Amgen v Sanofi/Regeneron in Japan: Function, sequence, mechanism
Amgen patents declared valid in Germany and Japan have also been deemed invalid by the European Patent Office, US Supreme Court, and in a separate Japan decision: but why? Takanori Abe of Abe & Partners examines the details.
Amgen owns two patents, JP5705288 and JP5906333, entitled ‘Antigen-binding protein against proprotein convertase subtilisin kexin type 9 (PCSK9)’ .
The amended claim 1 of JP5705288 (Amgen’s patent) is as follows:
An isolated monoclonal antibody, wherein the monoclonal antibody can neutralise binding of PCSK 9 to LDLR protein, and competes for binding to PCSK9 with an antibody (the judgment refers to it as ‘reference antibody’) including a heavy chain, including a heavy chain variable region consisting of an amino acid sequence in SEQ ID NO: 49 and a light chain, including a light chain variable region consisting of an amino acid sequence in SEQ ID NO: 23.
As the Federal Circuit named Amgen's US patent claims as double-function claims, the Japanese patent claims are similarly double-function claims. In other words, there are two types of antibodies: a neutralising antibody and an antibody competing with a reference antibody. Amgen’s Japanese claims are almost the same as the ones in Europe, and slightly different from the ones in the US, however, both Japanese and US claims are identified by the two functions.
In Japan and Germany, there are judgments affirming the validity of the patents and infringement. However, the EPO, another judgement in Japan and the US Supreme Court found invalidity grounds by different reasons. How does the judgment in Japan differ from the European and US ones? Why did the judgments in Japan change from valid to invalid?
Case 1-1: Judgment of December 27, 2018, IP High Court (Suit against JPO trial decision) (Patent Valid)
Sanofi filed a request for a trial for invalidation of Amgen’s patent. The JPO dismissed the request. Sanofi appealed to the IP High Court, seeking to revoke the decision. The IP High Court (Presiding judge Otaka) dismissed Sanofi's appeal, stating as follows.
(1) Support requirement
Sanofi alleges that the claim 1 is an invention without specifying the structure of the antibody, and thus it encompasses an enormous number or kind of binding neutralising antibodies with literally any possible structure, whereas there are as few as three groups or three kinds of antibodies described specifically in the
specification, and further it cannot be said that an antibody ‘competing’ with a reference antibody may neutralise the binding of PCSK9 and LDLR, and the ‘competition’ of an antibody with the reference antibody cannot be an indicator of ‘binding neutralisation’.
However, it is a matter of common general knowledge in the production process of a monoclonal antibody by an animal immunisation method, that an antibody specifically reacting with a specific antigen in animal body is produced, and hybridomas produced by use of the immunised animal are subjected to screening, and an amino acid sequence is identified in a process of identifying an antibody with specific binding properties. Therefore, it is not recognised as essential to specify the structure of the antibody (amino acid sequence) in advance in order to obtain an antibody with specific binding properties. Further, it is recognised that a skilled person could understand from the specification that a neutralising antibody competing with a reference antibody included in claim 1 of the patent could be obtained under the guidance of the specification without reference to the amino acid sequence of antibodies.
(2) Enablement requirement
Sanofi alleged that claim 1, taken literally, may include antibodies with various structures having totally different amino acid sequences from specific antibodies (three groups to three kinds of antibodies) that may be included in claim 1 in the detailed description of the invention in the specification, and of course may also include any antibodies that were totally unknown before. It also noted that excess trial and error would be expected by a skilled person in order to obtain every antibody encompassed into the claim 1.
However, it cannot be said as essential to specify a structure of the antibody (amino acid sequence) in advance to obtain an antibody with specific binding properties, and it is recognised that a skilled person could obtain a neutralising antibody competing with a reference antibody included in claim 1 under the guidance of the specification without reference to amino acid sequence of antibodies.
Case1-2: Supreme Court
Sanofi filed a petition for writ of certiorari to the Supreme Court. On April 24 2020, the Supreme Court denied certiorari.
Case 2-1: Judgment of October 30 2019, IP High Court (Infringement Suit) (Infringement affirmed/ Patent valid/ Injunction granted)
The IP High Court (Presiding Judge Takabe) affirmed the validity of Amgen’s patent and infringement, and granted the injunction.
(1) Whether it falls within the technical scope of the invention
Sanofi alleged that each of the inventions is a functional claim that specifies the invention only by a function to compete with reference antibody 1, and that a functional claim should be narrowly construed to have a scope within which a skilled person can perform from the specification. Therefore the technical scope of each of the inventions would be limited to specific antibodies described in the examples of each of the specifications or amino acid sequences in which one or several amino acids at a specific position are substituted in the antibody.
However, Sanofi’s allegation of each of the inventions being limited to the examples is not acceptable. Moreover, each of the inventions is not specified by an amino acid sequence. Thus there is no reason to construe them as being limited to the specific antibodies described in each of the specifications or antibodies having an amino acid sequence in which one or several amino acids are substituted at a specific position in the former antibodies.
(2) Support requirement, Enablement requirement
Regarding the support requirement and the enablement requirement, the judgment in Case 2-1 is almost the same as the judgment in Case1-1.
(3) Injunction
Sanofi submitted an expert opinion prepared by B (D33), alleging that the injunction of the manufacture and sale, etc. of Praluent causes patients currently administered or to be administered Praluent severe health risks or anxiety about future therapy.
D33 pointed out the problem with decreasing options for patients and expected that patients who use Praluent would l be confused by the injunction concerning sale, etc. of Praluent. However, it did not go so far as to point out a concrete health risk for patients from using products (Repatha) manufactured and sold by Amgen in place of Praluent. Thus, it cannot be said that specific facts have been demonstrated to prove that the injunction against the use of Praluent is contrary to the public interest.
Case 2-2: Supreme Court
Sanofi filed a petition for writ of certiorari to the Supreme Court, alleging as follows: ‘In the specification, Amgen disclosed the binding neutralising antibodies of PCSK9 and LDLR, which are only a few specific antibodies competing with a reference antibody. Even though there are numerous antibodies competing with a reference antibody, Amgen has only discovered the reference antibody and a limited number of antibodies competing with the reference antibody, and is trying to obtain exclusive rights to all antibodies competing with the reference antibody’. Several amicus curiae briefs by so-called large pharmaceutical companies were submitted, however, on April 24, 2020, the Supreme Court denied certiorari.
Case 3-1: Judgment of January 26 2023, IP High Court (Suit against JPO trial decision) (Patent Invalid)
On February 12, 2020, just before the above Supreme Court decision denying certiorari, Regeneron filed a request for a trial for invalidation of Amgen’s patent. The JPO dismissed the request. Regeneron appealed to the IP High Court, seeking to revoke the decision. The IP High Court (Presiding judge Kanno) dismissed Regeneron's appeal, stating as follows.
(1) Support requirement
The reference antibody itself can be acknowledged as a neutralising antibody which sterically interferes with binding between the PCSK9 and LDLR proteins and which strongly blocks the binding at a position which partially overlaps with a position of the EGFa domain of LDLR in the crystal structure. On this basis, the matter specifying an invention "which competes with the reference antibody for binding to PCSK9 " in the present invention also has a technical significance in that it is revealed that an antibody which competes with the reference antibody interferes with, blocks, reduces, or modulates the interaction between the PCSK9 and LDLR proteins by directly blocking the binding site of the LDLR protein (specifically, by the antibody binding to PCSK9 at a position which overlaps with a position of the EGFa domain of LDLR in the crystal structure) by a mechanism similar to that of the reference antibody.
It can be deemed to be highly probable that a group of antibodies with amino acid sequences having high identity with the reference antibody bind to PCSK9 at a position similar to the reference antibody. However, regarding several groups of antibodies with amino acid sequences other than the above, knowledge that a position where such an antibody binds on PCSK9 is revealed by the fact that the antibody was evaluated as competing in an assay such as epitope binning cannot be acknowledged to be common general knowledge. Thus, the position where the above antibody binds on PCSK9 cannot be deemed to be apparent.
Further, an antibody having a property "which competes with a reference antibody for binding to PCSK9 " of the present invention can include an antibody which prevents or inhibits (e.g., reduces) specific binding of the reference antibody to PCSK9 by binding to a site which differs from that where the reference antibody binds to PCSK9 and which differs from the position of the EGFa domain of LDLR in the crystal structure and bringing minor steric hindrance to the reference antibody.
Therefore, it cannot be deemed that an "antibody which competes with a reference antibody for binding to PCSK9" interferes with, blocks, reduces, or modulates the interaction between the PCSK9 and LDLR proteins by directly blocking the binding site of the LDLR protein (specifically, by the antibody binding to PCSK9 at a position which overlaps with a position of the EGFa domain of LDLR in the crystal structure) in a manner similar to the reference antibody. Thus, it also cannot be acknowledged that the "antibody which competes with a reference antibody for binding to PCSK9" has a functional property as a binding-neutralising antibody.
These points are supported by the results of the demonstration experiment by Dr. [A] and Affidavit (1) by Dr. [B]. In this demonstration experiment, 63 antibodies of Regeneron were tested for competition with the reference antibody and their binding-neutralising activity. It was confirmed that 13 antibodies competed with the reference antibody, among which 10 antibodies (about 80%) had no binding-neutralising activity.
The JPO decision determines that the present specification specifically demonstrates that a number of antibodies of the present invention are repeatedly identified with sufficiently high probability by performing the preparation and selection of immunised mice according to the procedure and schedule of the immunisation program as stated in the present specification, the production of hybridomas using the selected immunised mice, and the screening and epitope binning assay for identifying an antibody which strongly blocks the binding interaction between PCSK9 and LDLR as stated in the present specification from the beginning, repeatedly.
However, the second Expert Opinion by Professor [F] states that "It is impossible to generate and screen all possible candidate antibodies, because whether a particular mouse generates a particular antibody is controlled by luck," even if the production process of antibodies stated in the present specification has been undergone, it is "controlled by luck" what position on PCSK9 an antibody obtained in an immunised mouse will bind to.
Additional remarks will be made just to be sure.
Although judgments have been made in Europe and the US regarding this invention, needless to say, it is apparent that the judgments in other countries do not immediately affect the judgment in the present case. With regard to the US, even if the judgment of invalidation by the Court of Appeals for the Federal Circuit is overturned, the corresponding US patent is not directly relevant to the judgment on the present invention in either case, because it can be found that the "competition" with the reference antibody is not defined as a matter specifying the invention in the US patent.
In the previous case (Case 1-1), Sanofi's assertion of violation of the support requirement has been rejected. However, in the present suit, by new assertions based on new evidence such as the Affidavits by Dr. [A] and Dr. [B], the structural analysis by Professor [F]’s Expert Opinion, etc., there is a reasonable explanation why the conclusion of the present case differs from the previous case.
Case 3-2: Supreme Court
Amgen filed a petition for writ of certiorari to the Supreme Court. On September 14 2023, the Supreme Court denied certiorari. As a result, the IP High Court judgment of January 26, 2023, revoking the JPO's decision became final, and the JPO has reopened the case for examination.
Practical tips
The EPO found invalidity grounds of lack of inventive step and the US Supreme Court held the patent invalid for violation of the enablement requirement, while Japanese Case 3-1 held the patent invalid for violation of the support requirement.
Japanese Case 1-1 and Case 2-1 held the patent valid, emphasising that it is not essential to preliminarily specify a structure of the antibody (amino acid sequence) to obtain an antibody having specific binding properties. In contrast, the Japanese Case 3-1 specified the technical significance of the invention and referred to the mechanism, and held the patent invalid on the grounds of violation of the support requirements, considering also new evidence such as demonstration experiments and technical expert affidavits, which were not in line with the mechanism. It is unique in focusing on the mechanism rather than on the function or the amino acid sequence of the antibody. The focus on mechanism can be said to correspond to the ‘middle ground’ in Professor Lemley's article titled ‘The Antibody Patent Paradox’ (Mark A Lemley & Jacob S Sherkow, The Antibody Patent Paradox, 132(4) THE YALE LAW JOURNAL 994, 1000, 1053, 1064 (2023)).
Is the functional claim in antibody patents in Japan dead? Some analyses that describe the mechanism in the specification may avoid a violation of the support requirement (Jessica R. Sudbury & Carmela De Luca, Antibody Patents in the US & Japan: If You Claim More, You Better Enable More, Sep 6, 2023, last visited Mar 1, 2024). However, the difficult question of how the mechanism is identified remains.
Takanori Abe is the founder of Abe & Partners. He can be contacted at: abe@abe-law.com
