Takeda revokes Roche patent to glycosylation pattern extending to Entyvio


Takeda’s product Entyvio contains the monoclonal antibody vedolizumab and is used as a treatment for Crohn’s disease and ulcerative colitis. In the treatment of these autoimmune diseases, in order to prevent depletion of the patient’s white blood cells, it is preferable to block targeted T cells from entering the gut, rather than to cause death of those cells. Vedolizumab operates by binding to X4β7 integrin and prevents T cells expressing this integrin from migrating into the gut.

Roche’s patent EP (UK) 2 007 809 (“the Patent”) claimed a monoclonal human IgG1 or IgG3 antibody glycosylated at position Asn 297 characterised by the amount of fucose, N-glycolylneuraminic acid (NGNA) and N-terminal alpha 1,3 galactose (a-Gal) residues within the sugar chain. Antibodies with a high degree of fucosylation exhibit reduced antibody dependent cellular cytotoxicity (ADCC), an effector function which is triggered when an antibody binds to its target cell and ultimately results in cell death via lysis.

Roche alleged that Entyvio infringed the Patent and Takeda alleged that the Patent was invalid, on a large number of grounds. The case was heard by Birss J in the Patents Court, who gave a lengthy judgment dealing with a number of technical issues on antibody technology[1].

What did the claims mean?

There were two key areas of dispute between the parties on the meaning of the claims of the Patent: (i) how to establish the level of fucosylation required by the claim; and (ii) the meaning of the term “antibody” as used in the claim.

Fucosylation levels

The parties put forward several possible methods of calculating whether an antibody would meet the required 99% fucosylation level specified in the claims. The judge preferred a calculation (referred in the judgment as TRM) that involved taking into account all fucosylated glycans (for the numerator in the calculation) and all higher mannose species (for the denominator). Notwithstanding the expert evidence, which suggested that all the calculations proposed by the parties involved “crazy math”, the judge opted for what appeared to be the “least crazy” option. Adopting this calculation, vedolizumab satisfied the level of fucosylation required by the claims.

Antibody, as used in the Patent

In the specification, the Patent explained that an “antibody according to the invention contains at least a functionally active (FcR binding) Fc part of IgG1 or IgG3 type comprising glycosylated Asn 297.” Although the skilled person would be familiar with the term ‘antibody’ used in the claims, the judge found that in the Patent the word had a special meaning and included only antibodies having a functionally active (FcR binding) Fc region, with glycosylation at Asn 297.

The Fc region of vedolizumab contains a mutation referred to as the “LAGA mutation”. Takeda’s case was that the LAGA mutation had been engineered so as to disrupt Fc receptor binding (substantially reducing ADCC) and, accordingly, the Patent was not infringed. Although a small degree of binding may remain, and indeed this was shown to be the case by surface plasmon resonance (SPR) assay experiments conducted by Roche on vedolizumab, there was no evidence that a functional effect takes place as a result of the binding. Roche’s position was that its experiments showed some, albeit small, binding and vedolizumab therefore has an Fc region which is functionally active, falling within the claim. According to Roche, as vedolizumab exhibits no ADCC, the high level of fucosylation in the antibody contributed to the absence of ADCC given that some Fc region binding was present.

The judge found that the LAGA mutation was not regarded as something that would necessarily eliminate functions such as ADCC altogether. Roche’s experiments showed some binding to the receptor by vedolizumab. Although it was “a close call”, the judge held that on the evidence it was more likely than not that the level of fucosylation made some contribution to the absence of ADCC. Accordingly, vedolizumab would infringe the Patent, if valid.


An important preliminary finding made by the judge was that, at the priority date in 2006, the skilled team had within its common general knowledge the know-how to make an antibody to a given target antigen. While this required a lot of work it was not an undue burden. This was determinative in his assessment of whether certain prior art documents not only disclosed the invention, but also enabled the skilled person to work the invention without undue burden.


Takeda alleged that the Patent lacked novelty over three articles (Bihoreau, Shinkawa and Ferrara) and the prior use of a Novartis antibody called basilixumab (Simulect). Each of the pieces of prior art disclosed an antibody with the claimed levels of fucosylation. In the case of Simulect, the skilled person would have been able to analyse the glycosylation pattern of the antibody.

Roche argued that the prior art citations were not enabling, on the basis that the skilled person could not make the respective antibodies they disclosed. The judge found that this was not the correct test, however, and it was sufficient that the skilled person can produce a variation of the disclosure falling within the claim. Taking Bihoreau as an example, while the skilled person might not have been able to make the exact particular antibody disclosed in a prior art document that had the required level of fucosylation (in the absence of amino acid sequences) that does not mean the disclosure was not enabled. This piece of prior art reported the cell line used, even if not the expression vector or particular clones for producing the particular antibody disclosed. Whilst accepting that the skilled team would not be able to make the very antibody disclosed, the judge held that there was an enabling disclosure because the skilled team could make their own version of an antibody to the same target as disclosed in the prior document. The skilled team would express their own antibody sequence in the CHO-DG44 cells as described in the prior art document, they would make a number of subclones and screen for the level of fucose and galactose content. This would be a great deal of work, but the skilled person could make their own version of the antibody having the same fucose and galactose content as reported in the prior art document for the prior-disclosed antibody. The levels of NGNA and a-Gal were not disclosed, but the test for novelty was nonetheless satisfied, since the antibody produced by the skilled person would inevitably fall within the levels of NGNA and a-Gal set out in the claims. This was a result of the expression in CHO-DG44 cells.

Inventive step

The case highlights the difference between the tests for novelty and obviousness. Takeda did not run a conventional obviousness attack over Bihoreau, with the judge accepting Roche’s evidence, in the absence of evidence or submissions to the contrary from Takeda, that if the skilled team followed up the Bihoreau disclosure they would not have set about making antibodies with very high fucosylation levels.

The focus of the Bihoreau document was on fucose to galactose ratios and thus it was not obvious to produce an antibody with the very high fucose levels of the claims, even if there was one antibody that did have the claimed levels of fucose (referred to above under novelty).

Instead, Takeda argued that the Patent offered no technical contribution to the art and thus lacked inventive step on this basis. Takeda relied on the general principle, arising from the EPO Technical Boards of Appeal case Agrevo/Triazoles T 939/92, that the patent monopoly as defined by the claims should correspond to the technical contribution to the art. Roche sought to advance three technical contributions centred around the high level of fucosylation, how it can be achieved in practice and its therapeutic utility. The judge adopted five questions in his assessment – Is the alleged technical contribution disclosed in the patent? Is it plausible? Is it true? Is it a technical advance? Does it support claims of the breadth they are?

Each of Roche’s alleged technical advances failed on the basis that they either did not represent a technical advance, or were not rendered plausible by the Patent. Roche argued that a technical contribution made by the Patent was that 99% fucosylation of an antibody reduces ADCC to background. However, it was not clear that this was true at concentrations of antibody higher than the specific concentrations of antibody for which data were presented in the Patent and accordingly this was not plausible. In relation to the level of fucosylation, this feature alone could not support an inventive step since the contribution of the alleged technical advance was limited to antibodies expressed in CHO cells, and fucosylation was well known at the priority date to depend on cell type. Finally the idea of using increased fucose levels for a therapeutically useful purpose (the third alleged technical contribution) was not a technical advance over what was known as the idea that reducing ADCC was therapeutically useful for certain antibodies was part of the CGK and so too was the idea that increasing fucose would reduce ADCC. As a result the judge held that Patent invalid under the head of obviousness for lack of technical contribution.


Takeda’s case that the Patent was insufficient was based on allegations of ambiguity, breadth of claim (linked to the lack of technical contribution) and classic insufficiency. The judge found that the claims were truly ambiguous and invalid. This was grounded in the expert evidence that the skilled team, when given the patent, would think they could analyse peptides using two different analytical methods (two different types of liquid chromatography-mass spectrometry) and, depending on which one they used, they would get a different result on the characterisation of the antibody glycosylation pattern.

It will be interesting to see whether the judge’s approach to assessing novelty and the technical contribution will be adopted more widely.

[1] Click for the full judgment: Takeda UK v F. Hoffman-La Roche, [2019] EWHC 1911 (Pat)

Gregory Bacon


Chloe Dickson