1 February 2023
  • Share

The prosecution of patent applications relating to antibodies involves particular considerations. This is reflected in the EPO’s recent (arguably belated) inclusion of a dedicated section in its Guidelines to assist examiners working on patent applications in this area.

We have conducted a review of Board of Appeal decisions relating to antibody patents published since the start of 2020 to review recent developments in jurisprudence in this field and to determine how these developments might inform best practice for applicants.

A major consideration in patent applications related to antibodies is how the antibody should be defined. The EPO Guidelines provide a non-exhaustive list of characteristics by which antibodies can be defined:

            (i) by structure (i.e. sequence);

            (ii) by reference to the antigen or epitope recognised by the antibody;

            (iii) by structural and functional features;

            (iv) by antigen (or epitope) and functional features

            (v) by production process; and

            (vi) by the hybridoma which produces the antibody.

In this article, the first of a two-part series relating to recent developments in antibody case law, we will look at the decisions which explore these various ways of defining antibodies and the specific considerations which may apply. The second part of the series, to be published later this spring, will review the case law relating to inventive step of antibodies.

In general, in the decisions we have reviewed, the Boards of Appeal have approached the issues at hand in a similar way to our experience of Examining Divisions during routine prosecution, suggesting a reasonable level of consistency to the approach of the various levels of the EPO to the subject. Some points raised by the Boards provide guidance on how the EPO Guidelines should be applied, which we review below.

Definition by Reference to the Antigen Recognised by the Antibody

As set out in the Guidelines, it is in principle possible to define antibodies by reference to the antigen(s) to which they bind. Such claims have the advantage of a broad scope which encompasses any antibody which binds the defined antigen(s). However, this breadth is a double-edged sword, and means these claims have become progressively more challenging to obtain as the antibody field has developed and novel antibody targets become increasingly few and far between.

Claims relating to an antibody purely defined by the antigen to which it binds were considered in T 2423/18. The patent under appeal related to the use of antibodies against the MHC class I polypeptide-related sequence A protein (MICA) in cancer treatment.

MICA is a highly polymorphic protein, with over 50 different human alleles known. The claim considered by the Board was directed to an antibody that binds to four of these alleles on the surface of cells, for use in the treatment of cancer. The Board also considered a claim which additionally required the antibody to have a human heavy chain constant region, or to be conjugated or covalently bound to a toxic agent.

The board accepted that the closest prior art was a document related to the use of anti-MICA antibodies in treating cancer, without disclosing a concrete monoclonal anti-MICA antibody and without considering the polymorphism of MICA. The Board considered that the polymorphism was a matter of general knowledge and that the skilled person would therefore have been motivated to seek antibodies which bound to more than one allele. A second document related to a mouse antibody which was expressly taught to bind the same four alleles mentioned in the claims, albeit that this mouse antibody was taught only for in vitro and experimental use. The Board found that the skilled person would have considered “modelling suitable antibodies” for therapeutic use on the binding characteristics of this mouse antibody. They concluded that the claims lacked inventive step.

The applicant also argued that the claims required the antibody to bind all four specified alleles with a high enough affinity to treat cancer expressing them, whereas the mouse antibody bound one of the four alleles with too low an affinity to be effective in treating cancer. However, the Board disagreed with the applicant’s interpretation of the claims, taking the view that the wording of the claim separately required (any level of) binding to the four alleles on the one hand and treatment of cancer (e.g., expressing any one of the alleles) on the other. The applicant’s argument was therefore disregarded. Unfortunately we therefore have no guidance on what the outcome would have been if the Board had interpreted the claim to require binding characteristics distinct from the mouse antibody.

This case highlights the difficulty mentioned above, that while it is permissible at the EPO to define an antibody by its target antigen alone, ever fewer novel targets for antibodies exist. Indeed, as seen here, even defining an antibody as binding multiple specific targets may not be enough to avoid prior art issues. In addition, this case illustrates that prior art issues can arise from mouse or other experimental antibodies, and not just from antibodies that are expressly envisaged for the same therapeutic purpose.

Definition by Antigen and Functional Features

A recent case to have highlighted the acceptability of claims in which the combination of antigen plus function are used to define an antibody is T 0499/18. The patent which was the subject of the appeal related to antibodies against human R-spondin (RSPO) and human leucine-rich repeat-containing G protein-coupled receptor 5 (LGR5). Claim 1 of the Main Request was directed to an antibody defined as specifically binding RSPO, and having the functionalities of (i) inhibiting the growth of a solid tumour comprising solid tumour stem cells, and (ii) disrupting binding of RSPO to LGR5 and/or disrupting activation of LGR5 signalling. Claim 5 of the Main Request was directed to an antibody defined as specifically binding LGR5, and having the same functionalities as the antibody of claim 1.

According to the patentee, the patent under appeal demonstrated in vitro (in an assay utilising HEK293 cells) that RSPO and LGR5 interacted with each other to activate β-catenin signalling. The patentee argued that it was common knowledge that the β-catenin/Wnt signalling pathway was associated with cancer and that inhibition of activation of this pathway using an antibody against RSPO or LGR5 would be effective in cancer treatment.

The OD had revoked the patent after being persuaded by the opponent that it was not plausible that the data in the patent demonstrated an interaction between RSPO and LGR5, in view of data showing that HEK293T cells did not appear to express LGR5. The Board concluded that the OD had been wrong on the facts to come to that conclusion, and overturned their finding of insufficiency. No issues were raised with the functional language by which the antibody was defined. The patent was remitted to the OD for assessment of novelty and inventive step, which remains ongoing.

Definition Based on the Epitope Recognised and Functional Features

Although not ruled out in theory, claims based on the epitope which is bound are essentially impossible to obtain in the US due to strict written description requirements.

In comparison, the EPO’s Guidelines explicitly state that definition by an epitope is allowable, but including the cautionary message that “the application must provide an enabling disclosure allowing the skilled person to determine whether further antibodies bind this epitope. The application must also enable the production without undue burden of additional antibodies binding to the same epitope” (G-II-

The question of undue burden in producing antibodies against a specified target was considered in T 2416/18, an appeal by the patentee against the decision of the Opposition Division to maintain the underlying patent in amended form. The patent had originally been granted to an antibody against influenza A haemagglutinin (HA), capable of neutralising at least one H1 influenza A subtype and at least one H3 influenza A subtype (i.e. the antibody was defined based on the targets to which it could bind and its biological activity), but at opposition the claims had been limited to a specific antibody (defined by its variable domain sequences) with these capabilities. The patentee appealed, seeking to restore the claims as originally granted.

It was common ground between the parties that an antibody as claimed had to bind an epitope present on both H1 and H3 haemagglutinin subtypes. However, no such epitope was disclosed within the patent. Rather, a single antibody with the required cross-reactivity was disclosed, and its epitope was not determined.

While the patent acknowledged that such cross-reactive antibodies were rare, it disclosed selection criteria according to which individuals likely to produce antibodies with the required cross-reactivity could supposedly be identified. The patentee argued that these criteria enabled the reliable identification of further antibodies falling within the claims. The selection criteria included that the individuals had high antibody titres and high neutralising titres against the H1N1 and H3N2 influenza subtypes responsible for the seasonal epidemics in the previous 5 years, and detectable neutralising titres against one specific reference H1N1 virus and one specific reference H3N2 virus. However, the Board agreed with the Opponent that since these tests would be performed on sera containing a polyclonal mix of antibodies, these characteristics would most likely identify individuals who had generated separate immune responses to the different influenza subtypes, rather than a cross-neutralising immune response to both H1 and H3 influenza subtypes. The Board therefore held that the selection criteria in the patent did not constitute a reliable method to obtain further antibodies with the cross-neutralising activity required by the claims.

Moreover, due to the rarity of the antibodies, isolation of the single antibody disclosed in the patent which fell within the claims by following the disclosed criteria was considered a “single lucky event”.

In the absence of a reliable method to obtain further antibodies with the required cross-reactivity (which did not require luck), the claims were considered insufficient and the appeal was dismissed.

This decision highlights the message of the Guidelines quoted above, that antibodies can only be defined by reference to their epitope if further antibodies against the same epitope can be raised by the skilled person without an undue burden. If no reliable methods of generating and selecting antibodies with the requisite activity can be provided, it is unlikely to be possible to obtain claims which extend beyond the specific antibodies disclosed, as proved true in the case at hand.

Another approach sometimes used by applicants is to define antibodies as binding to the same epitope as a reference antibody (without this epitope being defined). Our experience is that EPO examiners can sometimes object to such language as unclear and/or insufficient, depending on the depth of disclosure in the application. We are aware of two recent Board of Appeal cases (T 2258/15 and T 1964/18) in which such language appeared in the claims. Unfortunately, in neither case were the arguments focused on the allowability of this language. Therefore, these cases do not offer specific reasoning which would allow applicants to push back against such objections. We are closely watching this space, and in the meantime, can advise on drafting strategies which may increase the likelihood of claims of this type being allowed. 


T2258/15 is interesting because it highlights the need to show that the functional features relied on for inventive step are actually related to the epitope defined in the claim. In that case, the antibodies were defined as binding the same epitope as one of two reference antibodies. The patentee argued that the reference antibodies in the claim displayed good internalisation properties and attempted to rely on that feature for inventive step. However, the Board held that there was no evidence that the internalisation properties were related to the epitopes bound, and hence disregarded this feature, leading to a finding of lack of inventive step.

Epitopes defined by reference to a linear sequence have generally been considered acceptable by the EPO in principle. For instance, in T 1872/16, the antibody was defined by two characteristics: (i) the epitope it binds to on hIL-13 and (ii) the ability to modulate the interaction between hIL-13 and its receptor. The epitope was defined as: "set forth in SEQ ID NO: 84 of SEQ ID NO: 9 between residues 103 and 107 inclusively". There was some debate in the case about the extent to which the claims encompass antibodies binding both within and beyond the defined region, highlighting that there may be complexities in exactly how the epitope should be defined. No final decision was reached on this point because the patent was found to lack inventive step for other reasons. However, it appears that the Board would have been willing to apply the usual principles of claim construction, had it been necessary to do so.

There is currently little case law from the Boards in relation to claims in which antibodies are defined as binding a discontinuous epitope, but there are cases heading to appeal which will be interesting to watch. Of key interest is the appeal T 0326/22, relating to EP 2812443. The patent relates to anti-CD47 antibodies and was granted with claims which defined the antibody as binding a discontinuous epitope comprising 15 residues spaced across a 70 residue stretch of the protein, and as having the function of preventing CD47 from interacting with SIRPα. The patent was opposed for lack of sufficiency with the Opponent arguing that the patent allegedly did not provide an enabling disclosure allowing the skilled person to determine whether further antibodies bind this epitope, and because finding further antibodies binding to the same epitope would represent a research project. The Opposition Division found claim 1 as granted to be sufficiently disclosed, because in their opinion only a limited number of paratopes would match the epitope; furthermore, the patent disclosed a particular antibody and the skilled person would be able to vary the CDRs and determine whether the changes have an impact on the recognised epitope. The Opponent has appealed and it is to be hoped that the decision will provide some clarity on this interesting and important issue, though the decision cannot be expected for a couple of years.

Definition by Structural and Functional Features

In the context of patent applications directed to new antibodies, it is most common for applicants to define the antibody by its sequence. A common question is how narrowly the antibody sequence must be defined.

The Guidelines state that, as a rule, when defining antibodies by sequence alone, the sequences of all six complementarity-determining regions (CDRs) must be specified. The Guidelines sets out exceptions for antibody formats which do not require six CDRs, or when it is experimentally shown that one or more of the six CDRs is not required for target binding (G‑II-

The Guidelines also state that it is possible to claim an antibody characterised by the sequences of both variable domains or CDRs with less than 100 % identity when combined with a clear functional feature. (G-II- However, in our experience, even when a functional feature is present, it is common for examiners to insist that all six CDRs are fully defined.

The extent of allowable variation was explored in T 0941/16, an appeal against the refusal of a patent application relating to antibodies against PSMA (Prostate Specific Membrane Antigen). The application disclosed specific murine anti-PSMA antibodies including as 3/F11 and 3/E7, and claim 1 of the Main Request was directed to an anti-PSMA antibody comprising “at least three” of the CDRs of antibody 3/F11 or antibody 3/E7 and having certain functional characteristics. The applicant also filed Auxiliary Requests which specified that the claimed antibody comprised “at least five” of the CDRs.

The applicant’s position was that the humanization process would allow the skilled person to modify up to three CDRs (or possibly even more) while still retaining the functional properties of the original antibodies. In support of their position the applicant submitted experimental evidence of humanised antibodies, showing that VHCDR2 of 3/F11 could tolerate at least 6 amino acid substitutions, and that VLCDR1 and VLCDR2 of 3/F11 could each tolerate at least 2 amino acid substitutions.

Interestingly, the Appeal Board did not entirely dismiss the idea that some variability in the CDRs may be permissible. Rather, they conducted a detailed technical analysis, and as a result of this concluded that the claimed antibody could not be obtained over the whole breadth of the claim. In particular, the Appeal Board cited publications which demonstrated a particular importance of VHCDR3 and VLCDR3 in antibody specificity and affinity, and objected that the Main Request did not require either the VLCDR3 or VHCDR3 of the exemplified antibodies to be retained in the claimed antibody. The applicant had not provided any evidence to demonstrate that either CDR3 sequence could be varied without loss of functionality. The Board concluded that it was not credible that an antibody with the properties of those exemplified would be obtained unless the original CDR3 sequences were present, and thus refused the application for insufficiency. The Auxiliary requests were rejected for the same reason.

The applicant also argued that antibody formats such as camelised antibodies were possible, containing only three CDRs. However, this argument was quickly dismissed by the Board based on the fact that the claims were not limited to such formats.

The detailed technical analysis undertaken by the Board in this case is interesting because it seems to allow for the possibility for some flexibility in allowing for variation in the CDRs when the scope of the claim is more closely aligned with the supporting evidence. For instance, while this is necessarily speculative because such a request was not present by the proprietor, it seems possible that claims which specified the CDR3 sequences, and allowed for some variability in the other CDRs, might have been allowed by the Board. Certainly this seems to offer some guidance for applicants in drafting applications.

Definition based on Hybridoma which Produces Antibody

The Guidelines state that it is possible to define an antibody based on the hybridoma which produces it. However, the decision in T 32/17 suggests that there are limitations to the allowability of such claims. In the case at hand, an antibody was claimed which was specified to (i) recognise 25-hydroxyvitamin D2 and 25-hydroxyvitamin D3, and (ii) to be produced from one of 5 specific deposited hybridomas.

Applying the existing principles of the case law of the Boards of Appeal relating to product by process claims, the Board held that the claims lacked novelty over a prior art antibody capable of binding the same target. The Board held that there was no evidence in the patent that antibodies produced from the specified hybridomas had a technical effect not achieved by the prior art antibody. There was also no evidence that the antibodies from the hybridomas had different sequences to the prior art antibody, and it was held that the burden to prove this was on the patentee. In the absence of such evidence, the claims were found to lack novelty.

The decision of the Board ultimately rested on a lack of evidence for the alleged distinguishing feature. On the face of it, this evidence could have been provided by the proprietor after filing, e.g., during the opposition proceedings. However, certain comments of the Board also appear to cast into question the fundamental suitability of hybridoma-based definitions, at least when no sequence information is provided in the patent. For instance, the Board states that “The skilled person following the teaching of the patent must inevitably achieve that characteristic and must be aware of that characteristic so that they can recognise the claimed product and discard any products not having it…At issue in the present case is thus whether or not the process feature in claim 2 "produced from a hybridoma selected from the group consisting of the hybridomas deposited (...)" gives rise to the specific amino acid sequence and chemical composition of the claimed antibodies and furthermore, whether the skilled person is made aware of these structural characteristics of the antibodies by the teaching of the patent.” (Emphasis added).

It is rare to see a claim these days which uses a definition based on a hybridoma deposit rather than a sequence. Nonetheless, these comments from the Board strongly suggest that sequence-based definitions should be the standard, despite the suggestion in the Guidelines that hybridoma-based definitions may also be used.


The last few years have seen several interesting Board of Appeal decisions relating to the patentability of antibodies and the different ways in which they may be defined in patent claims. Many of the decisions reviewed above reinforce the existing Guidelines relating to the requirements for defining an antibody, and some seem to give basis to push back against EPO Examiners who insist on an overly-strict structural definition. There also some significant cases pending in appeal, particularly relating to definition of antibodies in terms of their epitope. It is to be hoped that these will provide further clarity on when epitope-based definitions may be allowed and what sort of support is required in the patent. We will of course report on these decisions when they issue - watch this space!


About the authors

This article was written by Rebecca Tollervey, Alex Galbraith and Edward Couchman.


Alex headshot-1 (4)


Alex Galbraith 

Alex is a trainee patent attorney working in our life sciences team. He has a BSc degree in Biological Sciences from Durham University where he  was awarded the Biological Sciences Prize. His final year research project focused on the expression of a novel recombinant bio-pesticide protein,  while he also gained industrial experience working with CRISPR base-editing technology within a large pharmaceutical company during his degree.

Email: alex.galbraith@mewburn.com



Blank Author Circle


Edward Couchman

Ed is an Associate in our life sciences team. He is experienced in drafting patent applications and prosecuting them around the world, with a particular focus on the fields of immunotherapy and enzyme technology. Ed is also experienced in carrying out freedom-to-operate (FTO) analysis for clients looking to bring new products to market. His clients in the life sciences field are from the UK and internationally, from university spin-outs and SMEs to multinational biotech corporations, and universities and research institutes.

Email: edward.couchman@mewburn.com

Rebecca is a Partner and Patent Attorney at Mewburn Ellis. She focuses on European prosecution, opposition and appeal work in the life sciences sector. She has secured allowance for a number of European applications relating to antibody products in clinical trials and has a strong track record of success in both defensive and offensive opposition proceedings. Rebecca’s clients include multinational healthcare and biotechnology companies, as well as US and European life science companies of all sizes and US law firms.

Sign up to our newsletter: Forward - news, insights and features

Our People

Our IP specialists work at all stage of the IP life cycle and provide strategic advice about patent, trade mark and registered designs, as well as any IP-related disputes and legal and commercial requirements.


Contact Us

We have an easily-accessible office in central London, as well as a number of regional offices throughout the UK and an office in Munich, Germany.  We’d love to hear from you, so please get in touch.