihydrate polymorph patent that lasts until 2027. This is why it makes sense to me that Actavis may be pursuing both forms, at least for now, to see which one can get to market (by satisfying the FDA and not treading on any patents) first.
Perhaps the biggest development from Thursday's hearing was the concession by Actavis to Celgene's proposed constructions for the '740 and '569 methods of use patents, which last until April 2023 and October 2023, respectively. By consenting to Celgene's proposed constructions of those patents, Actavis' ability to argue their generic will not infringe those patents is substantially decreased. To put it another way, Celgene's lawyers would not propose a construction of those patents that did not make it as easy as possible to prove infringement later in the case. I mean, they're good lawyers, not idiots.
So, when Actavis threw in the towel on trying to argue those patents were narrower than what Celgene argued, that signaled to me Actavis is going to put all its effort on those patents to trying to prove they are invalid. This is what Actavis has already done with several other of the patents in the case, the '501, '720, '976, '977, '784 and '886 patents, which Actavis has stipulated it infringes. Actavis' only hope to win on those patents is to try to prove they are invalid, something they must do with clear and convincing evidence, a very high hurdle indeed. To defeat any argument of invalidity, Celgene only needs to raise enough of a defense of the patents to leave the judge with some doubt as to whether they are invalid. Even if she thinks the patents may be invalid, if she's not convinced, she must uphold them due to the presumption of validity and the clear and convincing evidence burden that is put on a patent challenger in court.
At this admittedly early stage of the case, absent a settlement between the parties, I don't see Actavis being able to introduce an anhydrous generic form until September 2024 when Celgene's '219 and '598 anhydrous polymorph patents expire, because I expect the judge will not narrowly construe the claims in those patents that do not expressly say "Form A" to only cover Celgene's Form A. If the judge construes those claims in the way I expect, Actavis' anhydrous form will likely be determined to infringe those patents, leaving Actavis with only a very difficult invalidity challenge to attempt. I do, though, expect the judge will limit the '357 patent to only Form A, meaning Actavis will be found to not infringe its claims and, thus, not have to wait until January 2025 when it expires to launch an anhydrous generic form.
With that said, I wonder whether Actavis will be able to get FDA approval of an anhydrous form, given that Celgene's approved version is a hemihydrate, because the two may not be considered bioequivalent. I am no expert on such issues, so don't have an opinion. If, however, Actavis is unable to get an anhydrous form approved, I don't expect it will be able to introduce a hemihydrate generic form until April 2027 when Celgene's '800 hemihydrate polymorph patent expires because, again, I expect the judge will not limit that patent to only cover Celgene's exact Form B. Instead, as I describe above, I expect the judge will interpret the '800 patent as Celgene suggests, which in all likelihood will lead to a finding of infringement by Actavis' proposed generic hemihydrate form. That would mean Actavis' only hope of launching a generic hemihydrate form before expiration of the '800 patent would rest on making a successful invalidity challenge, something I think is unlikely.
However, to play Devil's Advocate, let's assume for a moment that I'm wrong, and Actavis is indeed successful at having the polymorph patents either ruled invalid or construed narrowly so that its proposed generic form (anhydrous or hemihydrate) is determined not to infringe. Even if that happens, I see little chance that Actavis will be able to offer any generic form Revlimid before October 2023, when the last method of use patent expires, because Actavis has now acquiescenced to Celgene's broad construction of those patents and therefore will have a very hard, if not impossible, case to make for non-infringement, and a similarly difficult case to make for invalidity.
Lastly, another interesting question to consider is whether Actavis' launch of a generic form of Revlimid in 2023, 2024, or 2027 would necessarily cause economic harm to Celgene. This brings me back to my point above about Celgene's continually applying for more and more patents relating to the drug as it keeps working to improve and further develop it. Hopefully, for the sake of patients at least, the company will make great strides in the next 9, 10 or 13 years with lenalidomide such that today's version of Revlimid becomes outdated and replaced by a next generation version of the drug that is markedly safer and/or more effective.
A new Revlimid 2.0 product would surely have its own set of patents that would extend out until the 2030s. If Celgene can launch such a Revlimid 2.0 product before Actavis launches its generic form of today's Revlimid 1.0, then doctors may not switch away from Celgene's better Revlimid 2.0 product to substitute Actavis' inferior Revlimid 1.0 generic, even if there is cost savings, because the improved efficacy and safety make the price premium worthwhile. This is typical life cycle management in the pharma space and I expect Celgene is already working on ways to achieve it. So, one can not assume that an Actavis generic launch a decade from now will crush Celgene's revenues.
Disclaimer and No Warranty: While I express many opinions in this article, the case is still at an early stage and there are a lot of developments yet to come. Thus, my opinions may change. I'm simply giving here what my opinions are today given what I know, what I've read, and what I've observed.
I hope this overview of Celgene's Revlimid patents, the Markman process, and the issues discussed at Thursday's Markman hearing is helpful. If you have questions or issues, please feel free to leave a comment here or send me an email (dan[at]ravicher.com). I enjoy discussing these issues immensely.
Disclosure: I am long CELG. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: Due to the highly dynamic nature of legal matters in general, and the Markman hearing process in particular, I may change my position at any time for any reason. My position is disclosed as of publication of this article.
