UPC Court of Appeal clarifies approach for interpreting claims with “obvious” errors

Alexion Pharmaceuticals, Inc. v Samsung Bioepis NL B.V. [UPC_CoA_402/2024] –Court of Appeal of the UPC (Grabinski, Blok, Gougé, Enderlin, Hedberg) – 20 December 2024

Alexion Pharmaceuticals, Inc. v Amgen Technology (Ireland) Unlimited Company & Ors. [UPC_CoA_405/2024] –Court of Appeal of the UPC (Grabinski, Blok, Gougé, Enderlin, Hedberg) – 20 December 2024

The UPC Court of Appeal has upheld the Hamburg Local Division’s decisions to deny Alexion preliminary injunctions against Samsung Bioepis and Amgen for sales of biosimilars to Alexion’s blockbuster drug, Soliris®. Although the CoA agreed with the Hamburg LD that there was an insufficient degree of certainty that the patent was valid to justify preliminary injunctions, the CoA disagreed with the LD’s interpretation of the patent, calling into question the LD’s finding that the patent was likely to be infringed.  The case is interesting because the CoA interpreted errors in patent claims differently, instead adopting an approach more closely aligned with the Technical Board of Appeal’s during examination of the same patent before the EPO (T 1515/20).  Moreover, whereas the LD ignored assertions made by the patentee during examination because the patentee had “abandoned” them during the UPC proceedings, the CoA saw no justification for ignoring them. Instead, the CoA regarded these assertions as indicative of the views of a skilled person at the filing date to the detriment of the patentee’s desired interpretation of the patent.

Background

These cases are concerned with Alexion’s divisional patent, EP 3167888B1, to an antibody that binds a component of the immune system, C5.  Patients with blood diseases, such as paroxysmal nocturnal haemoglobinuria (PNH), lack the protective proteins required to prevent red blood cells from being attacked by C5.  Accordingly, by binding to C5, the antibody can prevent red blood cells from being destroyed by the patients’ own immune systems.

Since 2007, Alexion has marketed a drug, Soliris®, containing the antibody, eculizumab, as the active ingredient.  Soliris® is authorised for the treatment of a variety of rare diseases, including PNH.  In July 2023, Samsung and Amgen launched biosimilar products, Epysqli® and BEKEMV^®, respectively, in several Contracting Member States of the UPC.

Having initially opted-out the patent application during the sunrise period, the opt-out was withdrawn in January 2024.  The EPO communicated its intention to grant the patent on 18 March 2024, and on 19 March 2024, Alexion made an application for a preliminary injunction against Samsung Bioepis and 9 Amgen entities for infringement under claim 2 of the patent.  Alexion subsequently requested unitary effect for the patent on the date of grant (1 May 2024).  Currently, the UPC’s CMS does not show that any infringement proceedings on the merits have been brought against any party.

Samsung filed an EPO opposition against Alexion’s patent the day after grant, with one of the Amgen entities filing an opposition on 28 October 2024.  On 31 October 2024, a Swiss Amgen entity (not party to the application for the PI) filed an action for revocation before the Milan Central Division.   The 9-month opposition period does not expire until 3 February 2025, after which the opposition will be accelerated in view of the pending UPC proceedings.

Claim 2 of the patent was directed to a pharmaceutical composition comprising the C5-binding antibody.  The antibody was defined in terms of a light chain amino acid sequence, SEQ ID No:4, identified in the patent as a specific sequence of 236 amino acids.  The sequence in the patent, however, was incorrect as it contained 22 extra amino acids at the start.  This error was corrected in the Chemical Abstract Service (CAS) database registry in 2009, after the filing date of the patent.  However, during examination, the patentee was unable to correct the sequence in the patent itself because the TBA did not consider the error to be obvious[1], contrary to the requirements of Rule 139 EPC. Furthermore, the TBA held that there was no basis for amending the sequence, contrary to the requirements of Article 123(2) EPC[2].

First Instance Proceedings

Alexion applied for a preliminary injunction against Samsung and Amgen before the Hamburg LD.  In UPC_CFI_124/2024, the Hamburg LD dismissed Alexion’s application, ordering Alexion to pay the costs of the proceedings and setting the value of the dispute at the highest level of €100 million. Although the Hamburg LD was convinced with a sufficient degree of certainty that the patent was infringed, the court was not convinced of the validity of the patent with a sufficient degree of certainty.  In this connection, the Hamburg LD held that a decision on provisional injunctions could not be based solely on the court’s view of the validity of the patent in suit, but also on the likelihood that the opposition division of the EPO will revoke the patent.  In the Hamburg LD’s view, it was reasonably likely that the opposition division of the EPO would revoke the patent for lack of sufficiency because the patentee had admitted that, in the presence of the first 22 amino acids, the antibody was non-functional in the sense that it would not bind to C5, contrary to the requirements of the claim.

The CoA interprets claim differently from the LD

The key question on interpretation was whether a skilled person would interpret claim 2 as excluding the first 22 amino acids of the light chain amino acid sequence, SEQ ID No:4.  The Hamburg LD took the view that the skilled person would interpret the claim to provide a technical teaching, which, if carried out, would lead to the intended success of the invention.  In doing so, they would recognise that an antibody that included a “signal peptide” having the first 22 amino acids in the light chain would not allow the antibody to bind to C5.  Accordingly, in the Hamburg LD’s view, the skilled person would recognise typical features of signal peptide sequences that would not lead to the intended success of the invention and regard these as unnecessary.

The CoA held that the considerations of the Hamburg LD as “legally flawed”. Instead, the CoA held that a linguistic error, a spelling mistake or any other inaccuracy in a patent claim could only be corrected by way of interpretation of the patent claim if the existence of an error and the precise way to correct it were sufficiently certain to the average skilled person on the basis of the patent claim, taking into account the description and the drawings and using common general knowledge.  This had to be a “rather strict standard”, since the error implied legal uncertainty for third parties.

Existence of error not “sufficiently certain”

In the case at hand, neither the error nor its correction were deemed sufficiently certain to a skilled person.  Concurring with the TBA, the CoA held that there were no reasons presented as to why a skilled person would be prima facie alerted and consequently prompted to analyse the sequence to determine the functional parts of the amino acid sequence.  Although the patentee relied on expert statements, these statements presupposed that a skilled person would analyse the sequence; for example, the experts were asked whether the skilled person would recognise the first 22 amino acids as a “signal sequence”.  Thus, the statements failed to demonstrate that a skilled person would have analysed the sequence in the first place.  Moreover, even if a skilled person were to inspect the sequence as argued by Alexion, the CoA concurred with the TBA that a skilled person would not immediately have recognised that the depicted sequence contained an error.  Even if a skilled person could have recognised features typical of a signal peptide, a skilled person would not know whether these were an error or simply an unusual part of the antibody.

Alexion argued that the average skilled person would consider the inclusion of a signal peptide in the sequence to be an error because a skilled person would have known that signal peptides would be cleaved during the production of antibodies by ordinary cell-based methods.  However, Samsung countered that alternative production methods existed that would not result in the cleavage of signal peptides. As the patent expressly stated that production of the antibody was not limited to any specific method, the CoA was not convinced that a skilled person would regard the cleavage of signal peptides as necessary.

The CoA also took the view that assertions made by Alexion during examination cast doubt on the sufficient certainty of the existence of the error.  Specifically, in arguing for sufficiency during examination, Alexion asserted that “the position of the three respective CDR sequences in SEQ ID NO: 4, which are instrumental for the specific binding properties required by the claim, is sufficiently distanced from the N-terminal signal peptide to dissuade the skilled person from having doubts that this longer light chain would also bind to C5”. This argument was accepted and formed the basis of the TBA’s positive finding on sufficiency.  The CoA saw this as confirmation that it was not sufficiently certain that a skilled person would see the longer light chain as an error. Although Alexion abandoned this line of argument before the UPC and argued instead that the amino acid sequence would not bind in the presence of the first 22 amino acids, the CoA held that a patent claim must be interpreted from the perspective of the person skilled in the art. Alexion’s assertions during grant and the TBA’s endorsement of them were seen as an indication of the view of the person skilled in the art at the filing date.

Conclusions

These Judgments confirm that, where there is an error in a patent claim, the claim can only be corrected by way of interpretation if the existence of the error and the precise way to correct the error are sufficiently certain.  This is necessarily a “rather strict standard”, as errors inevitably give rise to uncertainty for third parties. By concurring with TBA’s reasoning in T 1515/20 on why the error in the sequence could not be corrected under Rule 139 EPC, these Judgments suggest that a claim containing an error is likely to be interpreted with the error in place unless the existence of the error and its correction are obvious with “sufficient certainty”. This clearly has implications on the care that should be taken when drafting patent applications.  Practitioners should also be mindful of how assertions made during examination can have a detrimental effect on later proceedings; in this case, assertions made by the patentee in favour of validity during examination at the EPO adversely affected the patentee’s chances of their claim being corrected by interpretation during proceedings at the UPC.

[1] Reasons, 11 to 12 of T 1515/20

[2] Reasons, 14 of T 1515/20

This article was prepared by Partner and Patent Attorney Hsu Min Chung and Rachel Fetches.