The Patent Trial and Appeal Board (PTAB) has benefited, particularly after enactment of the Leahy-Smith America Invents Act, from the deference to its factual findings mandated by the Supreme Court's interpretation in Dickenson v. Zurko of the application (deferential for factual determinations) of the standard-of-review provisions in the Administrative Procedures Act.  This deference does not extend to the Board's decisions based on legal standards nor can the PTAB escape from Federal Circuit review of its reasoning (or failure to provide its reasoning), requirements illustrated in the Court's decision in In re Theripion.

The case arose as an appeal from a decision from the PTAB in ex parte prosecution of U.S. Application No. 15/909,314, a U.S. national phase application from International Publication No. WO 2017/044424.  The Board affirmed the Examiner's determination that claims 1-13, 16, 22-24, and 27 of the '314 application were obvious.  Claim 1 is representative of the claims reviewed by the Court:

1.  A fusion polypeptide comprising, from an amino-terminal position to a carboxyl-terminal position, ApoA1-L1-D, wherein:
    ApoA1 is a first polypeptide segment comprising an amino acid sequence having at least 95% identity with amino acid residues 19-267 or 25-267 of SEQ ID NO:2, wherein said first polypeptide segment has cholesterol efflux activity;
    L1 is a first polypeptide linker consisting of from 10 to 40 amino acid residues; and
    D is an immunoglobulin Fc region,
    wherein the fusion polypeptide has increased cholesterol efflux activity as compared to the ApoA1-L1-D fusion polypeptide in which L1 is a two amino acid linker or is absent.

As explained in the opinion, there is a correlation between low blood levels of high-density lipoprotein (HDL) and the risk of stroke and heart attack, which are related to atherosclerosis.  Therapeutic intervention accordingly have been directed towards increasing HDL levels.  Apolipoprotein A-1 ("ApoA1") is "the principal protein component of HDL."  Thus, one stratagem for increasing HDL levels is to introduce ApoA1 into an individual having low HDL levels.  A drawback to this approach is that ApoA1 only remains intact in a human for a short time (it has a short "half-life").  The invention claimed in the '314 application is directed to modifying the protein to increase its half-life by linking it through a polypeptide linker to a portion of an antibody molecule (comprising the immunoglobulin Fc region).  (As explained in the opinion, antibody molecules have a relatively long half-life in vivo in contrast to ApoA1.)  The opinion sets out the understanding in the art that fusion proteins comprising immunoglobulin Fc regions were known to have improved half-lives, the art disclosing an ApoA1-Fc fusion (but in contrast to the claimed invention not having a polypeptide linker between these two components of the fusion protein).  This distinction is relevant, because the prior art fusion protein was deficient in one of ApoA1-HDL's important functions, increasing the process whereby the body removes free cholesterol from the bloodstream (termed reverse cholesterol transport, RCT).  Theripion's claims were directed to ApoA1-Fc fusion proteins joined by a polypeptide linker comprising 10-40 amino acids that had such beneficial RCT properties.

The Examiner rejected the '314 application's claims for obviousness based on the combination of 4 references:  Knudson et al., 2011, U.S. Patent Publication No. 2011/0178029; Ledbetter et al., 2015, U.S. Patent No. 8,937,157; Bacus et al., 2009, U.S. Patent Publication No. 2009/0318346; and Lagerstedt et al., 2015, U.S. Patent Publication No. 2015/0353626.  (The Examiner applied other references to certain of the claims but these were not considered by the panel on appeal.)  The Board affirmed these rejections and this appeal followed.

The Federal Circuit vacated and remanded, in an opinion by Judge Stark, joined by Judges Hughes and Cunningham.  Importantly for the panel's decision, the Board is required to "make the necessary findings and have an adequate evidentiary basis for its findings" to enable the Court to review them, under In re Nuvasive, Inc., 842 F.3d 1376, 1382 (Fed. Cir. 2016).  In addition, the Board must "articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made" concerning obviousness, Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983), although the amount of such explanation will vary case-by-case "depending on the complexity of the matter and the issues raised in the record," Pers. Web Techs., LLC v. Apple, Inc., 848 F.3d 987, 992 (Fed. Cir. 2017).

The basis for Theripion's appeal (with which the panel agreed) is that the Board erred in its analysis of whether the claimed invention showed unexpected results (one of the objective indicia of non-obviousness) and whether the skilled worker would have had a motivation to combine the cited art with a reasonable expectation of success in achieving the claimed invention.  But the Court's agreement was limited by "the extent that we find the Board failed to adequately explain how it determined the '314 application's claims are obvious in light of the totality of the record before it" according to the opinion.  One example of the dearth in the Board's explanation involved whether including the linker would produce the improved RCT activity of the claimed fusion proteins, Theripion's contention that the Board found unpersuasive.  The opinion states that the panel was unable to discern the Board's reasoning of why this result would have been expected on the record before it, which was limited to the declarative statement that "[t]hus, one of ordinary skill in the art would similarly expect that use of a peptide linker, such as Gly4Ser4, would increase the biological activity (i.e., cholesterol efflux activity) of a Fc fusion protein with ApoA1."  In contrast, Theripion had submitted a declaration from one of the named inventors (and the lead author of one of the cited references) that "common practice in the art of Fc fusion protein engineering [was] to not include a peptide linker, or to include only a short, one or two amino acid linker, between the N-terminus of an Fc region and the C-terminus of a fusion partner."  In addition, Theripion provided prior art examples of linker-comprising fusion proteins that did not necessarily increase (and in at least one instance, decreased) protein function.  Finally, Example 1 of the '314 application showed a side-by-side comparison between ApoA1-Fc fusion proteins comprising a 26 amino acid sequence linker and one having none or a 2-amino acid linker, wherein increased RCT activity was seen only in the embodiment having the long linker.  While the Court did not hold that this evidence was insufficient or that the Board erred in finding it unpersuasive, the opinion states that "we cannot discern from [the Board's] opinion the reasons for such a conclusion [of nonpersuasiveness].  Nor can we determine whether the Board adequately considered the totality of Theripion's evidence."

Illumination of the Board's reasoning was also not present in its explanation of why the Examiner was persuaded by the combination of Bacus and Ledbetter, particularly regarding the disclosure in Bacus of fusion proteins comprising linkers that involved "an entirely different type of protein."  Similarly, the panel found no reasoning on the record of why Ledbetter's disclosure that longer linkers were associated with increased DNase activity (another entirely different protein) using "yet another type of linker" was persuasive.  According to the opinion, "[t]he Board did not provide any rationale for why a skilled artisan would have expected Ledbetter's data relating to DNase fusion protein activity to be predictive of ApoA1 activity in the fusion proteins claimed in the '314 application.  After all, ApoA1 is an entirely different kind of protein than Bacus' ErbB or Ledbetter's DNase."  Finally, the Board apparently entirely ignored (at least on the record) prior art evidence that including a linker could reduce protein activity, particularly because the protein in these studies (DNase) was the same as in the Ledbetter reference relied upon by the Examiner to support the obviousness rejection.  And while recognizing that neither the Board nor the Examiner were obliged to "explicitly discuss every issue or every piece of evidence" and that failure to do so "does not alone establish that [the Board] did not consider [any such evidence]," Novartis AG v. Torrent Pharms. Ltd., 853 F.3d 1316, 1328 (Fed. Cir. 2017), the Board "must examine the relevant data and articulate a satisfactory explanation for its action[,] including a rational connection between the facts found and the choice made" under Motor Vehicle Mfrs. Ass'n. Reaching the heart of the matter, the opinion concludes this portion of its assessment of the Board's performance in this case by stating that "we are concerned that the Board may have improperly used the claims of the '314 application 'as a template for its own reconstruction,'" citing Sensonics, Inc. v. Aerosonic Corp., 81 F.3d 1566, 1570 (Fed. Cir. 1996).

Theripion also contended that the Board erred its assessment of unexpected results by considering the Bacus and Ledbetter references (which as noted above were directed to different fusion proteins) rather than the reference by Knudson regarding ApoA1-Fc fusion proteins; this reference Theripion argued was the closest prior art which on that basis should have been considered, citing Kao Corp. v. Unilever U.S., Inc., 441 F.3d 963, 970 (Fed. Cir. 2006), and Adapt Pharma Ops. Ltd. v. Teva Pharms. USA, Inc., 25 F.4th 1354, 1372-73 (Fed. Cir. 2022).  Once again the panel states that while it appears the Board believed the Bacus and Ledbetter references were the closest prior art "the Board never expressly says so" and on these grounds the panel instructs the Board on remand to "determine which prior art is the closest prior art and explain why that is, and then consider and explain whether Theripion's results are unexpected relative to that closest prior art."

Further, regarding the question of unexpected results, the Board failed to address (and thus the Court "had no analysis to review") whether there was a problem known in the prior art regarding low RCT activity in ApoA1-Fc fusion proteins having polypeptide linkers shorter than the 10-40 amino acids recited in Theripion's claims.  The Court recognized that "the '314 application solv[ing] a problem that was not recognized in the prior art, if true, could support a finding of nonobviousness, as it would support Theripion's contention that its results were unexpected," citing In re Gruskin, 234 F.2d 493, 498 (C.C.P.A. 1956).

Turning to the question of the existence of a motivation to combine by the person of ordinary skill in the art, the panel noted that the Board asserted the combination of the Knudson, Bacus, Ledbetter, and Lagerstedt references, and Theripion argued that the "art fails to teach or suggest any functional relationship between linker length and protein function in the context of an ApoA1-Fc fusion."  Further, Teripion provided declaration evidence that the N-terminus of the Fc portion of the fusion protein "has generally been viewed in the art as a natural linker region that does not require further extension when constructing an Fc fusion protein."  This was evidence, Theripion contended before the Examiner, the Board, and the Federal Circuit that showed the skilled artisan would have had no motivation to combine the cited references.  As it did throughout the opinion, the panel found that the Board provided no basis for its conclusion to the contrary; indeed the opinion states that the Board "made findings regarding what each prior art reference taught in isolation and failed to articulate any reason why a skilled artisan would have modified Knudsen's system with Ledbetter's linker - other than the unexplained assertion that 'incorporation of such a linker increases biological activity of the fusion partner,' for which there is little, if any, support in the record."  This explanation is insufficient, the opinion asserts, because the Board "must articulate a reason why" a person of skill would be motivated to combine references," citing Nuvasive (emphasis in opinion) which in this case the Board did not.

The Federal Circuit concludes the opinion with an explanation why it vacated and remanded rather than reversed, as Theripion had advocated they do.  The opinion states that "[g]iven that we are presently unsure of the bases on which the Board reached its subsidiary factual findings supporting its ultimate legal conclusion of obviousness, we are not in a position to determine whether there is substantial evidence to support the Board's determination," citing Alacritech, Inc. v. Intel Corp., 966 F.3d 1367, 1370-73 (Fed. Cir. 2020).

In re Theripion (Fed. Cir. 2023)
Panel: Circuit Judges Hughes, Cunningham, and Stark
Opinion by Circuoit Judge Stark