Originally posted on IPKat.
The recent UK High Court decision in Samsung v Alexion [2025] EWHC 1240 (Pat) is a timely reminder that you can never be too careful with patent sequence data. Mr Justice Meade’s ruling in this case highlights that simple mistakes in the sequence information can be fatal for both the enforcement and validity of the claims. The case, relating to the blockbuster rare disease antibody Soliris, is a sorry saga of expensive errors by the Patentee that highlights both the importance of ensuring accurate sequence information as well as the difficulty of rectifying any errors after a patent application has been filed.
Defining a molecule by its sequence
Ensuring that the sequences of a claimed biological molecule are correct in a patent application is absolutely critical but not always straightforward. Sequence information often forms the very substance of a biologic invention. A mistake of even a single letter in an antibody sequence can thus lead to the patent defining entirely different subject matter than intended, potentially meaning the patent doesn’t even cover the commercial product. Interestingly, sequences are one of the only situations where the subject matter of the invention is necessarily not defined by the claim language per se. The term “SEQ ID NO: 1” means very little without a description and sequence listing defining and providing the actual sequence referred to as SEQ ID NO: 1. Claims specifying SEQ ID NO’s therefore can only be interpreted by reference to the description.
As EPO Board of Appeal cases have highlighted, errors in the sequence information can have fatal consequences for a patent (T 1213/05). Given that antibody sequences can be thousands of amino acids long, and a full sequence listing might contain vast numbers of such sequences, it’s remarkably easy to make an error but almost impossible to manually detect it. This places a huge burden on applicants to ensure 100% accuracy of their patent sequence data. Sequence alignment software is needed to ensure absolute accuracy, as well as triple checking with the inventors that the correct sequences have been added to the patent application.
Alexion’s Soliris patent: Where did it all go wrong?
The dispute in this case related to Alexion’s rare disease blockbuster eculizumab, marketed as Soliris, and its associated patent EP (UK) 3167888. Alexion brought an infringement action against Samsung Bioepis and Amgen who were looking to launch their own eculizumab biosimilars. The patent specification described the TRIUMPH clinical trial relating to the use of eculizumab to treat the rare disease Paroxysmal Nocturnal Hemoglobinuria (PNH). Importantly however, for reasons not gone into in the judgement, Alexion could not maintain its priority claim for the sequence of eculizumab. Consequently, a key publication relating to the trial was determined to be key prior art. This publication described eculizumab as a humanized monoclonal antibody that targets the terminal complement protein C5, but did not disclose the amino acid sequence of eculizumab.
In view of the prior art, Alexion shifted its patent strategy to claiming eculizumab per se instead of its medical use. The original patent and SPC for eculizumab defined a genus of C5 binding antibodies, and did not disclose the sequence of eculizumab specifically. The granted patent at issue in this case claimed “[a]n antibody that binds C5 comprising a heavy chain consisting of SEQ ID NO: 2 and a light chain consisting of SEQ ID NO: 4.” Alexion argued that the sequence of eculizumab had never properly formed part of the state of the art due to an earlier erroneous sequence submission to the Chemical Abstracts Service (CAS). The original sequence for eculizumab submitted to CAS by Alexion, for the purpose of its International Non-proprietary Name submission for eculizumab, included a “major” error. Whilst Alexion later corrected the CAS error, the correct sequence only became publicly available after the filing date of the patent.
Unfortunately for Alexion, however, its sequence errors did not end with the CAS submission. SEQ ID NO: 4, as defined in the patent, was also not actually the sequence of eculizumab. However, this was a different error to that in the CAS submission. SEQ ID NO: 4 in the patent included an additional 22 amino acids leader sequence not present in the final antibody product. Antibody leader sequences are short stretches of amino acids that function to direct the newly synthesized antibody amino acid chains to the correct intracellular compartments for proper assembly into the final antibody molecule. Importantly, the leader sequence is typically cleaved off after it has served its purpose and does not become part of the mature antibody protein. In this case, for example, the leader sequence for eculizumab did not form part of eculizumab. However, the EPO did not permit amendment of the claims to remove the leader sequence, finding that this would add matter to the application as filed (T 1515/20). Both Soliris and the Defendants’ biosimilar products do not contain the extra 22 amino acids specified in SEQ ID NO: 4, and so were not covered by the claims of the patent as literally construed.
Claim construction: what did SEQ ID NO: 4 mean?
The primary battleground between the parties was therefore construction of the SEQ ID NO: 4 sequence reference in the claim. Alexion contended that claim 1, despite specifying a light chain “consisting of SEQ ID NO: 4” which included the leader sequence, should be construed purposively to cover the mature eculizumab light chain without the 22 extra amino acids. Alexion argued that the skilled person would realise that the extra 22 amino acids were a leader sequence and would understand the claim to require a functional light chain without this leader sequence. The Defendants argued that the claims were limited to what they literally said, meaning the 22 amino acid leader sequence was a required feature. They pointed to the precise language “consisting of” and the definite nature of a sequence identity.
Mr Justice Meade sided with the Defendants. He found that “[t]he language actually used by the claims is of course a critical consideration. Here, the two most important features are that SEQ ID NO: 4 is a use of scientific language that would generally be expected to be precise, and the use of the drafting convention that “consisting of” specifies exactly what must be present. These matters are strongly in the Defendants’ favour.” (para. 208).
Crucially, the judge also noted that the description of the patent itself indicated that the prior art disclosed eculizumab as a whole antibody. This language was, of course, a result of the original application having been directed to the medial use of eculizumab, as opposed to eculizumab per se. The judge reasoned that “[i]n the present case, the patentee said that eculizumab was old […] the whole specification reads as if eculizumab was old and what was being contributed was new ways of using it” (para. 209). The skilled person reading the patent, the judge found, would not expect claims to cover something acknowledged as old. Alexion had also argued that the “incorporation by reference” terminology used in the specification to refer to the prior art patent would lead the skilled person to look at the prior art and realise that it did not in fact disclose eculizumab. However, the judge was not impressed with this argument, finding that “incorporation by reference” was just boilerplate that had “the unrealistic effect of incorporating very large amounts of information willy-nilly into the Patent.” (para. 210).
Alexion further argued that it was just more scientifically sensible that the claim was intended to be directed to eculizumab without the erroneous leader sequence. However, the judge was also not convinced by this argument, noting that “Patents do sometimes claim things which are suboptimal or whose purpose is not entirely evident. That is not a licence to rewrite the claims.” (para. 220).
The judge finally considered the potential relevance of the prosecution history in claim construction. Memorably, the Supreme Court in Actavis v Eli Lilly & Co [2017] UKSC 48 advised a sceptical, but not absolutist, attitude to prosecution history. According to the Supreme Court, reference to the file history would generally only be appropriate where i) the point at issue is truly unclear if one confines oneself to the specification and claims of the patent, and the contents of the file unambiguously resolve the point, or (ii) it would be contrary to the public interest for the contents of the file to be ignored, see for example Regen Lab v Estar [2019] EWHC 63 (Pat). In the present case, the judge did not find the claims “truly unclear” to an extent warranting extensive reference to the file history. Furthermore, the judge found that even if prosecution history were to be considered, it did not in any case unambiguously resolve the construction issue, nor did it show Alexion had disavowed the interpretation it now advanced for the granted claims.
The judge thus construed the claim as not covering the eculizumab sequence. Alexion’s infringement case therefore failed, given that the biosimilars also lacked the 22-amino acid leader sequence. Alexion also submitted its “pragmatic” concession that if its construction was wrong and the patent was not infringed (given that it would also thereby be worthless) (para. 26). The decision noted the potential insufficiency attack based on the position that an antibody with the leader sequence would be non-functional, although this was not the judge’s own view.
Even if the court had agreed with Alexion’s interpretation of its patent claims, meaning they covered the final eculizumab drug, Mr Justice Meade still found the patent would not have been inventive. The judge reasoned that the antibody sequence could have been easily arrived at by a skilled person from the prior art mentions of eculizumab as a C5 targeting antibody, prior disclosures of very similar antibodies and a skilled person’s common general knowledge of antibody engineering techniques.
Analysis
This decision serves as a potent reminder of the fundamental importance of ensuring the accuracy of sequence information in patents. In this case, an error in a sequence listing, coupled with admissions in the specification about the prior art status of the intended product, created a nightmare of claim construction for the patentee. On first impressions, it is hard to have much sympathy for the Patentee. Alexion not only got the sequence of their antibody wrong in their patent, they also got the sequence wrong in their regulatory submissions, on top of which they lost their priority claim (Alexion has a notable history with priority claims, being the patentee in the recent G 1/22 decision of the Enlarged Board of Appeal (EBA) on priority).
The decision in this case also highlights that, whilst purposive construction is a core component of UK patent law, it may not rescue a patentee from clear linguistic limitations in the claims. This is especially the case when those limitations find context and support within the description. The clear language provided in the description in this case, describing the eculizumab as part of the prior art, was influential in swinging the judge’s interpretation of the claims away from a construction that defined an arguably non-functional antibody as the invention.
Ultimately, Samsung v Alexion is a harsh but clear lesson that the devil is indeed in the detail of patent sequence data. Alexion’s unfortunate saga of sequence errors highlights that even sophisticated legal arguments centred on the clear purpose of an invention may not rescue a sequence inaccuracy in the application as filed. For biologic inventions, where the sequence is the invention, there is simply no substitute for getting it right, right from the start. If nothing else, this decision is also worth reading for the judge’s clear exasperation with the facts of the case, including his use of such phrases as “willy-nilly”. This is never a phrase you want associated with your patent.
Further reading
Is it time for patent offices to enter the bioinformatic age?
This post is based on a previous article on IPKat.
