dissenting.
The only issue decided by the panel majority is the district court’s ruling of inequitable conduct, during patent prosecution.1 I respectfully dissent, for my colleagues apply incorrect law and add confusion to precedent.
To establish “inequitable conduct” in patent prosecution, both materiality and deceptive intent must be proved
“Inequitable conduct” arises when material references were intentionally withheld by the patent applicant in order to deceive or mislead the examiner into granting the patent. Both materiality and intent must be proved by clear and convincing evidence. Therasense, Inc. v. Becton, Dickinson & Co., 649 F.3d 1276, 1287 (Fed. Cir. 2011). Intent to deceive cannot be inferred; yet here, the district court inferred intent to deceive during prosecution and .invalidated the patent, as a sanction for purported attorney misconduct during this litigation.
The district court found that certain un-cited references were “but-for material” to patentability—although the court did not And the ’018 patent claims invalid on the substantive content of these references. The district court then declined to decide the question of specific intent to deceive the patent examiner. Instead, the court cancelled the scheduled trial on the question of intent, adopted an “inference” of intent, and held the’018 patent unenforceable on grounds of inequitable conduct as a sanction for Regeneron’s “litigation misconduct” relating to discovery and the privilege log during this litigation.
The panel majority, acknowledges that “the district court never held a second trial to determine if Regeneron acted with the specific intent to deceive the PTO during prosecution.” Maj. Op. at 1356. This absence of trial and trial findings on this critical issue cannot be substituted by inference.
Nor is the appellate role to scour the Appendix to fill the gap and make our own appellate finding of “intent to deceive.” Here, no evidentiary record was developed on intent to deceive, with no testimony and no opportunity for examination and cross-examination of witnesses. The panel majority instead engages in innuendo based on its careful selections from documents not admitted into evidence. The panel majority thus convicts Regeneron, its counsel, and its scientists, with no trial, no- evidence, and no opportunity to respond in their defense.
Materiality does not establish intent; deliberate withholding of but-for invalidating prior art, with the intent to deceive the examiner, must be established by clear and convincing evidence. The majority’s mechanism whereby dispositive facts are found for the first time on appeal, with no right of traverse by the affected party, is contrary to fundamental fairness and judicial process. If the panel majority indeed believes that the four “uncited” reférences are but-for material to ■ patentability, we should at least require trial of the question of intent.
Whether or not counsel’s discovery and privileye disputes were justifiable, invalidation of the patent is not an available remedy for such disputes
Instead of requiring proof of intent to deceive the examiner during patent prose-*1366ration, the panel majority upholds the district court’s “adverse inference” in light of “widespread litigation misconduct.” Maj. Op. at 1364. Misconduct during litigation— as the district court viewed counsel’s actions concerning discovery and the privilege log—cannot substitute for evidence of intent to deceive by withholding but-for material prior art during patent prosecution.
Precedent is long-standing, unambiguous, and binding. In Keystone Driller Co. v. General Excavator Co., 290 U.S. 240, 54 S.Ct. 146, 78 L.Ed. 293 (1933), the Court established that litigation misconduct can support the dismissal of the suit, whereas patent invalidity or unenforceability must be established on the law of validity or enforceability. Applying Keystone Driller, in Aptix Corp. v. Quickturn Design Systems, Inc., 269 F.3d 1369 (Fed. Cir. 2001), this court held that:
[T]he remedies for litigation misconduct bar the malfeasant who committed the misconduct. The property right itself remains independent of the conduct of a litigant.
Id. at 1375. This court elaborated:
Leaving the patent right intact, the Supreme Court repeatedly stressed that litigation misconduct bars the litigant. Again in Hazel-Atlas Glass Co. v. Hartford-Empire Co., 322 U.S. 238 [64 S.Ct. 997, 88 L.Ed. 1250] (1944), overruled on other grounds by Standard Oil Co. v. United States, 429 U.S. 17, 18 [97 S.Ct. 31, 50 L.Ed.2d 21] (1976), another instance of extreme litigation misconduct, the Supreme Court “require[d] that Hartford be denied relief,” but left the patent right intact. Id. at 251 [64 S.Ct. 997].
Id. We continued to explain that in order to invalidate the patent, the inequitable conduct must have occurred in patent prosecution:
Litigation misconduct, while serving as a basis to dismiss the wrongful litigant, does not infect, or even affect, the original grant of the property right.
Id. We concluded:
No case law from the Supreme Court or this court provides a basis for nullifying property rights granted by the United States when such property rights did not themselves accrue through inequitable conduct.
Id. at 1377.
The Aptix holding has been applied in trial forums across the nation. E.g., Kimberly-Clark Worldwide, Inc. v. First Quality Baby Prod., LLC, 2011 WL 679337, at *6 (E.D. Wis. Feb. 16, 2011) (“[Alleged litigation misconduct is not sufficient to support a counterclaim of unenforceablity of a patent.”); MedPointe Healthcare Inc. v. Hi-Tech Pharmacol Co., 380 F.Supp.2d 457, 467 (D.N.J. 2005) (“[Because the alleged misconduct involved conduct before the court and not before the patent office during the procurement of the patent, it does not taint the property right ab initio to render the patent unenforceable.”); Honeywell Int’l, Inc. v. Universal Avionics Sys. Corp., 398 F.Supp.2d 305, 311 (D. Del. 2005) (“If the wrongdoing occurs during the prosecution of the patent, in the furtherance of obtaining a patent right, then it can render the patent unenforceable. Alternatively, if unclean hands occurs during litigation, it bars any recovery by the offending party.”).
The panel majority dismisses Aptix as “inapposite,” Maj. Op. at 1364, because Regeneron was “accused ... of engaging in inequitable conduct during prosecution,” id. Our system of justice is bottomed upon proof, not upon bare accusation. Intent to deceive is not established by accusation and innuendo. It is only established by evidence. That evidence “must be sufficient to require a finding of deceitful intent *1367in the light of all the circumstances.” Therasense, 649 F.3d at 1290 (quoting Kingsdown Med. Consultants Ltd. v. Hollister Inc., 863 F.2d 867, 873 (Fed. Cir. 1988) (emphasis original)).
The panel majority also states that “the district court did not punish Regeneron’s litigation misconduct by holding the patent unenforceable.” Maj. Op. at 1364. However, the district court stated that it “impose[d] the sanction of an adverse inference as to the intent of Smeland and Murphy with regard to inequitable conduct during patent prosecution.” Dist. Ct. Op. at 595. A sanction, by definition, is punishment; here, in holding the patent unenforceable. This is a further departure from binding precedent, as equitable doctrines are not a source of a power to punish. Feltner v. Columbia Pictures Television, Inc., 523 U.S. 340, 352-53, 118 S.Ct. 1279, 140 L.Ed.2d 438 (1998); Tull v. United States, 481 U.S. 412, 422, 107 S.Ct. 1831, 95 L.Ed.2d 365 (1987) (“Remedies intended to punish culpable individuals, as opposed to those intended simply to extract compensation or restore the status quo, were issued by courts of law, not courts of equity.”).
In its attempt to the Supreme Court precedent or principles of equity underlying the holding. Nor does the panel majority cite a single case—at any level of the federal system—in which litigation misconduct was part of a finding of inequitable conduct. An unbroken line of precedent strictly limits the inequitable conduct inquiry to a patentee’s conduct before the examiner.
Aptix instructs that litigation misconduct in the infringement suit “does not infect, or even affect” the patent right. 269 F.3d at 1375. The panel majority errs in “infecting” its analysis- of inequitable conduct with counsel’s purported litigation misconduct years later in the infringement trial.
I also review the court’s treatment of the four purportedly withheld references, for they do not impart unpatentability to the claims, and thus are not but-for material.
The references cited by the examiner were fully explored during patent prosecution; the additional references do not add invalidating information
The ’018 patent is one of a family of patents directed to Regeneron’s Veloci-Gene technology, which uses quantitative assays to screen for DNA recombination events. During prosecution the examiner cited seven references, including U.S. Application 11/009,873 (“Lonberg”) and U.S. Patent No. 6,114,598 (“Kucherlapati”), and considered U.S. Patent No. 6,130,364 (“Ja-kobovits”). The examiner rejected all the claims of the ’018 application as anticipated by Lonberg, and obvious over Lonberg in combination with three other references, including a Briiggemann reference dated four years after the allegedly withheld Briiggemann reference, discussed post.
Lonberg was the examiner’s primary reference, and teaches the introduction of immunoglobulin transgenes into mouse cells. Lonberg specifically discloses “constructing” a transgene composed of at least one variable gene segment, one joining gene segment, and one constant region gene segment, preferably of human origin. Lonberg, [0031]2. These segments are “un-rearranged” in that they are not “rearranged as to encode a functional immuno-globulin light or heavy chain,” but are not in germline configuration. Id. The Lonberg transgene constructs may include regulatory sequences from either the host (i.e., murine) or a related animal, or from the exogenous (i.e., human) species. Id. at *1368[0033]. These transgenes are randomly integrated into the host (mouse) genome, id. at [0292], and the resulting animals are then crossed with “knockout” mice—i.e„ mice with a disrupted immunoglobulin locus, id. at [0296]. The result is that the cross-bred mice produce heterologous (i.e., non-host) antibodies.
Kucherlapati teaches methods of producing transgenic animals in which the host endogenous immunoglobulin locus is “substituted by a portion of, or an entire, xenogeneic immunoglobulin locus, or may have a xenogeneic immunoglobulin locus inserted into a chromosome of the host cell and an inactivated .endogenous immu-noglobulin region.” Kucherlapati, col. 3,11. 51-55. Kucherlapati teaches both random integration and targeted insertion of the immunoglobulin, locus. Such xenogeneic immunoglobulin loci are described as “human, constant and/or variable regions.” Id. at col. 5, 11. 51-54. Kucherlapati teaches that the xenogeneic locus “will be placed in substantially the same position as the analogous host locus, so that any regulation associated with the position of the locus will be substantially the same for the xenogeneic locus.” Id. at col. 10, 11, 51-55, As an example, Kucherlapati teaches retaining promoter and regulatory regions of the host DNA. Id. at col. 10, 1. 64-col. 11, 1.2.
The district court referred to Regener-on’s arguments before the European Patent Office about whether Kucherlapati was enabled. Dist. Ct. Op. at 577-78 (citing Merus’s expert). The panel majority cites Regeneron’s arguments about Kucherlapa-ti’s enablement in the prosecution of a different patent application, U.S. Application No. 13/719,819.3 Maj. Op. at 1355-56. However, argument of Kueherlapati’s en-ablement does not appear in the prosecution record of the ’018 application. “United States patents—even those only asserted as prior art in an invalidity defense—are presumed enabled.” Amgen Inc. v. Hoechst Marion Roussel, Inc., 314 F.3d 1313, 1354 (Fed. Cir. 2003). Kucherlapati was thus presumed enabled before the examiner.
The Jakobovits reference teaches the “use of Cre-mediated site-specific recombination for modifying immunoglobulin loci, for instance, to replace ¿11 or' a part of either the constant region or variable region of an antibody molecule.” Jakobovits, col. 1,11.11-14. That is, Jakobovits teaches a. method for targeted insertion at an im-munoglobulin locus.
The examiner in the “reasons for allowance”- stated that “the prior art does not teach or suggest a genetically modified mouse comprising, in its germline cells, human unrearranged variable region- gene segments inserted at an endogenous mouse immunoglobulin locus.” J.A. 531. No error has been ascribed' to this finding.
The purportedly withheld references were not more material than the cited references
None of the purportedly withheld references provides teachings more material than in the cited references. No purportedly withheld information was identified by the district court or the panel majority to teach a missing limitation or provide a motivation missing in the art.
Despite this failure, the district court held that the following uncited references and information were material to patenta-bility:
1. Marianne Brüggeman & Michael S. Neuberger, “Strategies for Expressing Human Antibody Repertoires in Transgenic Mice,” 17(8) Review Immunology Today 391 (1996) (“Brüg-geman”)
*13692. Shinsuke Taki et al., “Targeted Insertion of a Variable Region Gene into the Immunoglobulin Heavy Chain Locus,” 262 Science 1268 (1993) (“Taki”)
3. Yong-Rui Zou et al., “Cre-loxP-me-diated Gene Replacement: A Mouse Strain Producing Humanized Antibodies,” 4(12) Current Biology 1099 (1994) (“Zou”)
4. WO 91/00906 (“Wood”)
5. Certain opposition briefs filed by third parties in the European Patent Office contesting patentability of EP No. 1 360 287 (EP ’287) ■
The test for materiality is not whether references are directed to similar subject matter; the test is whether “the PTO would not have allowed a claim had it been aware of the undisclosed prior art.” Therasense, 649 F.3d at 1291, That standard is not met here.
Neither the district court nor my colleagues find that any uncited reference was closer to the claimed subject matter than the cited references, or filled gaps in the cited references, or related to additional limitations in the claims. Nor did the district court find invalidity based on the uncited references; invalidity was based on the court’s finding of indefiniteness, not on obviousness over cited or uncited prior art.4
The uncited references do not provide additional information of but-for materiality with respect to the claimed technology. My colleagues suggest that because these four references were later cited by Regen-eran in the prosecution, of related cases, this is an admission that the. references are material. Surely it was prudent for Regen-eran to submit these citations to the examiner for consideration in any still-pending applications, and Regeneran states that it also submitted the district court’s opinion. That action cannot be taken as an admission of but-for materiality.
The partiés debate several aspects of the broadest reasonable interpretation of claim terms, but neither the district court’s nor my colleagues’ analysis shows that any “withheld reference” is more material than the cited references. Under the district court’s “broadest reasonable interpretation,” the ’018 claims require a genetically modified mouse,' the genes' of which have been modified using the particular large targeting vector method described in the specification, by the insertion of human variable region DNA in its germline configuration into or next to the endogenous mouse immunoglobulin locus. Dist. Ct. Op, at 664-67. The “withheld references” indeed relate to genetic modification, but they are not but-for material as compared with the references before the examiner.
The district court does not establish that the allegedly withheld references lead to unpatentability. Instead, the district court states that the references disclose motivations, benefits, and cumulative teachings. That is correct; but the references do not provide but-for materiality, whether taken alone, or with the cited references.
The VelociGene project arose in a field of complex and unpredictable science, with no consensus on how to produce therapeutically effective antibodies. The predictability of the state of the science relates to the materiality determination, as the court has explained:
The methodology of science and the advance of technology are founded on the *1370investigator’s educated application of what is known, to intelligent exploration of what is not known. Each case must be decided in its particular context, including the characteristics of the science or technology, its state of advance, the nature of the known choices, the specificity or generality of the prior art, and the predictability of results in the area of interest.
Abbott Labs. v. Sandoz, Inc., 544 F.3d 1341, 1352 (Fed. Cir. 2008). Recognition of the value of providing a murine source of antibodies with therapeutic effect in humans does not render the achievement obvious when it is ultimately successful. See Cardiac Pacemakers, Inc. v. St. Jude Med., Inc., 381 F.3d 1371, 1377 (Fed. Cir. 2004) (“Recognition of a need does not render obvious the achievement that meets that need.”).
Nonetheless, my colleagues find that these four cumulative references are but-for material and were intentionally withheld in order to deceive the examiner. That is insupportable, as review demonstrates:
i. Brüggemann
Brüggemann is a 1996 review paper that collects the then-published methods of integrating immunoglobulin transgenes into murine genomes. Brüggemann concludes with a statement of hope for future achievement:
[A]n attractive alternative would be to replace the mouse Ig loci with the human Ig loci; in this way it might also be possible to retain and exploit any possible regulatory sequences in the mouse loci that are located distal to protein-coding regions. While such ambitions have not yet been realized, successful replacement of small portions of the mouse genome have been described.
Brüggemann at 394. Brüggemann also states:
[I]t is far from clear whether this [Ig loci replacement] will be the best way to create a mouse strain giving rise to a wide-range of high-affinity antibodies.
Id. at 397. The district court found that Brüggemann taught (1) replacing “much of’ the mouse Ig locus with human DNA; (2) an “explicit motivation” to exploit endogenous regulatory sequences; and (3) retaining an entirely human gene segment and an entirely murine gene segment. Dist. Ct. Op. at 572, 575. The district court ignored Brüggemann’s statements that these results had not been achieved, as well as that these elements are not required by the claims. See SRI Int’l v. Matsushita Elec. Corp., 775 F.2d 1107, 1121 (Fed. Cir. 1985) (en banc) (“It is the claims that measure the invention.”).
Brüggemann does not teach unrear-ranged variable region gene segments in the germline configuration, nor does it teach any method—much less the LTVEC method required by the claims. Indeed, the district court’s finding of materiality of Brüggemann is in conflict with the district court’s rejection of Regeneron’s arguments that the claims require retaining the mu-rine constant region and require functional murine regulatory elements. Brüggem-ann’s statement of “unrealized ambitions” of targeted replacement' of the immunoglo-bulin locus does not impart invalidating materiality when the ambitions are accomplished by Regeneron.
The Jakobovits reference teaches “replacing] all or a part of either the constant region or variable region of an antibody molecule.” Jakobovits, col. 1,11. 11-14. Ku-cherlapati, also cited by the examiner, teaches retaining promoter and regulatory regions of the host DNA. Kucherlapati, col. 10, 1. 64-col. 11, 1. 2. The district court found that Kucherlapati and Brüggemann were not cumulative, stating:
*1371Briiggemann teaches the benefits of targeted insertion as taking advantage of the regulatory regions distal to the protein-coding regions and the expectation that mouse regulatory sequences distal to the protein coding regions will remain intact. In contrast, Kucherlapati states that “the xenogeneic locus will be placed substantially in the same region as the analogous host locus, so that any regulation associated with the position of the locus will be substantially the same for the xenogeneic locus.”
Dist. Ct. Op. at 578 (internal citations omitted). The district court does not explain how this distinction converts Brüg-gemann into an invalidating reference.
The panel majority adopts different and flawed reasoning, finding that Briiggem-ann shows “targeted gene replacement” while Kucherlapati shows “wholesale replacement.” Maj. Op. at 1355-56. These “unrealized ambitions” are not teachings of this long-sought result, as the references readily demonstrate. Moreover, Kucherla-pati states that the host endogenous im-munoglobulin locus is “substituted by a portion of, or an entire, xenogeneic immu-noglobulin locus,” Kucherlapati, col. 3, 11. 55, and describes the inserted DNA as “human, constant and/or variable regions,” id. at col. 5, 11. 51-54, as does Briiggem-ann.
The panel majority also incorrectly states that Briiggemann suggests the method of Zou to accomplish retaining and exploiting regulatory elements. The method of Zou is cited only as an example of the “successful replacement of small portions of the mouse genome,” as opposed to a method to accomplish the “possibility” of inserting larger portions of the immuno-globulin loci. Briiggemann at 394. The panel majority’s statement that Zou is described as a method to retain and exploit regulatory sequences is a misreading of both Zou and Briiggemann.
ii Taki
Taki is a 1993 article describing the then-knowledge of targeted insertion of a rearranged murine variable region construct at the immunoglobulin locus. The rearranged gene inserted in the Taki reference, Vh15, is derived from a murine antibody to phosphorylcholine. Taki at 1268. In that early work, the Taki transgenic mouse produced fully murine antibodies to this particular antigen. The goal of this research was “exploration of im-munoregulatory mechanisms,” id., not the development of therapeutically useful human antibodies.
The district court found that Taki taught “the motivation to target human variable region DNA into the mouse Ig locus.” Dist. Ct. Op. at 574. Taki indeed mentions this long-sought ambition. The panel majority agrees, stating that the “fact that Taki teaches using exogenous mouse DNA instead of exogenous human DNA does not detract from the motivation Taki provides to target the mouse Ig locus with exogenous DNA.” Maj. Op. at 1354. However, a “motivation” to solve a known scientific problem is not a teaching of how to achieve that solution. “Knowledge of the goal does not render its achievement obvious.” Abbott Labs., 544 F.3d at 1352.
The claims of the ’018 patent require human DNA, not mouse DNA or any exogenous DNA. Neither the district court nor the panel majority addresses the enormous difference between Taki’s use of a single rearranged variable region gene and the unrearranged variable region gene segment in the ’018 patent. Taki does not teach a mouse with unrearranged variable region DNA capable of recombination to create innumerable immune responses. Taki does not teach the LTVEC method or human unrearranged variable region gene segments in their germline configuration. At most, Taki teaches targeted insertion of *1372a single gene of mouse DNA at the immu-noglobulin, locus.
The district court recognized that Taki “provides different motivations” than Ku-cherlapati, Dist. Ct, Op. at 578. Taki reflects the early work in this field; it has been superseded by the teachings of Ku-cherlapati and the other cited references. The record does not support the district court’s finding of materiality. The panel majority errs in holding otherwise..
in, Zou
Zou teaches the targeted insertion of a human constant region gene segment, and uses the Cre-loxP system to “replace the mouse gene, Cyl, which encodes the constant region of the heavy chain of IgGl antibodies, with its human counterpart.” Zou at 1099. The district court found Zou to be but-for material because it “provides significant motivation to target the mouse Ig locus with human Ig DNA.” Dist. Ct. Op. at 575. The district court’s error was in equating the motivation to solve a known problem with teaching the solution to the problem.
The district court found that Zou, along with Taki, taught a “method” for inserting human unrearranged variable region gene segments into an endogenous mouse im-munoglobulin locus. Dist, Ct. Op. at 575. Zou is cumulative of at least Kucherlapati, as well as Jakobovits who teaches the same Oe-to*P-mediated targeting of the immunoglobulin locus as utilized by both Zou and the ’018 patent.5 Jakobovits, col. 1, .11. 11-14. Kucherlapati teaches that the xenogeneic (human) locus is “substituted” in “substantially the^ same region as the analogous host locus.” Kucherlapati, col. 10, 11. 50-55. Zou does not add but-for material information to these references. Zou and Jakobovits use the same method of targeted insertion; Zou is not alleged to teach a missing limitation, but only to provide a “motivation” to target the immuno-globulin locus. Again, “[kjnowledge of the goal does not render its achievement obvious.” Abbott Labs., 544 F.3d at 1352. The district court’s contrary ruling is incorrect, as is the panel majority’s endorsement of that ruling.6
iv. Wood
Wood describes a transgenic mouse having’ únreárranged human DNA fragments incorporated into its germline. Wood teaches the use of either constructed unre-arranged gene fragments or the use of contiguous unrearranged human DNA. Wood, col. 16, 11, 14-22. Wood does not describe how such gene fragments are “introduced” or “integrated” into the germ-line of the described mouse; Wood does not teach targéted insertion.
The district court found that Wood teaches the “insertion of human variable region gene segments upstream of an endogenous mouse constant region, to produce a genetically modified mouse” and “motivates a person of ordinary skill to use an endogenous mouse constant p, (mu) region for purposes of allelic exclusion.” Dist. Ct. Op. at 572-73. Both the district court and the panel majority misread Wood.
*1373Wood teaches a “DNA fragment construct” with murine constant regions upstream from the human variable region gene segments. Building a DNA construct in a particular order to be later inserted is not the same as describing the targeted insertion of that construct into germline DNA. Wood does not describe any targeted insertion method described elsewhere in the prior art, such as Cre-loxP. The district court excuses this absence, because Wood “is appropriately understood as including but not limiting insertion at the Ig locus.” Dist. Ct. Dp. at 573.
Wood’s teaching of a “DNA construct” was misread as teaching the targeted insertion of that construct at a particular portion of the endogenous locus. The Wood teaching of “integration” into the genome is cumulative of Lonberg and other references which broadly teach “integration” into the genome. Lonberg, [0292]. There is no support in Wood for' the leap from a broad, unspecified disclosure of “integration” somewhere into the genome, to the district court’s finding of disclosure of targeted insertion at the Ig locus.
Neither my colleagues nor the district court explains how an examiner would have tied together- the conflicting approaches and unrealized ambitions of the four purportedly omitted references to render obvious the method described and claimed in the ’018 patent.
v. European Opposition Briefs
The European Opposition Briefs were filed in the European Patent Office, in an opposition proceeding associated with EP ’287, a counterpart of the Regeneran technology. The Merus opposition brief cited the references cited by the United States examiner, and additional references in this busy field of science, including the same Brüggeman, Taki, Zou, and Wood references. The district court stated that the “faithful” “description” of the allegedly withheld references in the European opposition would “have led inexorably to an understanding of their relevance and but-for materiality.” Dist. Ct. Op; at-577.
It is noteworthy that the European Technical Board of Appeals ruled that EP ’287 was patentable over these allegedly withheld references. See Decision in Appeal No. T2220/14-3.3.08, at 67-68 (Taki); 71-72 (Brüggemann); 72-77 (Wood); and 77-78 (Zou), available at http://www.epo. org/law-practice/case-law-appeals/pdf/t 142220eul.pdf. These determinations negate but-for materiality, as well as the district court’s analysis. Perhaps this is why the panel majority chose not to discuss the European Opposition. Maj, Op. at 1350 n.3.
There is no support—legally or factually—for the district court’s reliance on the European opposition briefs to find these four references material to patentability. The European tribunal, with these references before it, did not find the claims unpatentable. Nor did the district court. The panel majority upholds a finding Of but-for materiality without finding the claims invalid based on these purported but-for material references. It is not disputed that the information in those references did not solve the problem that was ultimately solved by the ’018 patent.
Conclusion
The controlling precedent of Aptix v. Quicktum, supra, and Keystone Driller, supra, cannot be ignored by this panel. Although my colleagues make much of the purported “litigation misconduct” relating to the privilege log and discovery in this infringement litigation, this has no relation to whether there was inequitable conduct in the prosecution before the patent examiner. Intent to -deceive the examiner cannot be inferred from purported litigation misconduct several years later.
*1374The premises of the law of inequitable conduct have not been established by clear and convincing evidence. Intent to withhold material references in order to deceive the examiner was not found by the district court, and cannot be inferred. These four additional references were not but-for material to patentability, and specific intent to deceive was not shown. From my colleagues’ contrary ruling, I respectfully dissent.
. Regeneron Pharmaceuticals v. Merus B.V., 144 F.Supp.3d 530 (S.D.N.Y. 2015) (‘‘Dist. Ct. Op”).
. The bracketed paragraph citation format is retained from the reference.
. I note that this application was recently allowed over both, Kucherlapati and Taki.
. The references, cited and uncited, all recognize the goal of providing antibodies for utility in human therapies—a goal not achieved. The district court recognized that the references state the motivation for development of the science, but it appears undisputed that the problem was not solved until the Regeneran scientist's succeeded, as reported in the '018 patent.
. Although the district court found that Jako-bovits taught targeting only for the insertion of lox sites, that is incorrect, for Jakobovits refers to the "use of Cre-mediated site-specific recombination for modifying immunoglobulin loci, for instance, to replace all or a part of either the constant region or variable region of an antibody molecule.” Jakobovits, col. 1, 11. 11-14. '
. The district court referred in a footnote to Regeneron's internal email discussion of citation to Zou in preparing a scientific publication, and found these conversations "relevant” to materiality. Dist. Ct. Op. at 557 n.21. This discussion has no bearing on the status of Zou as but-for material prior art.