Otsuka Pharmaceutical Co., Ltd. v. Lupin Ltd.

• Court: Court of Appeals for the Federal Circuit
• Decided: May 21, 2026
• Docket: 24-2297
• Status: Unpublished

Opinion

Case: 24-2297 Document: 46 Page: 1 Filed: 05/21/2026

NOTE: This disposition is nonprecedential.

United States Court of Appeals for the Federal Circuit ______________________

OTSUKA PHARMACEUTICAL CO., LTD., Plaintiff-Appellant

v.

LUPIN LTD., LUPIN PHARMACEUTICALS, INC., Defendants-Appellees ______________________

2024-2297 ______________________

Appeal from the United States District Court for the District of Delaware in No. 1:21-cv-00900-RGA, Judge Richard G. Andrews. ______________________

Decided: May 21, 2026 ______________________

ZACHARY L. GARRETT, Venable LLP, New York, NY, ar- gued for plaintiff-appellant. Also represented by JOHN D. MURNANE, JOSHUA ROTHMAN, ALICIA ALEXANDRA ROSE RUSSO; MEGAN S. WOODWORTH, Venable LLP, Washington, DC.

WILLIAM R. ZIMMERMAN, Knobbe, Martens, Olson & Bear, LLP, Washington, DC, argued for defendants-appel- lees. Also represented by JARED C. BUNKER, Irvine, CA; CAROL PITZEL CRUZ, Bellevue, WA. Case: 24-2297 Document: 46 Page: 2 Filed: 05/21/2026

______________________

Before HUGHES and CUNNINGHAM, Circuit Judges, and BURROUGHS, District Judge. † HUGHES, Circuit Judge. Otsuka Pharmaceutical Co., Ltd. appeals a final judg- ment of the United States District Court for the District of Delaware. The district court held that certain claims of U.S. Patent Nos. 8,273,735 and 8,501,730 were not in- fringed by Lupin Ltd. and Lupin Pharmaceuticals, Inc.’s manufacturing process incorporated in Abbreviated New Drug Application No. 216063. The court also held that cer- tain claims of U.S. Patent No. 8,273,735 were invalid for obviousness. For the following reasons, we affirm. I Otsuka Pharmaceutical Co., Ltd. (Otsuka) is the owner of U.S. Patent Nos. 8,501,730 and 8,273,735 (collectively, patents-in-suit). Respectively, the patents-in-suit claim highly pure tolvaptan—a compound used to treat Autoso- mal Dominant Polycystic Kidney Disease (ADPKD)—and improved methods for synthesizing tolvaptan. While previ- ous synthesis methods for tolvaptan led to the production of an impurity known as the dechlorinated impurity, the innovation of the patents-in-suit is that, by reducing the amount of a key hydrogenating reagent in the synthesis process—sodium borohydride—the amount of the dechlo- rinated impurity is reduced. Otsuka uses this innovation in manufacturing its ADPKD treatment JYNARQUE®. In May 2021, Lupin Ltd. notified Otsuka that it had submitted an Abbreviated New Drug Application (ANDA)

† Honorable Allison D. Burroughs, District Judge, United States District Court for the District of Massachu- setts, sitting by designation. Case: 24-2297 Document: 46 Page: 3 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 3

to the Food and Drug Administration, seeking approval to market generic versions of JYNARQUE®. Otsuka then brought an action for infringement of the ’730 patent against Lupin Ltd. and its wholly owned subsidiary Lupin Pharmaceuticals, Inc. (collectively, Lupin) under 35 U.S.C. § 271(a), (e)(2)(A), and (g). Later, Otsuka amended its com- plaint to also assert infringement of the ’735 patent. Prior to and at trial, the issues were narrowed to include in- fringement and invalidity for obviousness. Otsuka asserted claims 1, 2, 4, and 5 of the ’730 patent and claims 7, 8, and 10 of the ’735 patent against Lupin. The asserted claims of the ’730 patent are all independent product-by-process claims, and they require reduction of a benzazepine compound in the presence of a hydrogenating agent. The claims further require that this hydrogenating agent be present in an amount of either 0.25–1 or 0.25–0.5 molar equivalent per 1 mole of benzazepine pre- cursor compound. For example, claim 1 of the ’730 patent recites: 1. A highly pure 7-chloro-5-hydroxy-1-[2-me- thyl-4-(2-methylbenzoylamino)benzoyl]-2,3,4,5 -tetrahydro-1H-1-benzazepine having a purity of more than 99.5%, or a salt thereof, which is produced by the process which comprises re- ducing a benzazepine compound of the formula (1):

wherein X1 is a halogen atom, R1 and R2 are in- dependently a lower alkyl group, or a salt Case: 24-2297 Document: 46 Page: 4 Filed: 05/21/2026

thereof in the presence of a hydrogenating agent selected from the group consisting of lith- ium aluminum hydride, sodium borohydride, zinc borohydride, and diborane in an amount of 0.25 to 1 mole per 1 mole of the compound (1). ’730 Patent, 29:9–32. Similarly, the asserted claims of the ’735 patent are method claims, all of which also require the reduction of a benzazepine compound in the presence of a hydrogenating agent in an amount of either 0.25–1 or 0.25–0.5 molar equivalent per 1 mole of benzazepine compound. For exam- ple, claim 7, which depends from unasserted claim 6, re- cites (reproduced with claim 6 below for reference): 6. A process for producing a 2,3,4,5-tetrahydro- 1H-1-benzazepine compound of the formula (10):

wherein X1 is a halogen atom, R1 and R2 are in- dependently a lower alkyl group, or a salt thereof, which comprises reducing a benzaze- pine compound of the formula (1): Case: 24-2297 Document: 46 Page: 5 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 5

wherein R1, R2 and X1 are as defined above, or a salt thereof in the presence of a hydrogenat- ing agent selected from the group consisting of lithium aluminum hydride, sodium borohy- dride, zinc borohydride, and diborane in an amount of 0.25 to 1 mole per 1 mole of the com- pound (1). 7. The process according to claim 6, wherein the hydrogenating agent is sodium borohydride which is used in an amount of 0.25 to 1 mole per 1 mole of the compound (1). ’735 Patent, 31:61–32:36. Lupin’s ANDA submission incorporates Drug Master File (DMF) No. 036263, which describes the process by which Lupin intends to synthesize its generic tolvaptan ANDA product. Lupin’s DMF indicates that its tolvaptan synthesis process also makes use of a reduction reaction like the one claimed by Otsuka, but where Otsuka’s claimed process generally uses 1 molar equivalent or less of hydrogenating agent such as sodium borohydride per 1 mole of precursor compound, Lupin’s process uses at least 1.2 molar equivalents of sodium borohydride per 1 mole of precursor. See, e.g., J.A. 2571–73. In Lupin’s process, after this sodium borohydride has been added, two samples are taken, one after 15 minutes and one after 75 minutes. At each point, the sample is tested to see if no more than 0.05% of the original amount of precursor compound re- mains in the reaction mixture. If these tests indicate that 0.05% or less of the original amount of precursor compound remains, then Lupin quenches the reaction by adding wa- ter and hydrochloric acid. This destroys the chemical bonds in the sodium borohydride, preventing any further reac- tion. If these tests indicate that more than 0.05% of the amount of precursor compound remains, more sodium bo- rohydride is added. Case: 24-2297 Document: 46 Page: 6 Filed: 05/21/2026

After a bench trial, the district court held that Lupin’s DMF method for producing tolvaptan did not infringe the asserted claims of the patents-in-suit. Separately, the dis- trict court considered evidence of the invalidity of the pa- tents-in-suit for obviousness over several pieces of prior art, ultimately concluding that “Lupin has shown by clear and convincing evidence that a POSA would have found the claimed invention [of the ’735 patent] obvious.” Otsuka Pharm. Co. v. Lupin Ltd., No. 21-cv-00900, 2024 WL 3618123, at *11 (D. Del. July 31, 2024) (Bench Trial Opin- ion). The district court also found, however, that Lupin did not demonstrate the invalidity of the asserted ’730 patent claims under its obviousness theory. Otsuka appeals. We have jurisdiction under 28 U.S.C. § 1295(a)(1). II On appeal, Otsuka challenges the district court’s in- fringement conclusions, its conclusion that Lupin’s expert at trial qualified as a skilled artisan, and its invalidity con- clusions regarding the ’735 patent. We address each argu- ment in turn. A We begin with Otsuka’s challenge to the district court’s infringement conclusions. “Following a bench trial, we re- view the district court’s conclusions of law de novo and its fact-findings for clear error.” Merck Sharp & Dohme Corp. v. Amneal Pharms. LLC, 881 F.3d 1376, 1384 (Fed. Cir. 2018). Infringement is a question of fact reviewed for clear error after a bench trial. United Therapeutics Corp. v. Liquidia Techs., Inc., 74 F.4th 1360, 1367 (Fed. Cir. 2023). “A factual finding is clearly erroneous when, despite some supporting evidence, we are left with a definite and firm conviction that the district court was in error.” Packet Intel. LLC v. NetScout Sys., Inc., 965 F.3d 1299, 1305 (Fed. Cir. Case: 24-2297 Document: 46 Page: 7 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 7

2020) (quoting Alcon Rsch. Ltd. v. Barr Lab’ys, Inc., 745 F.3d 1180, 1186 (Fed. Cir. 2014)). Otsuka’s primary infringement argument on appeal is that the district court adopted a “practical completion” con- struction for the claim term “amount” but applied a differ- ent and contradictory “absolute completion” construction in determining that Lupin’s DMF process does not infringe. Under the proper understanding of “amount,” Otsuka claims that Lupin infringes because the reaction is practi- cally complete before 1 molar equivalent of hydrogenating agent is added to the reaction mixture. During claim construction, the parties disputed how to construe the term “amount” in the asserted claims. While Lupin offered a construction of “amount” that would make the relevant amount of hydrogenating agent the total amount added into the reaction chamber, see J.A. 136, 25:1–4, the district court disagreed, finding that the claim language itself required that the claimed amount of hydro- genating agent be “the amount while the reaction is taking place,” see J.A. 138–39, 27:22–28:9. Later, in post-trial briefing, the parties presented conflicting visions of when the reduction reaction actually takes place. While Otsuka argued that the claim term referred to “the amount of hy- drogenating agent added before the reduction reaction reaches ‘practical completion’ or is ‘complete in a practical sense,’” Bench Trial Opinion, 2024 WL 3618123, at *6 (quoting J.A. 1987), Lupin argued that “so long as unre- acted ketone precursor and sodium borohydride are pre- sent, the reaction will continue to proceed,” id. (quoting J.A. 2025). The district court agreed with Otsuka, finding that Lupin’s proposed construction was directed to the “theoretical possibility” of “absolute completion” of the re- action. Id. The district court therefore noted that it under- stood “the reduction reaction to be ‘taking place’ until it reaches some point of practical completion.” Id. Case: 24-2297 Document: 46 Page: 8 Filed: 05/21/2026

Having decided that the “amount” of hydrogenating agent was the amount present until the claimed reduction reached “practical completion,” the district court deter- mined that Otsuka failed to show infringement for two rea- sons. First, the district court found that, even if Otsuka was correct that practical completion is reached when “so little precursor ketone remains that no further reduction is observed,” Otsuka did not adduce evidence that showed that the reaction was practically complete by the time the amount of precursor compound fell to 0.05% of its original level, the point at which Lupin decides to end the reaction by initiating the quenching step. See id. at *7 (quoting J.A. 1985). Second, the district court determined that, even if Lupin’s reaction did reach practical completion at the 0.05% precursor level, Otsuka failed to demonstrate that “no more than 1 molar equivalent of sodium borohydride or less has been added by that point.” Id. Assuming without deciding that practical completion is reached when Otsuka claims it is—that is, when the amount of precursor compound has fallen to 0.05% of its original level—we agree with the district court that Otsuka has failed to prove infringement of the patents-in-suit by Lupin’s DMF process. Otsuka offers two main pieces of ev- idence to suggest infringement of the patents-in- suit: (1) two experiments performed by Lupin while devel- oping its DMF process, Experiments 109 and 115, which Otsuka contends show practical completion before 1 molar equivalent of sodium borohydride is added; and (2) testi- mony from its expert Dr. William Roush that the details of Lupin’s DMF process indicate that the reduction reaction will be complete long before the addition of 1 molar equiv- alent of sodium borohydride. Ultimately, neither persuades us that the district court clearly erred. First, as the district court found, the underlying data for Experiments 109 and 115 “contained anomalies that might be accounted for by some unspecified ‘margin of er- ror’ or a lack of ‘appropriate quality control.’” Id. (citations Case: 24-2297 Document: 46 Page: 9 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 9

omitted). For example, as per the testimony from Otsuka’s expert Dr. Roush, after the full amount of hydrogenating reagent was added in Experiment 115, the precursor mole- cule appeared to be regenerating, despite this being impos- sible based on the reaction’s chemistry. See J.A. 324, 88:7–16; J.A. 354–56, 118:3–120:25. While Otsuka at- tempts to convince us that these experiments nonetheless could form a reliable basis for Dr. Roush’s infringement opinions, we do not find clear error in the district court’s decision to not rely on experimental data that is, by Dr. Roush’s own admission, flawed. Second, Dr. Roush’s separate testimony that Lupin’s DMF process is practically complete before the addition of 1 molar equivalent of hydrogenating agent, based on his “vast experience performing sodium borohydride reac- tions,” similarly fails to persuade us that the district court clearly erred. J.A. 312, 76:15–16. Otsuka claims that Dr. Roush’s expertise, combined with certain details about Lupin’s DMF process, demonstrates that the DMF reaction will proceed quickly and run to completion before even 0.5 molar equivalents of hydrogenating agent are added. But Dr. Roush’s DMF conclusions also meaningfully rely on data from Lupin’s Experiments 109 and 115. See, e.g., J.A. 1993 (characterizing Dr. Roush’s testimony as “based on Experiments 109/115, and taking into account the dif- ferences in process parameters between Experi- ments 109/115 and the DMF process”). And we find no clear error in the district court’s conclusion that these ex- periments, on their own or in combination with Dr. Roush’s expertise, shed little light on when the reduction reaction in Lupin’s DMF process is practically complete. For one, record evidence from both experts supports the district court’s conclusion that Experiments 109 and 115 were per- formed under different conditions than the DMF process. See, e.g., J.A. 327, 91:23–25 (Dr. Roush); J.A. 432–33, 196:13–197:5 (Dr. Dichtel). Further, notwithstanding Dr. Roush’s testimony that these differences should be Case: 24-2297 Document: 46 Page: 10 Filed: 05/21/2026

immaterial, see J.A. 327–34, 91:23–98:13, Dr. Dichtel tes- tified that these differences would be material and impact any potential infringement conclusions resulting from the extrapolation of these experiments to the DMF process, see J.A. 432–33, 196:13–197:24. Considering the totality of the evidence before the dis- trict court, we do not find clear error in its conclusion that “Otsuka fail[ed] to prove by a preponderance of the evi- dence that the reduction reaction in Lupin’s tolvaptan syn- thesis process reaches completion before more than 1 molar equivalent of sodium borohydride has been added.” Bench Trial Opinion, 2024 WL 3618123, at *7. B We next turn to Otsuka’s argument that the district court’s infringement and invalidity analyses were flawed because Lupin’s expert Dr. Dichtel does not qualify as a person having ordinary skill in the art. Otsuka notes that the district court’s own definition of the skilled artisan in this case is someone who “ha[s] a relevant doctorate degree and ‘at least two years of experience in the synthesis, re- search, and development of medicinal compounds.’” Appel- lant Br. 49 (citing Bench Trial Opinion, 2024 WL 3618123, at *3). Otsuka then argues that because Dr. Dichtel lacks the requisite experience, his testimony should not have been admitted. Otsuka presented these same arguments to the district court after trial. Ultimately, however, the district court did not entertain the merits of Otsuka’s arguments, finding that because Otsuka did not timely object to Dr. Dichtel’s testimony before or during trial, it had not preserved any challenge to the admissibility of Dr. Dichtel’s testimony. Bench Trial Opinion, 2024 WL 3618123, at *5 (citing Fed. R. Evid. 103(a)). We review procedural issues not unique to patent law, such as forfeiture, under the law of the regional circuit. See Case: 24-2297 Document: 46 Page: 11 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 11

Innogenetics, N.V. v. Abbott Lab’ys, 512 F.3d 1363, 1371 (Fed. Cir. 2008). The Third Circuit “review[s] for abuse of discretion a district court’s determination that a party for- feited an argument by failing to raise it earlier in the pro- ceedings.” Harbor Bus. Compliance Corp. v. Firstbase.io, Inc., 152 F.4th 516, 527 (3d Cir. 2025). If an argument is found to be forfeited, it is reviewable only in “truly ‘excep- tional circumstances.’” Id. at 529 (citation omitted). Under the law of the Third Circuit, a party who fails to object to errors at trial forfeits the right to complain about those alleged errors later. United States v. Rivas, 493 F.3d 131, 136 (3d Cir. 2007). And while Otsuka is correct that it argued Dr. Dichtel was not a skilled artisan in an elliptical fashion in its opening statement, see, e.g., J.A. 244–45, 8:3–9:21 (“[W]e think Dr. Dichtel is straying from his ex- pertise.”), it did not object at trial when Dr. Dichtel stated that he felt comfortable opining on the patents-in-suit from the perspective of a skilled artisan, see J.A. 445–46, 209:21–210:12. Nor did Otsuka file a Daubert motion seek- ing to exclude Dr. Dichtel as an expert before trial. On these facts, we cannot find that the district court abused its discretion in finding that Otsuka forfeited its challenges to Dr. Dichtel’s status as a skilled artisan. We therefore up- hold the district court’s finding without reaching the merits of this issue. C Finally, we consider Otsuka’s arguments that the dis- trict court erred in invalidating the asserted claims of the ’735 patent for obviousness. “Obviousness is a question of law based on underlying facts, and ‘[o]n appeal from a bench trial, this court reviews the district court’s conclu- sions of law de novo and findings of fact for clear error.’” Eli Lilly & Co. v. Teva Parenteral Meds., Inc., 845 F.3d 1357, 1372 (Fed. Cir. 2017) (quoting Prometheus Lab’ys, Inc. v. Roxane Lab’ys, Inc., 805 F.3d 1092, 1097 (Fed. Cir. 2015)). The underlying factual determinations within the Case: 24-2297 Document: 46 Page: 12 Filed: 05/21/2026

obviousness analysis include: “(1) the scope and content of the prior art; (2) differences between the prior art and the claims at issue; (3) the level of ordinary skill in the perti- nent art; and (4) secondary considerations such as commer- cial success, long felt but unsolved needs, and failure of others.” Incept LLC v. Palette Life Scis., Inc., 77 F.4th 1366, 1371 (Fed. Cir. 2023). In its bench trial opinion, the district court first con- cluded that a skilled artisan would have been motivated to select Kondo, 1 a reference which teaches a process for small-scale synthesis of tolvaptan, as a starting point in seeking to achieve the claimed invention. Next, the district court found that within the Kondo process, a skilled artisan would have been motivated to specifically select the reduc- tion step for modification based on testimony from Dr. Dichtel. The district court then concluded that a skilled artisan would have been motivated to reduce the amount of sodium borohydride used in Kondo’s reduction step be- fore going on to find that a skilled artisan would also be motivated to reach the claimed invention either in light of his or her background knowledge or in light of two other pieces of prior art. After establishing this motivation to combine, the district court also found a reasonable expec- tation of success in developing the invention claimed by the asserted claims of the ’735 patent. Finally, the district court considered various secondary considerations of non- obviousness and concluded that these considerations did not overcome the strong evidence in favor of obviousness. On appeal, Otsuka primarily attacks the court’s moti- vation analysis, arguing that Dr. Dichtel’s focus on

1 Kazumi Kondo et al., 7-Chloro-5-hydroxy-1-[2-me- thyl-4-(2-methylbenzoylamino)benzoyl]-2,3,4,5-tetrahydro- 1H-1-benzazepine (OPC-41061): A Potent, Orally Active Nonpeptide Arginine Vasopressin V2 Receptor Antagonist, 7 Bioorganic & Med. Chemistry 1743, 1743–54 (1999). Case: 24-2297 Document: 46 Page: 13 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 13

Kondo—and, specifically, Kondo’s reduction step—is indic- ative of hindsight bias that should have rendered his testi- mony not credible. Otsuka also argues that the district court erred by dismissing its secondary consideration evi- dence, specifically evidence of unexpectedly high tolvaptan purity and yield over Kondo. We are unpersuaded that the district court clearly erred in its analysis of the skilled artisan’s motivation to start with and modify Kondo. While Otsuka argues that Dr. Dichtel focused solely on Kondo, reflecting hindsight bias, Dr. Dichtel testified that he “did [his] own literature search” and looked at several other tolvaptan-synthesis references in addition to Kondo, but that he considered Kondo to be a particularly promising reference. See J.A. 544–45, 308:21–309:13. So too does Otsuka’s argu- ment downplay evidence, credited by the district court, that a separate, non-prior-art reference, Zard, 2 referred to Kondo as a standard method for synthesizing tolvaptan at the time of the claimed invention. See J.A. 677–78, 441:18–442:21. While Otsuka is correct that Zard acknowl- edged “major obstacle[s]” with Kondo’s process, see J.A. 618–19, 382:4–383:10, considering all of the evidence, we do not find that these considerations undermine the dis- trict court’s conclusion that “Lupin . . . presented clear and convincing evidence that a POSA would have recognized Kondo as a viable starting point,” Bench Trial Opinion, 2024 WL 3618123, at *12. See In re Mouttet, 686 F.3d 1322, 1334 (Fed. Cir. 2012) (recognizing that even “inferior com- bination[s]” of prior art may be appropriately considered in obviousness inquiry). Similarly, we find no clear error in the district court’s analysis of the skilled artisan’s motivation to modify the reduction step of Kondo specifically. As the district court

2 U.S. Patent Application Publication No. 2007/0185323. Case: 24-2297 Document: 46 Page: 14 Filed: 05/21/2026

noted, both experts agreed that many of the steps of Kondo would have been considered for modification, including the reduction step. See J.A. 546–48, 310:13–312:16 (Dr. Dichtel); J.A. 665, 429:7–16 (Dr. Roush). And Dr. Dichtel presented significant testimony about why a skilled artisan would have sought to modify Kondo’s reduc- tion step, including cost savings, improvement of the reac- tion’s safety profile, post-processing simplification, and knowledge that the hydrogenating agent was already being used in molar excess. See, e.g., J.A. 457–63, 221:1–227:12. This evidence, tethered to the chemical realities of the re- action, is a far cry from the conclusory expert testimony our case law rejects. Cf. TQ Delta LLC v. CISCO Sys., Inc., 942 F.3d 1352, 1361–62 (Fed. Cir. 2019). We therefore de- cline to find clear error in the district court’s choice to credit this testimony. Lastly, we fail to find reversible error in the district court’s conclusions regarding secondary considerations of non-obviousness. Otsuka purports to demonstrate both high yield and high purity by relying in part on a declara- tion from one of its chemists, Hirotaka Yukawa. See J.A. 4106–16. However, in analyzing purity, the Yukawa declaration compares the claimed methods not to Kondo it- self, but to a process that uses double Kondo’s amount of hydrogenating agent. Compare J.A. 449–50, 213:21–214:9 (reflecting that Kondo uses 1.5 molar equivalents of hydro- genating agent), with J.A. 4111 (indicating that Yukawa’s process used “3 fold molar amount of [sodium borohydride] per 1 mole of the [precursor molecule]”). The district court did not therefore clearly err by failing to credit the Yukawa declaration as accurately reporting the results of Kondo’s purity. See Bench Trial Opinion, 2024 WL 3618123, at *21. Nor do we find clear error in the district court’s conclusion that this difference in concentration of hydrogenating agent was material because “Otsuka’s internal documents report the presence of a dechlorinated impurity when using 2 molar equivalents of sodium borohydride . . . but not Case: 24-2297 Document: 46 Page: 15 Filed: 05/21/2026

OTSUKA PHARMACEUTICAL CO., LTD. v. LUPIN LTD. 15

when using 1.5 molar equivalents as disclosed in Kondo.” Id. at *10, *21. While Otsuka argues that this interpreta- tion of the evidence of record rests on a misinterpretation of Otsuka’s internal documents, we do not find clear error in the district court’s contrary conclusion, based on testi- mony from Dr. Dichtel that the court explicitly found “more credible.” Id. at *21. Regarding unexpectedly high yield, it appears that the district court overlooked evidence from Dr. Dichtel’s testi- mony, Kondo itself, and the Yukawa declaration reporting that the yield of Kondo’s process using 1.5 molar equiva- lents of hydrogenating agent was 30%, as opposed to the claimed process’s 82–93% yield. Nonetheless, we conclude that any error in evaluating this secondary consideration is harmless. After all, “weak secondary considerations gen- erally do not overcome a strong prima facie case of obvious- ness.” Genentech, Inc. v. Sandoz Inc., 55 F.4th 1368, 1378 (Fed. Cir. 2022) (citation modified). And we have already declined to disturb the district court’s findings underpin- ning the prima facie case. Accordingly, any evidence intro- duced of unexpected yield “is not sufficient to overcome the strong case of obviousness as a matter of law.” Adapt Pharma Operations Ltd. v. Teva Pharms. USA, Inc., 25 F.4th 1354, 1375 (Fed. Cir. 2022). We therefore affirm the district court’s obviousness determinations regarding the asserted claims of the ’735 patent. III We have considered the parties’ remaining arguments and find them unpersuasive. For the foregoing reasons, we affirm the district court’s judgment of non-infringement of the patents-in-suit, its determination that Otsuka forfeited its skilled-artisan arguments, and its judgment that the as- serted claims of the ’735 patent are invalid for obviousness. AFFIRMED