Illumina v MGI Part 2: Has the UK lost its way on the doctrine of equivalents?

The High Court decision in Illumina v MGI ([2021] EWHC 57 (Pat) hit the IP headlines for its application of the Supreme Court decision in Regeneron v Kymab ([2020] UKSC 27) (see IPKat: Illumina v MGI Part 1: Mr Justice Birss on sufficiency, DNA sequencing and chocolate teapots). Those who make it to paragraph [307] of the over 500 paragraph judgment are also rewarded with Mr Justice Birss' thought-provoking application of the doctrine of equivalents under Actavis v Eli Lilly ([2017] UKSC 48).  Unusually for a decision applying Actavis, Mr Justice Birss in Illumina v MGI provided guidance on how to approach the third Actavis question. However, Mr Justice Birss' approach may lead to some odd consequences for how the scope of a patent is understood, particularly with regards to the pre-eminence of the claims over the description.  

Legal Background: The 3rd Actavis question

The Actavis questions are used to determine whether an alleged infringement that does not fall under the literal (or normal, purposive) construction of a patent claim, none-the-less falls under the scope of the claim according to the doctrine of equivalents (DoE) as established in Actavis v Eli Lilly ([2017] UKSC 48). The first two Actavis questions ask 1) whether a variant (the alleged infringement) achieves substantially the same result and substantially the same way as the claimed invention and 2) whether this would be obvious to a skilled person at the priority date. If the answer to both questions 1) and 2) is "yes", the 3rd Actavis question asks whether the skilled person would none-the-less have understood the patentee as intending strict compliance with the literal meaning of the claims, and thus as not covering the variant. The variant infringes under DoE if the answer to question 3 is "no". 

The 3rd Actavis question can cause cognitive dissonance for patent attorneys trained to draft patent claims which include only those features necessary for defining the core inventive concept. To a patent attorney, the intention for a feature to be limiting might thus be found in the very fact of the inclusion of that feature in the claims. In his guidance on the 3rd Actavis questions, Mr Justice Birss in Illumina v MGI adds a further complication. Mr Justice Birss found that, when answering the 3rd Actavis question, a strict definition in the description excluding a variant was evidence of the patentee's intention for strict compliance with the literal meaning of the claims. This creates a peculiar situation, in which definitional language in a claim may be non-limiting whilst definitional language in the description is limiting. 

Case Background: Cool Massively Parallel Sequencing (CoolMPS)

The patents at issue in Illumina v MGI related to improved modified nucleotides for use in method of DNA sequencing. Illumina submitted that MGI's Cool Massively Parallel Sequencing (CoolMPS) system infringed their EP(UK) patent EP 1530578 either under a normal purposive construction of the claims, or under the doctrine of equivalents (DoE) as established in Actavis v Eli Lilly ([2017] UKSC 48). 

MGI argued that their CoolMPS system differed from the sequencing method described in Illumina's patent. The Illumina patent claim specified the use of nucleotides that are linked to a detectable label. The label allows detection and identification of the nucleotide (i.e. as a A, C, T or G) once it has been incorporated into a sequence. In the CoolMPS system, the nucleotide itself is not linked to a label. Instead, once the nucleotide has been incorporated into a sequence, the nucleotide is detected by a labelled antibody. See here for a video explanation of CoolMPS. 

Infringement under a literal construction of the claims

MGI argued that two phrases in the claims of the patents excluded their CoolMPS system from the claim scope. First, MGI argued that the phrase "cleavable linker" in the claim required the nucleotide to be covalently linked to a label and not merely detected by a labelled antibody. Second, MGI argued that the phrase "incorporation of a nucleotide" required the nucleotide and label to be incorporated in a single step, and did not cover two-step incorporation of first the nucleotide and second the label via the binding of a labelled antibody. 

Mr Justice Birss was not convinced by MGI's arguments, and found infringement of each of the relevant claims under a literal construction of the claims. In case he was incorrect on normal infringement, Mr Justice Birss also considered infringement under DoE. In particular, Mr Justice Birss considered a) whether the bound antibody of the CoolMPS system should be considered equivalent to a "cleavable linker" construed as requiring a covalent bond and b) whether a two-step incorporation step using a nucleotide and labelled antibody was equivalent to the claimed one step incorporation step.

DNA sequencing

On both of these questions, there was no dispute that the first two Actavis questions should be answered in the affirmative, i.e. that i) MGI's labelled antibodies worked in substantially the same way to achieve the substantially same result as Illumina's claimed method; and ii) this would have been obvious to a skilled person at the priority date. 

"Cleavable linker": The 3rd Actavis question and sufficiency

The 3rd Actavis question asks iii) whether a skilled person would conclude that the patentee intended strict compliance with a literal meaning of the claims as an essential requirement of the invention. Mr Justice Birss was not convinced that the patent specification demonstrated an intention from Illumina to limit "cleavable linker" to covalent bonds. MGI argued that a skilled person would understand "cleavable linker" as not extending to antibodies because the skilled person would have had to generate the antibodies. Mr Justice Birss, however, accepted the expert evidence that generation of the necessary antibodies would be common general knowledge (CGK) for the skilled person. 

The consideration of the role of CGK prompts the question of the overlap (or lack of) between the DoE and sufficiency. As discussed in Part 1, Regeneron (and Biogen) sufficiency requires a claim to be enabled across the entire scope of a claim (for features relating to the essence of the invention). The "scope of the claim" for the purposes of sufficiency is normally understood to be the scope under a literal construction of the claim language. However, the question following Actavis is whether a claim must also be sufficiently disclosed with respect to equivalents. 

Mr Justice Birss briefly mentioned sufficiency in his DoE analysis (para. 339): "one might believe that strict compliance [with the wording of the claims] was essential to avoid a problem with sufficiency". However, Mr Justice Birss did not consider the point relevant to the case in question in the absence of convincing arguments from MGI that the proposed equivalents were not enabled. The question of whether infringement under the DoE requires sufficiency of disclosure with regards to the variant therefore remains open, as does the related question of whether Regeneron sufficiency requires sufficiency across the entire scope of the claim including equivalents. 

"Incorporation": The 3rd Actavis question and a less-is-more approach to patent drafting

In CoolMPS, an unlabelled nucleotide is first incorporated. A label is subsequently attached via binding of a labelled antibody to the incorporated nucleotide. MGI argued that the CoolMPS process could not be considered equivalent to the single step process by which the labelled nucleotides were "incorporated" in the patents. 

Equivalence to the claimed feature "incorporation of a nucleotide" was considered with respect to two Illumina patents. The first of these was the same patent for which the "cleavable linkers" had been considered, describing sequencing methods using labelled nucleotides (EP 1530578).  The second Illumina patent (EP 1828412) related to sequencing buffers and their use in DNA sequencing. As with the first patent, the claim language of the second patent specified "incorporation" of labelled nucleotides in the sequencing method. 

Crucially the specification of the first patent defined "incorporation" in only general terms, as "becoming part of". In view of this broad definition, Mr Justice Birss was unconvinced that a skilled person might understand from the specification an intention from the patentee to limit the claims to single-step incorporation of the nucleotides. Infringement by equivalence by the CooMPS system was therefore found with respect to the first patent. 

In contrast to the first patent, the second patent specified in the description that "incorporation" referred to the joining of a nucleotide into a nucleic acid strand via a covalent. Because of this definition, Mr Justice Birss considered a two-step method of incorporation explicitly excluded from the literal meaning of the claim. But was a two-step method equivalent? On this issue, Mr Justice Birss strikingly found that the skilled person would consider strict compliance with the normal meaning of the claim as essential.  This was for the simple reason of the clear definition of "incorporated" provided in the description. For Mr Justice Birss this definition was evidence that the patent was "drafted in such a way as to exclude" the variant represented by the CoolMPS system (para. 396). 

The take-home message from the aspect of the infringement decision in Illumina v MGI is clear: those drafting a new patent application would be advised to follow a "less is more" approach and avoid unnecessary definitions in the description that could be interpreted as excluding, even as yet unforeseen, immaterial variants from claim scope. The consequences of Illumina v MGI are thus likely to include the careful avoidance of unnecessary definition and boilerplate language in patent specifications. Patent applicants amending the description according to the new EPO guidelines are also likely to prefer deletion of embodiments as opposed to inclusion of explicit language that excludes an embodiment from the scope of the invention (IPKat). 

Have we lost our way on the doctrine of equivalents?

An advantage of the DoE for patentees is that infringement by equivalence does not require the patent to delineate immaterial from essential features of the invention in the claims (see for example Regen v Estar ([2019] EWHC 63 (Pat), IPKat)). Decisions applying Actavis have so-far set a high bar for evidence of an intention by the patentee for strict limitation of the claims. Prosecution history has not yet been shown to provide any such evidence. In this context, the decision in Illumina v MGI may be seen as providing a counter-balance to these earlier decisions, by at least providing an example as to what evidence of a patentee's intention may entail. 

However, Illumina v MGI, in combination with earlier decisions of the lower courts to apply Actavis, places the UK in a strangely paradoxical position on patent claim interpretation. The principle of Actavis is that defining an immaterial feature in the claims does not necessarily limit the scope of the claims. Mr Justice Birss observed in Illumina v MGI that a limiting definition of an immaterial feature in the description, however, does limit the scope of the claim i.e. by demonstrating a clear intention by the patentee that the feature should be limiting. Thus, it appears that the definition of an immaterial feature in the description may be limiting, whilst the definition (i.e. literal interpretation) of the same immaterial feature in the claims is not. 

In Illumina v MGI we thus find ourselves in very strange territory indeed, where a limitation in the description may hold more weight than a limitation in the claims. The underlying direction of the courts in applying Actavis is directed by the emphasis in the UK on the core essence of the invention, as opposed to the language of the claims, as a descriptor of patent scope. However, it is worth remembering that the advantage of using claim language to determine claim scope is that it enables parties to understand clearly whether or not they are infringing a patent. Thus it seems that the more Actavis is applied in the lower courts, and the more the central importance of claim language in defining the scope of protection is eroded, the more difficult it becomes to accurately evaluate freedom to operate in view of a granted patent.