Enbrel® (entanercept) is Amgen’s blockbuster, anti-inflammatory biologic drug indicated for treatment rheumatoid arthritis, psoriasis, and other inflammatory conditions. Enbrel® was first approved by the FDA in 1998, and yet still faces no “generic” or biosimilar competition. The drug has enjoyed a monopoly for close to 20 years.
Amgen holds two patents purportedly covering its fusion protein, entanercept. Those patents are important to Enbrel® because they will not expire for another 10 years. The problem with those patents, however, is that they do not explicitly describe the fusion protein that they claim. A recent Federal Circuit decision, also involving Amgen, may exacerbate that problem because it tightened the standard for satisfying the written description requirement for antibody patents.
In the case, Amgen v. Sanofi, the Court vacated an injunction Amgen obtained against a competing drug to its new PCSK9-inhibitor. The Court’s decision turned on a finding that the jury was improperly instructed on the criteria for invalidating a patent directed to an antibody for lack of written description.
Thus, will the precedent recently established in Amgen’s PCSK9 case doom the validity of its patents covering Enbrel®? There are likely two ways that the decision in Amgen v. Sanofi made a validity challenge to Enbrel®’s patents easier.
Amgen’s Entanercept Patents
In September 2015, Sandoz filed an aBLA (abbreviated Biologics License Application) with FDA for approval to sell a biosimilar version of Enbrel®. Under the BPCIA (Biologics Price Competition and Innovation Act), Sandoz’s aBLA triggered a lawsuit commenced by Amgen in February 2016. In the suit, brought in the District of New Jersey, Amgen asserted five patents. Three of the patents (the ‘225, ‘605 and ‘631) essentially cover psoriasis indications for the entanercept antibody. They expire in approximately 2019, and thus, are not critical to Enbrel®’s continued monopoly.
By contrast, two other patents cover the entanercept protein itself (U.S. Patent No. 8,063,182) or a method of producing it (U.S. Patent No. 8,163,522). These two patents are very important to Amgen’s fight against Sandoz because they do not expire until 2028/2029. That means that if Amgen can successfully enforce the ‘182 and ‘522 patents, it stands to gain another 10 years on its Enbrel® monopoly. That would mean a 30 year monopoly since first FDA approval.
Enbrel® (entanercept) is a TNF inhibitor. The drug is indicated for treatment of inflammatory conditions by targeting Tumor Necrosis Factor (TNF), which is an inflammation causing substance. In healthy individuals, excess TNF is naturally blocked, but patients suffering from inflammatory conditions, such as rheumatoid arthritis, experience an excess of TNF.
Entanercept is a fusion protein, which fuses a fragment of the TNF receptor to the human immunoglobin G antibody (IgG). More specifically, entanercept constitutes a soluble fragment of the 75 kilodalton human tissue TNF receptor that is fused to the hinge, CH2-CH3 region of the heavy chain of IgG. The purported benefits of Enbrel®’s fusion protein is that fusing a receptor with IgG proteins yields increased affinity for the TNF antigen. The TNFr:IgG protein dimerizes, and therefore displays two copies of the receptor, thus increasing affinity. In addition, the half-life of the fusion protein is also purportedly much greater than the TNF receptor alone.
Prosecution of Amgen’s Entanercept Patents
Coherus Biosciences recently filed IPRs attacking the validity of the ‘182 and ‘522 patents. The petition states that the ‘182 and ‘522 patents never specifically describe a fusion protein comprising a soluble fragment of the 75 kDa TNFR and the hinge-CH2-CH3 region of a human IgG, and contains no description or examples describing etanercept. (IPR2017-2066, Paper 1 at 10; IPR2017-1916, Paper 1 at 10). The suggestion is that the patents suffer from inadequate written description. Coherus does not actually argue in its IPRs for invalidating the patents for lack of written description. Yet, that is likely because that type of argument is precluded in an IPR, and the ‘182 and ‘522 patents are ineligible for Post-Grant Review, where those arguments would otherwise be permitted.
Written description was nonetheless challenged during prosecution of the ‘182 and ‘522 patents. Both patents were prosecuted in parallel. The ‘182 patent was only allowed after an appeal to the Board, and the applicants relied upon the outcome of that appeal to gain allowance of the ‘522 patent as well. Although the patents were allowed over written description rejections appealed to the Board, that Board decision was not likely decisive enough to preclude a challenge from Sandoz.
The patent’s specification does not actually disclose the full-length of the 75 kD TNF receptor (p75 TNF receptor.) Rather, it discloses a fragment of that receptor in FIG. 4 (specifically, amino acids 49-439 of the 439-amino-acid receptor.) During prosecution, the Examiner argued that because this was the only p75 TNF receptor fragment disclosed, the claims lacked written description for anything broader. That would include a fusion protein based on a p75 TNF receptor fragment including more than what is disclosed in FIG. 4. And since the claims can be construed to include a “vast genus” of fragments, ranging in size from the entire extracellular domain of p75 TNF receptor (amino acids 1-235) to a single amino acid, the Examiner asserted they lacked adequate written description.
In response, the applicants argued that existing knowledge in the art would have counseled persons of skill that there were existing constraints on the size of the claimed fragment. For instance, to successfully bind TNF, applicants pointed to references teaching that certain portions of amino acids 10-54 must be included and at least a portion of amino acids 142-162 must be included. Together, this indicated that at least amino acids 54-142 must be included. All together, applicants argued that this shows that the variations of claimed fragments are not a “vast genus.”
In its decision, the Board essentially side-stepped the issue. It held that because the entire p75 TNF receptor sequence was known in the art, the claims were adequately described, even though FIG. 4 was only a portion of that receptor sequence. The Board did not necessarily address the issue of whether the patent disclosed sufficient examples to be representative of the genus of p75 TNF receptor fragments that are claimed. Sandoz can thus still likely attack the validity of the entanercept patents on this ground.
The Amgen v. Sanofi Decision
In Amgen’s case against Sanofi and Regeneron, Amgen asserted several patents covering its PCSK9-inhibitor. Regeneron argued that the Amgen’s patents were essentially functional claims, namely they described the claimed antibody not by what is (its amino-acid structure,) but rather by what it does (binds to PCSK9 at certain residues). The patent specifications only disclosed two example antibodies that satisfy the claims, and Regeneron argued they were not representative of the broadly claimed genus. Regeneron attempted to show the jury its own protein, so as to illustrate how different it was from the two disclosed examples. The District Court barred that evidence, and the jury found Regeneron failed to prove the patents were invalid. The jury instructions also suggested that characterization of the antigen itself may satisfy written description for claims directed to an antibody, known as the “newly characterized antigen” test.
On appeal, the parties disputed the propriety of the “newly characterized antigen” test. PTO Guidelines first published in 2000 indicated that functional claiming of an antibody (e.g., an antibody capable of binding to antigen X) is adequately described where the specification fully characterizes the protein X, even if the patent lacks examples of any such antibodies. The guidelines themselves purportedly referenced a 1976 immunology text, and they appear to have presumed that production of antibodies for well-characterized antigens would be conventional or routine in light of existing antibody technology.
In two subsequent cases, Enzo Biochem, Inc. v. Gen-Probe Inc., 323 F.3d 956, 960 (Fed. Cir. 2002) and Noelle v. Lederman, 355 F.3d 1343, 1349 (Fed. Cir. 2004), the Federal Circuit suggested tacit endorsement, albeit in dicta, of the “newly characterized antigen” test. The suggestion was that fully describing an antigen may be sufficient to describe and claim an antibody that binds to it.
In a more recent case, Centocor Ortho Biotech, Inc. v. Abbott Labs., 636 F.3d 1341 (Fed. Cir. 2011), the Court finally examined the “newly characterized antigen” test in more detail. In doing so, the Court questioned its propriety. The Court stated that it is routine and conventional to produce antibodies for some antigens. For instance, it may be routine to produce antibodies for a known protein “simply by injecting that protein into a host animal that is a different species . . . [and] [t]he host generates antibodies to the foreign protein.” 636 F.3d at 1351 n.4. The Court noted, however, that the same may not be true for human antibodies. For one thing, producing human antibodies in this way may be unethical. Putting that aside, the Court pointed out that some antibodies are “self” proteins, and a human host may not produce effective antibodies against the antigen. Id. Indeed, in Centocor, which also addressed the TNF protein, i.e., the antigen for Enbrel®, the Court noted that the TNF protein is such a “self” protein.
More recently, in Amgen v. Sanofi, the Federal Circuit once against addressed the “newly characterized antigen” since the jury instruction in the underlying District Court case appeared to endorse it. The Court reiterated that adequate written description must “contain enough information about the actual makeup of the claimed products . . . .” The Court simultaneously suggested that the “newly characterized antigen” test “flouts” section 112 because it “allows patentees to claim antibodies by describing something that is not the invention, i.e. the antigen.” The Court concluded that for written description of an antibody to be adequate when presented with “functional” terminology, there must be an established correlation in the art between structure and function.
For instance, citing to Centocor, the Court analogized an antigen and antibody to a lock and a key. For an antigen where there is only a finite number of binding antibodies, discovering those antibodies may be routine and conventional, and description of the antigen alone may be sufficient. By contrast, for antigens with millions of keys, or millions of potentially binding antibodies, description of the antigen and even a couple of examples may be far from sufficient.
Amgen v. Sanofi’s Impact on the Enbrel Patents
In Amgen v. Sanofi, the takeaway was, for functionally claimed antibodies, the state of the art must show the correlation between the structure of the antibody and the function of binding to an antigen to satisfy §112. The proteins of the ‘182 and ‘522 patents are claimed with functional terminology, because they recite, “human tumor necrosis factor (TNF)-binding soluble fragment”. On the other hand, they are not entirely functional. They do recite the specific domains of IgG, and they do recite that the claimed protein “comprises” a partial sequences for the 75-kDa TNFR.
Yet, written description will still likely haunt these patents, and Sandoz is likely challenge them on that basis. According to the Coherus IPRs, the patents never specifically describe a fusion protein comprising a soluble fragment of the 75 kDa TNFR and the hinge-CH2-CH3 region of a human IgG. They also purportedly contain no description or examples describing etanercept. Rather, they describe how to make a fusion protein from IgG and the p55 TNF receptor, which they suggested should be sufficient to describe the claimed fusion of IgG and the p75 TNF receptor. Focusing on the claimed p75 fragment alone, the patent only discloses FIG. 4 (amino acids 49-249). Yet, during prosecution, the applicants nevertheless argued that the scope of the claims “can include additional sequences beyond Fig. 4.” (Application No. 08/444,790, Reply Brief, May 26, 2009 at 12).
Given all this, there are two aspects of the decision in Amgen v. Sanofi that have likely bolstered a written description challenge against the entanercept patents.
First, Amgen v. Sanofi held that for “functional” terminology in antibody patents, the state of the art must show established correlation between structure and function. If Sandoz can show that it was neither routine nor conventional to deduce 75p TNF receptor fragments that bind to TNF and encompassed in the claims, that could jeopardize the adequacy of the patents’ written description. For one thing, in Centocor, the Federal Circuit observed that it would not likely be routine to produce antibodies for TNF. Also, during prosecution of Amgen’s entanercept patents, applicants attempted to show that, based on the state of the art, persons of skill knew which parts of the p75 TNF receptor bind to TNF. The relied upon two references to show that. In one case, they argued that the state of the art knew that certain amino acids (portions of 10-54) needed to be included in the fragment to bind to TNF. Applicants argued, thus, the patents did not cover a “vast genus” that was not adequately described. But the reference cited by applicants (Chan 2000) was dated 10 years after the priority date of the patent. Thus, it is an open question whether it actually reflected the state of the art 10 years earlier at the patents’ priority date.
Second, a lot will may come down to how different Sandoz’s protein is compared to Amgen’s protein. In Amgen v. Sanofi, Regeneron tried to show the jury its own protein, so that it could illustrate how different it was from the example proteins disclosed in Amgen’s patents. The District Court excluded that evidence on the ground that Regeneron’s protein post-dated the filing date of Amgen’s patents. On appeal, the Federal Circuit held that was error, and remanded the case for a new trial, wherein Regeneron will be permitted to show its protein to the jury. Likewise, in the Enbrel® case, Sandoz will likely be permitted to use its own protein to show that the examples (or lack thereof) of TNF receptor fragments disclosed in the ‘182 and ‘522 patents are not, in fact, representative of the claimed genus.
The case between Amgen and Sandoz is scheduled to go to trial in April 2018. It is likely that Sandoz stakes out its stance on the written description of Amgen’s Enbrel® patents before then, and the impact of Amgen v. Sanofi may become more clear.