Prost, Chief Judge.
This is a consolidated appeal from two Patent Trial and Appeal Board ("Board") decisions in inter partes reviews ("IPRs") of U.S. Patent No. 8,679,487 ("the '487 patent"), owned by Immunex Corp. ("Immunex"). Sanofi-Aventis U.S. LLC, Genzyme Corp., and Regeneron Pharmaceuticals, Inc. (collectively, "Sanofi") challenged the '487 patent, which covers isolated human antibodies that bind the human interleukin-4 receptor. The Board invalidated all challenged claims in one of the IPRs, No. IPR2017-01884. Immunex appeals, contesting the construction of the claim term "human antibodies." In the other IPR, No. IPR2017-01879, involving a subset of the same claims, the Board did not invalidate the patents for reasons of inventorship. Sanofi appeals, contesting the Board's inventorship determination. We consolidated the cases in the nature of an appeal and a cross-appeal. For the reasons below, we agree with the Board's claim construction in No. IPR2017-01884 (here, "the appeal"). Accordingly, we affirm that invalidity decision. Because this leaves valid no claims at issue in the second IPR, we dismiss Sanofi's inventorship appeal from No. IPR2017-01879 (here, the "cross-appeal").
The '487 patent is directed to antibodies that bind to the human interleukin-4 ("IL-4") receptor, the resulting inhibition of which is significant for treating various inflammatory disorders, such as arthritis, dermatitis, and asthma. See '487 patent col. 3 ll. 15-31; J.A. 3-4.
Claim 1 reads:
'487 patent (emphasis added). This appeal concerns what "human antibody" means in this patent.
First, the relevant science. Antibodies are proteins. Like all proteins, they are composed of numerous individual amino acids chained together in a particular sequence. Antibodies are roughly Y-shaped, made of four chains—two "heavy" and two "light." Each chain can be further divided into a "variable region" and a "constant region." And each variable region contains three relatively small "complementarity-determining regions" (CDRs) situated at the tips of the Y. The remainder of the variable regions are the "framework regions."
Particular antibody regions have particular biological implications. For instance, it is primarily the CDRs that give an antibody its ability to bind selectively to specific targets (i.e., antigens), despite making up just a sliver of its structure. See J.A. 1501, 7042-43. To that end, an antibody's exact amino acid sequence determines what the antibody binds to, which affects the antibody's therapeutic usefulness. The amino acid sequence of an antibody also determines whether the human immune system recognizes and rejects it as "non-human." Amino acid sequences that are human in origin—that is, sequences "consistent with the amino acid sequences of antibodies produced naturally by the human immune system," see Appellant's Br. 4—can avoid triggering immune responses.
Early efforts at therapeutic antibody development started with mice. For example, researchers could inject a mouse with an antigen, the mouse would generate antibodies to the antigen, and those antibodies would be harvested. In that case, the entire amino acid sequence was murine (i.e., from mice). These antibodies, disappointingly, tended to plague patients with "undesirable and harmful immune reactions." See Appellant's Br. 7-8. Too much of each antibody was "mouse" in origin, to the consternation of the human immune system.
Through various techniques, the proportion of an antibody that is recognized as "mouse" can be decreased. In "chimeric" antibodies, for instance, the constant regions tend to be human in origin, and the variable regions, including the CDRs, tend to be nonhuman—making the antibodies' amino acid sequences mostly human in origin. Appellant's Br. 8-9. In "humanized" antibodies, only the CDRs are nonhuman —the antibodies' amino acid sequences, including the portions responsible for immune reaction, are almost entirely human in origin.
Here, some of the disclosed embodiments are "partially human" and some are "completely human." E.g., '487 patent col. 19 ll. 38-44, col. 21 ll. 6-14. Among the former, the specification's embodiments specifically include humanized and chimeric antibodies. Id. at col. 18 ll. 36-37, col. 19 ll. 21-37.
The claim construction dispute is this: in the context of this patent, must a "human antibody" be entirely human? Or may it also be "partially human," including "humanized"?
Amid infringement litigation, Sanofi filed three IPR petitions challenging claims 1-17 of the '487 patent. Two were instituted.
In one final written decision, the Board concluded that claims 1-17 were unpatentable as obvious over two references, Hart and Schering-Plough. Sanofi-Aventis v. Immunex, No. IPR2017-01884, Paper 96, 2019 WL 643041 (P.T.A.B. Feb. 14, 2019) ("Final Written Decision").
Hart describes a commercially available murine antibody that purportedly meets all the limitations of claim 1—except that it is fully murine, not human at all. Final Written Decision, 2019 WL 643041, at *7-8. But Schering-Plough teaches humanizing such murine antibodies by "grafting" their CDRs onto an otherwise fully human antibody. Id. Sanofi therefore argued that the claims were obvious in light of the humanized antibody that would result from this combination. Further, Sanofi argued in a second obviousness ground that the gap between "humanized" and "fully human" could be closed using the teachings of a third reference, Hoogenboom. J.A. 1095. The Board reached only the first ground, finding that the "humanized" antibody met its construction of "human antibody." Final Written Decision, 2019 WL 643041, at *9, *12. On appeal, Immunex insists only that the Board erred in this construction.
In the second final written decision, the Board concluded that Sanofi had not shown by a preponderance of the evidence that claims 1-14, 16, and 17 were anticipated by one of Immunex's own publications. Sanofi-Aventis v. Immunex, No. IPR2017-01879, Paper 88, 2019 WL 643024 (P.T.A.B. Feb. 14, 2019). Sanofi appealed, contending that the Board erred in determining that the disclosure was not § 102(e) prior art "by another." We consolidated Immunex's appeal and Sanofi's appeal in the nature of an appeal and a cross-appeal, respectively. See Order (July 10, 2019), ECF No. 21.
We have jurisdiction under 28 U.S.C. § 1295(a)(4)(A).
First, we consider the applicable claim construction standard in light of a post-briefing terminal disclaimer.
After appellate briefing was complete, Immunex filed with the Patent and Trademark Office ("PTO") a terminal disclaimer of its patent. The PTO promptly accepted it, and Immunex's patent therefore expired on May 26, 2020, just over two months before oral argument.
Immunex then filed a citation of supplemental authority under Federal Rule of Appellate Procedure 28(j), apprising us of (but not explaining the reason for) its terminal disclaimer and asking us to change the applicable claim construction standard. See Citation of Suppl. Authority (Apr. 10, 2020), ECF No. 66. Sanofi and the PTO insist that Immunex has waived the Phillips issue. We need not reach waiver, determining for the following reasons that the BRI standard applies.
Today, in all newly filed IPRs, the Board applies the Phillips district-court claim construction standard. 37 C.F.R § 42.100(b) (2020); Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc).
We have also applied the Phillips standard when a patent expired on appeal. See PTO Resp. Letter (Apr. 30, 2020), ECF No. 72 (citing Apple Inc. v. Andrea Elecs. Corp., 949 F.3d 697, 707 (Fed. Cir. 2020)). But we do not read Andrea Electronics to mean that whenever a patent expires on appeal, at any time and for any reason, Phillips applies. In Andrea Electronics, the patent's term expired as expected. It was not cut short by a litigant's terminal disclaimer. And, importantly, the expected expiration happened before appellate briefing began. The parties knew this at the outset, as the expiration date was part of the record before the Board, and were able to fully brief the consequences. Not so here, where the patentee shortened the term abruptly after the parties had already fully briefed claim construction under the BRI standard.
This court "shall review the decision from which an appeal is taken on the record before the Patent and Trademark Office." 35 U.S.C. § 144. Our predecessor court has refused to consider terminal disclaimers filed after the Board's decision. In re Thorington, 418 F.2d 528, 533-34 (CCPA 1969); In re Heyl, 379 F.2d 1018 (CCPA 1967). In this situation, we do the same.
Accordingly, in this case we will review the Board's claim construction under the BRI standard.
Next, we address the Board's claim construction. Immunex contends that the Board erred by construing the term "human antibody" to encompass not only "fully human" but also "partially human" antibodies.
Claim 1 of the '487 patent recites a "human antibody." The Board determined that the BRI of "human antibody" "includes both fully human and partially human antibodies." Final Written Decision, 2019 WL 643041, at *7. As relevant to its obviousness rejection, the Board's construction includes "humanized" antibodies. Id. at *9. According to Immunex, however, "humanized" is not "human." For the reasons below, we disagree with Immunex and agree with the Board.
We review the Board's claim construction de novo and any underlying factual findings for substantial evidence. Kaken Pharm. Co. v. Iancu, 952 F.3d 1346, 1350 (Fed. Cir. 2020) (citing Teva Pharms. USA, Inc. v. Sandoz, Inc., 574 U.S. 318, 135 S.Ct. 831, 190 L.Ed.2d 719 (2015); Wasica, 853 F.3d at 1278). In this case, claim terms are given their "broadest reasonable construction in light of the specification of the patent." 37 C.F.R. § 42.100(b) (2016); Cuozzo Speed Techs., LLC v. Lee,
We review the Board's claim construction according to the Supreme Court's decision in Teva. Accordingly, we review the Board's evaluation of the intrinsic record de novo. See Teva, 574 U.S. at 331, 135 S.Ct. 831; Knowles Elecs. LLC v. Cirrus Logic, Inc., 883 F.3d 1358, 1361-62 (Fed. Cir. 2018). But "[w]e review underlying factual determinations concerning extrinsic evidence for substantial evidence." In re Cuozzo Speed Techs., LLC, 793 F.3d 1268, 1279-80 (Fed. Cir. 2015), aff'd sub nom. Cuozzo Speed Techs., LLC v. Lee, ___ U.S. ___, 136 S.Ct. 2131, 195 L.Ed.2d 423 (2016); Teva, 574 U.S. at 331-32, 135 S.Ct. 831; Knowles, 883 F.3d at 1361-62.
First, we turn to the intrinsic record. Personalized Media Commc'ns, LLC v. Apple Inc., 952 F.3d 1336, 1340 (Fed. Cir. 2020) ("When construing claim terms, we first look to, and primarily rely on, the intrinsic evidence, including the claims themselves, the specification, and the prosecution history of the patent, which is usually dispositive." (quoting Sunovion Pharms., Inc. v. Teva Pharms. USA, Inc., 731 F.3d 1271, 1276 (Fed. Cir. 2013))). As discussed below, the intrinsic evidence supports the correctness of the Board's construction.
We begin claim construction by looking to the language of the claim itself. Allergan Sales, LLC v. Sandoz, Inc., 935 F.3d 1370, 1374 (Fed. Cir. 2019). But nothing in the claim's language restricts "human antibodies" to those that are fully human. This is not surprising: antibodies, amid a rapidly evolving scientific background, are a frequent subject of claim-construction disputes that stretch beyond plain meaning. E.g., Baxalta Inc. v. Genentech, Inc., 972 F.3d 1341, 1345-49 (Fed. Cir. 2020) (construing "antibody"); UCB, Inc. v. Yeda Rsch. & Dev. Co., 837 F.3d 1256, 1259-61 (Fed. Cir. 2016) (construing "monoclonal antibody"); Biogen Idec, Inc. v. GlaxoSmithKline LLC, 713 F.3d 1090, 1095-97 (Fed. Cir. 2013) (construing "anti-CD20 antibody"). Nor is the claim context helpful, as the dependent claims provide no further guidance.
Accordingly, we consult the rest of the intrinsic record. Indeed, the specification is key—it is "highly relevant to the claim construction analysis" and the "single best guide to the meaning of a disputed term." Phillips, 415 F.3d at 1315 (quoting Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed. Cir. 1996)); see also In re Translogic Tech., Inc., 504 F.3d 1249, 1256-58 (Fed. Cir. 2007) (endorsing Phillips "best practices" in the BRI context).
Many patentees do expressly define "human antibody." See, e.g., Abbott GMBH & Co. v. Centocor Ortho Biotech, Inc., 870 F.Supp.2d 206, 247 (D. Mass. 2012) (noting express definition of "human antibody"). Here, however, we are without an express definition. But the usage of "human" throughout the specification confirms its breadth.
The specification contrasts "partially human" with "fully" or "completely human." E.g., '487 patent col. 19 ll. 41-44, col. 20 ll. 57-60, col. 21 ll. 1-2. For example, the specification states that "[a]ntibodies of the invention include, but are not limited to, partially human (preferably fully human) monoclonal antibodies." Id. at col. 20 ll. 57-60. And elsewhere, it notes that "[t]he desired antibodies are at least partially
Still further, the specification reads:
'487 patent col. 19 ll. 38-44 (emphases added). Again, here the specification makes clear that "human antibodies" is a broad category encompassing both partially and completely human antibodies.
Immunex disagrees with this reading: it protests that the phrase "the antibodies," as italicized above, refers not to "human antibodies"—one sentence back—but to "antibodies directed against the IL-4R polypeptide" —two sentences back. We are unpersuaded. Immunex's proposed interpretation would contort the logical and grammatical reading of the passage.
The specification also repeatedly clarifies that some "human" antibodies are "fully human":
'487 patent col. 21 ll. 6-13 (emphases added); see also id. at col. 43 ll. 26-27 ("[Antibody] 12B5 was determined to be an IgG1 antibody, and to be fully human." (emphasis added)). If "human antibodies" were already understood to mean "fully human," no clarification would be necessary. This usage confirms that a reader would take "human monoclonal antibodies" to be broader.
Consistent with this usage, the abstract and the summary each simply refer to "human" antibodies. See '487 patent Abstract ("Particular antibodies provided herein include human monoclonal antibodies generated by procedures involving immunization of transgenic mice. Such human antibodies may be raised against human IL-4 receptor."); id. at col. 2 ll. 42-46 ("Particular antibodies provided herein include human monoclonal antibodies generated by procedures involving immunization of transgenic mice. Such human antibodies may be directed against human IL-4 receptor, for example.").
Accordingly, the language of the specification confirms a broadest reasonable interpretation of "human antibodies" that includes those that are partially human— including "humanized" antibodies.
Next, we turn to the prosecution history. The Board found the prosecution history to be "equivocal, at best." Final Written Decision, 2019 WL 643041, at *6. Immunex
First, we note that Immunex used both "fully human" and "human" within the same claim set in another patent application in the same family.
Second, "there is a strong presumption against a claim construction that excludes a disclosed embodiment." Nobel Biocare Servs. AG v. Instradent USA, Inc., 903 F.3d 1365, 1381 (Fed. Cir. 2018) (quoting In re Katz Call Processing Pat. Litig., 639 F.3d 1303, 1324 (Fed. Cir. 2011)). We noted above that the specification's embodiments include partially human antibodies—both humanized and chimeric. And the prosecution history here illustrates why the presumption against their exclusion from the claims is not overcome.
As initially filed, claim 1 recited simply "an isolated antibody." J.A. 2409. The word "human" was added later, at the same time that dependent claim 11, which recited "a human, partially human, humanized, or chimeric antibody," was canceled.
Immunex suggests instead that the amendment "surrender[ed]" the partially human embodiments. E.g., Appellant's Br. 26; see also id. at 36 (arguing that Immunex "unambiguously amended the claims to remove antibodies that are not fully human"). We disagree. "Because the claim language does not require the exclusion of those embodiments, and there is no basis in the intrinsic record for excluding them," Immunex "has not overcome [the] presumption" against their exclusion. Nobel Biocare, 903 F.3d at 1381; see also, e.g., Baxalta, 972 F.3d at 1348 ("[D]isavowal must be clear and unmistakable.").
As the Board noted, "human" was added to overcome an anticipating reference that disclosed nonhuman murine antibodies—a far cry from "humanized" antibodies. Final Written Decision, 2019 WL 643041, at *5; Appellee's Br. 45-46.
We agree with the Board that nothing indicates that Immunex added "human" to limit the scope to fully human. There was no apparent need to do so in light of the rejection, and no evidence that anyone understood Immunex to be casting aside subject matter that was not at issue. Final Written Decision, 2019 WL 643041, at *5-6.
Third, in a post-amendment office action, the examiner expressly wrote that the amended "human" antibodies encompassed "humanized" antibodies. J.A. 2211. Immunex suggests without substantiation that this was a "copy and paste error." See Appellant's Br. 36. But if so, Immunex made no effort to disabuse the examiner of this understanding. And, while hardly dispositive, this uncontested characterization is consistent with the Board's construction.
Accordingly, the prosecution history also supports the Board's construction.
Next, we address the role of extrinsic evidence in the Board's construction.
Immunex argues that the Board "failed to establish how a person of ordinary skill in the art would have understood the term `human antibody.'" Appellant's Br. 47. That is, Immunex contends that the Board did not adequately consult its extrinsic evidence —its experts' testimony, product catalogs, and a selection of journal articles— to establish whether "human antibody" had an established meaning to a person of ordinary skill in the art, independent of the specification. We disagree.
It is true that we seek the meaning of claim terms from the perspective of the person of ordinary skill in the art. The key, however, is that we look to how that person would have understood a term in view of the specification. See, e.g., In re Am. Acad. of Sci. Tech Ctr., 367 F.3d 1359, 1364 (Fed. Cir. 2004) ("[C]laims are to be given their broadest reasonable interpretation consistent with the specification, and claim language should be read in light of the specification as it would be interpreted by one of ordinary skill in the art." (cleaned up)).
While extrinsic evidence may sometimes illuminate a well-understood technical meaning, Teva, 574 U.S. at 331-32, 135 S.Ct. 831, that does not mean that litigants can introduce ambiguity in a way that disregards language usage in the patent itself. The patent drafter controls the content of the specification, writes the claims, and responds to office actions. The drafter, then, is in the best position to anticipate ambiguity or questions of scope and to write the patent accordingly. Indeed, we give the intrinsic evidence "priority," see, e.g., Knowles, 883 F.3d at 1361-62, over extrinsic evidence with which it is "inconsistent," Tempo Lighting, Inc. v. Tivoli, LLC, 742 F.3d 973, 977 (Fed. Cir. 2014) (emphasis omitted); see, e.g., Celgene Corp. v. Peter, 931 F.3d 1342, 1350-51 (Fed. Cir. 2019) (holding that the Board
Immunex's extrinsic evidence included the testimony of its two experts, who discussed their views in light of a handful of journal articles, catalogs, and other documents. The Board cited this evidence, and clearly considered it. Final Written Decision, 2019 WL 643041, at *6-7. But the Board found nothing credible to call its interpretation into question. To the contrary, it credited a prior art reference and expert testimony that were squarely consistent with "humanized" being understood as a subset of "human." See id. at *6 (citing J.A. 5099-100 ¶¶ 9-10 as "Ex. 1477").
At any rate, the intrinsic evidence here decides the issue. Extrinsic evidence may be of assistance if the intrinsic record is equivocal, leaving us looking for further guidance. See Helmsderfer v. Bobrick Washroom Equip., Inc., 527 F.3d 1379, 1382 (Fed. Cir. 2008). But here, the meaning of "human antibody" as discerned from the intrinsic evidence squarely conflicts with the meaning that Immunex would distill from its selected extrinsic evidence. Id. ("A court may look to extrinsic evidence so long as the extrinsic evidence does not contradict the meaning otherwise apparent from the intrinsic record."). Accordingly, the intrinsic record trumps.
Finally, we turn to the matter of the Board's departure from an earlier court's claim construction.
In litigation that prompted this IPR, a district court construed "human" to mean "fully human" only. See Immunex Corp. v. Sanofi, No. CV 17-02613 SJO, 2018 WL 6252460, at *12-14 (C.D. Cal. Aug. 24, 2018). That claim construction order issued two months before the oral hearing in this IPR, and the parties discussed it in their briefing and at oral argument before the Board.
The Board did not adopt the district court's construction. After conducting a full analysis of the parties' arguments, the Board concluded that it reached a different interpretation "based on the broader applicable case law." Final Written Decision, 2019 WL 643041, at *7.
Immunex chides the Board for not explaining more fully its departure from the district court's narrower Phillips-based construction. Citing Power Integrations, Inc. v. Lee, 797 F.3d 1318, 1326 (Fed. Cir. 2015), Immunex contends that the Board must explain in detail why, under a broader legal standard, it reaches a broader construction than a district court does.
The Board's misstep in Power Integrations, however, was not merely failing to explain the difference between a Phillips construction and the BRI. Rather, the Board there both "failed to acknowledge the district court's claim construction" and "devoted a substantial portion of its analysis" to an issue not raised by the parties,
Regardless, in Power Integrations we reiterated that the Board "is not generally bound by a previous judicial construction of a claim term." Id. at 1326; see also Mayne Pharma Int'l Pty. Ltd. v. Merck Sharp & Dohme Corp., 927 F.3d 1232, 1242 (Fed. Cir. 2019) ("[W]e are not persuaded that the Board erred in discounting the district court's construction because the court construed the claims under the narrower, Phillips standard."). And we emphasized that the Board need not "in all cases assess a previous judicial interpretation of a disputed claim term." Power Integrations, 797 F.3d at 1327. Rather, we require the Board to provide "reasoning in sufficient detail to permit meaningful appellate review." Id. And the Board's opinion was sufficiently detailed to permit meaningful appellate review. We conclude that the Board did not err by not saying more.
* * *
In summary, the Board's construction was correct. Accordingly, we affirm the Board's invalidity judgment predicated on that claim construction.
Last, we turn to Sanofi's cross-appeal. Sanofi had alleged in its petition for IPR2017-01879 that certain claims of the '487 patent were anticipated by the disclosure of mAb 6-2, an isolated human antibody, in an earlier publication of Immunex's. That reference, U.S. Patent Application Publication No. 2002/0002132 ("the '132 publication"), is within the same prosecution family as the '487 patent. But Sanofi contested the listed inventorship, insisting that mAb 6-2 was invented "by another"—namely, by research technician Norman Boiani, not the '487 patent's inventors —and therefore that the this disclosure was prior art under § 102(e). The Board disagreed, concluding that Mr. Boiani was not an inventor of mAb 6-2. Sanofi cross-appeals this determination.
Because we affirm the Board's invalidity judgment in the other IPR, which implicates the same claims, it is unnecessary to reach this issue.
We have considered the parties' other arguments but find them unpersuasive.