Round up of recent EPO Board of Appeal decisions relating to antibodies

We have carried out our annual review of EPO Board of Appeal decisions published up to the end of 2025 where the claimed subject matter relates to antibodies. We have selected those decisions that have points of interest and summarised these under the headings of (i) clarity, (ii) novelty, (iii) inventive step, and (iv) sufficiency of disclosure.

Our review of these decisions has identified two areas of particular interest which deserve highlighting as they may well help us not only in drafting and prosecuting antibody cases, but also in framing oppositions and appeals to maximum advantage.

Key Areas of Interest

1. Use of post-filed data following G2/G1

This Enlarged Board decision allowed applicants and proprietors to use post-filed data as evidence of a technical effect relied upon for inventive step, if the skilled person would have considered that technical effect was “encompassed by the technical teaching [of the application as filed] and embodied by the same originally disclosed invention”. While that decision primarily concerned inventive step, the Enlarged Board also commented on the admissibility of post-filing data to support sufficiency of disclosure when the technical effect is recited in the claims (for example in medical use claims, where attainment of the claimed therapeutic effect is considered a limiting technical feature of the claims). The Enlarged Board held that the scope for reliance on post-filing evidence of a technical effect is much narrower for sufficiency than it is for inventive step.

Here are two cases published recently which illustrate this as post-published supporting data was not allowed in one case (T294/20) and was in the other (T1515/23).

In T294/20 the claim was directed to a second medical use (antibody for use in treating disease X) and so the technical effect relied on for patentability was considered under sufficiency of disclosure. In this case, the board held that G2/21 confirmed that post-published evidence cannot be used to remedy a lack of sufficiency of disclosure in respect of a medical use if it would not be credible to the skilled person from the application as filed that the therapeutic effect is achieved. They even suggested that the Enlarged Board decision could be understood to suggest that a high standard of proof was required, i.e. “beyond reasonable doubt” as opposed to on “the balance of probabilities”, which could mean direct experimental proof of the effect was needed. Though they later conceded that G2/G1 did not address the question of proof, they stated that this has been part of the consideration of “plausibility” in the board of appeal case law that was considered by the Enlarged Board. The Board concluded that the principles set out in T609/02 remained valid and that the “suitability” of the product for the claimed therapeutic application must be derivable from the application (unless it is already known in the art) and that suitability requires that the claimed compound has a direct effect on the metabolic mechanism specifically involved in the disease. In this case, the claims were directed to a PD-1 blockading antibody for use in treating a particular form of Hodgkin Lymphoma (NLPHL). The Proprietor relied on statements in the description concerning the enhancement of an immune response to an infectious agent using a compound that reduced PD-1 expression/activity. They argued that Example 1 showed chronic viral infections were associated with PD-1 expression; exhausted T cells were associated with PD-1 expression; and inhibition of PD-1 expression resulted in restored T-cell activity. The Proprietor argued that the link to the treatment of NLPHL was provided in Example 9 which showed that PD-1 expression was seen in 14 cancer samples and that the skilled person would have understood from the art that the chronic infection model used in Example 1 was used for investigating immune responses in an environment of exhausted T cells which could equally apply cancer as well as infections. However, board found that this amounted to no more than a “vague indication of the suitability of the claimed compound for the therapeutic effect” and was not sufficient to allow post-filed evidence to remedy the deficiency and the claim was held insufficiently disclosed.

In T1515/23, the Board considered a claim directed to an antibody defined by its sequence. Compared to the closest prior art document, the claimed antibody had an engineered heavy chain constant region with several amino acid differences. The Applicant argued that the technical effect of the claimed antibody was a reduced ability to stimulate undesired immune responses and provided post-filed evidence of this effect. The Board accepted the post-filed evidence in line with G2/G1 because the application as filed made a statement “In certain embodiments, the antibody….. comprises an altered constant region, wherein said antibody…exhibits decreased effector function relative to an anti-CDCP1 antibody with a native constant region.” No data was provided in the application to support this statement, but the Board found it sufficient to allow post-filed data to support the technical effect and held that the claim was inventive. The Applicant was able to tread a relatively fine line between it being plausible that constant region mutations would reduce effector function (such that the unevidenced statement was not simply disregarded), and the specific claimed mutations being non-obvious from the prior art on the facts of the case (including that the closest prior art antibody used IgG4 constant region because of its lack of complement-inducing activity).

We know that the technical effect relied on for inventive step or sufficiency of disclosure can change during prosecution or opposition depending on the cited prior art and/or amendments to the claims (for example by amendment from product claims to medical use claims). It is therefore important when drafting applications to try to go beyond simply foreshadowing possible technical effects, but to support them by clear statements (preferably in the general part of the description) and, in the case of therapeutic applications, an explanation of how the antibody directly affects a metabolic mechanism that is specifically linked to the therapeutic application.

2. Care in defining epitopes

The EPO Examination Guidelines explicitly acknowledge that antibodies can be defined in terms of the epitope to which it binds. Earlier versions of the Guidelines discussed the different considerations for linear and discontinuous epitopes, but perhaps surprisingly, this has been removed from the 2024 and 2025 versions.

A number of recent EPO decisions have considered the disclosure requirements for a claim which defines the antibodies in relation to a specific epitope. Some consider whether the technical effect mentioned in the claim is achievable across the scope of the claim (inventive step) and others whether the invention can be worked across the scope of the claim (sufficiency of disclosure). However, the same considerations are made in both.

In T12552/22, the claim related to an antibody which bound to an epitope of human LAG-3, the epitope comprising a specific amino acid sequence (SEQ ID NO: 77) provided in the claim. The Proprietor asserted that this was associated with an improved technical effect over the closest prior art antibody (8B7), which bound to an adjacent epitope. In fact the two epitopes were overlapping but shifted by one amino acid meaning that an N-terminal histidine was present in the claimed epitope but not in the epitope listed for antibody 8B7.  There was a lot of discussion around claim construction and what the skilled person would understand by the functional requirement of “binds to an epitope”. The patent provided evidence that 8B7 bound to a different epitope by assigning SEQ ID NO: 77 as the epitope for their antibody 25F7 and SEQ ID NO: 78 to antibody 8B7 on the basis of a Peptide Scan assay.  The patent also stated that the peptide scan results indicated that “25F7 and 8B7 antibodies bind to different although closely located epitopes within human LAG-3".  The Proprietor argued that this meant that antibody 8B7 was not an embodiment of claim 1 because it did not bind to an epitope comprising SEQ ID NO: 77 (i.e. at least the N-terminal H in SEQ ID No. 77 was not bound by 8B7). The Proprietor further argued that the skilled person would understand this to mean specific, not arbitrary binding and as the antibody provided by the opponent (8B7) did not bind to the first residue of the sequence provided in the claim (unlike the Proprietors exemplified antibody, 25F7), it was not covered by the scope of the claim. However, the Board found that antibody 8B7, which did not exhibit the same improvement relied on for technical effect, was also within the scope of the claim, despite a difference in binding to one amino acid residue in the epitope. The Board stated that the skilled person would understand that not all residues provided in the epitope sequence would be expected to be bound by the antibody and that an absence in the claim of a method to determine the epitope and the extent of binding to the epitope, the broadest technically sensible interpretation of the term “epitope” has to be adopted and this included the non-working embodiment antibody 8B7. The Board decided that 8B7 “binds an epitope of human LAG-3 comprising the amino acid sequence of SEQ ID NO: 77”. The Board therefore held that the asserted technical problem was not solved across the scope of the claim, so the technical problem was broadened to the provision of alternative antibodies and the claims were found obvious. Surprisingly, in relation to an auxiliary request, the Board also considered that antibody 8B7 was encompassed by the language: “binds to the same epitope” on human LAG-3 as 25F7.

Similarly, in T435/20, a claim to antibody binding to an epitope defined by two group of amino acid sequences (discontinuous or conformational epitope) lacked sufficiency of disclosure. In this case the board held that the claimed antibodies are functionally defined by their ability to bind human IL-23p19 and contact several amino acids residues in both stretches of amino acid sequences specified in the claim. The board confirmed that the same specificity of the antibody exemplified in the patent was required but the antibodies may be structurally unrelated. When considering whether the skilled person would be faced with an undue burden in working the invention, the board held that the lack of a suitable antigen and a suitable screening method to select antibodies meant that the claimed subject matter was insufficiently disclosed. The board concluded that this “critical” information was missing from the application as filed and therefore no reliable method for working the invention was disclosed.

T 0326/22 reaffirms that claims to antibodies defined by their specific epitope binding and functional properties can be viable under the EPO approach to both sufficiency and inventive step. The Board upheld claims to anti-CD47 antibodies defined by their binding to a discontinuous epitope comprising 14 specified amino acids, and which prevent CD47 from interacting with SIRPα without causing significant cell agglutination.

The board rejected arguments that the generation of such antibodies would require undue burden. While the Board acknowledged that generating antibodies through immunisation involves some unpredictability, the patent provided guidance through its description of a complete process for generating and characterising the CD47 antibody 2A1. Crucially, the examples described details such as the antigen used for immunisation, screening assays for selecting SIRPalpha-blocking and non-cell agglutinating CD47 antibodies, structural information for exemplary antibodies, and cross-competitive binding assays for epitope binning. A final step of X-ray crystallography was therefore required only to confirm that screened antibodies bound to the specified epitope, not as a screening step.

This decision further underscores that discontinuous epitope claims remain viable in Europe in principle, though the amount of detail about the method used to arrive at the claimed antibody is perhaps unusual. It is crucial that the patent application, or established scientific knowledge, provides sufficient detail so a skilled person can produce new antibodies within the scope of the claims without undue difficulty.

In contrast with T 0326/22, T 1103/22 provides an example of an unsuccessful attempt to pursue protection for an antibody directed to a functionally-defined epitope. The antibody was defined as binding “at the active site located in the light chain region of factor XI”. The board held there was insufficient information in the application to permit the identification the location of the active site in the light chain region of factor XI at the relevant date, or the identification of antibodies binding to it.

These decisions underscore the importance for applicants pursuing epitope claims to provide as much information as possible about the epitope, and to describe how they obtained one or more specific antibodies which bind the claimed epitope, ideally describing how the antibodies were raised and any screening performed to identify them from a broader pool.

Since binding to an epitope can be assessed in different ways, it is of course also essential to specify the method(s) that are to be used to determine whether an antibody falls within the claim.

Summary of Selected Board of Appeal Decisions

(i) Clarity

Following the trend of antibody cases relating to antibody variants, T 1815/22 was concerned with a composition comprising an individual antibody (mepolizumab) in particular percentage proportions of oxidised variants and deamidated variants. The examining division initially rejected the application as unclear due to the absence of a method to measure the percentage of variants in the claim. At the Board of Appeal, the applicant successfully argued that such definitions were common in the relevant field and that established methods for measuring the percentage of variants existed, making it unnecessary to specify a measurement method in the claim. This is a common situation for applicants pursuing antibody claims which recite parameters, and this case may be a helpful one to cite when faced with such requests, particularly in situations where the parameter is well understood in the art.

(ii) Novelty

In T 0203/22 the board examined whether the bispecific T-cell engager defined in claim 5 was novel over prior art disclosing various multispecific antibodies targeting different tumour antigens. The board held that to arrive at the particular bispecific T-cell engager in claim 5, several selections from alternative antibody formats and targets in the prior art would be required. As there was no clear preference or explicit disclosure of the claimed combination in the prior art, the board concluded that claim 5 was novel. While the reasoning seems relatively straightforward, this is a decision which may be helpful to cite for applicants pursuing protection for bispecific antibodies who often face prior art that discloses the two target antigens but does not specifically point towards the combination of these two specificities.

Decisions T 1731/22 and T 0358/22 both related to situations where the novelty of antibody-related composition of matter claims were debated in light of prior art that described methods of production of antibodies.

The claim at issue in T 1731/22 was directed to an antibody defined with reference to multiple functional effects. The board considered that following the method disclosed in the closest prior art would inevitably lead to the identification of an antibody (scFv) which would meet the claimed functional criteria. The proprietor argued that starting from a high number of phage particles in the libraries used in the closest prior art document, it was not certain that the skilled person would obtain an antibody having the features as claimed, as each of the steps required to obtain this result was associated with uncertainties, making the end result unpredictable.  The board dismissed this argument, highlighting that the completeness and enablement of the disclosure is not affected by the fact that several steps are needed in order to arrive at the final result or the fact that that these steps select clones from a large number of initial phages in the library. The board held that the skilled person, even in the event of an isolated failure, would repeat the method to arrive at an antibody having the claimed features.

In contrast, in T 0358/22 the opponents pointed to the disclosure of standard recombinant production methods for pertuzumab in CHO cells and argued that such processes would inevitably result in antibody compositions containing acidic variants as defined in the claims. The board referenced the established case law which holds that a product is considered to be disclosed even if it is not cited expressis verbis in a prior art document, if it is the inevitable result of a process defined in the prior art document (see T 12/81, Headnote 1). However, the board noted that it is essential in this context that the process be adequately and credibly described in the prior publication, including the starting substances and reaction conditions. The board found that the disclosure of the production methods in D1 was only in broad, general terms (without specifying the individualised process conditions necessary to assert that a particular, well-defined product will always result). Therefore, the board held it could not be concluded that the claimed composition containing disulfide reduced variants inevitably resulted from the instructions in D1.

These decisions provide some guidance to applicants as to how to navigate prior art disclosing methods of production which may lead to the claimed antibody/antibody composition. Important distinctions leading to the differing outcomes in these cases appear to be the level of detail in the disclosure of the prior art method, and the fact that in T 1731/22 the prior art document was aimed at obtaining an antibody with the claimed features, whereas there was no such teaching in the prior art of T 0358/22.

(iii) Inventive Step

Also mentioned above in the novelty section, T 0203/22 is also notable for the reasoning on inventive step of claim 1, which was directed to an anti-B7-H6 chimeric antigen receptor (CAR). The board accepted that skilled person would not have expected the improved IFNgamma production observed with anti-B7-H6 CAR T cells compared with NKp30 CAR T cells. This was despite the fact that B7-H6-specific antibodies capable of blocking NKp30 CAR-dependent T cell activation were known and assumed to be suitable for constructing a CAR. Also, the structure of CARs was common general knowledge at that time (points 22 and 23 of the reasons). Interestingly, in this decision the B7-H6-specific antibodies were not considered the closest prior art to the B7-H6-specific CAR. Instead, the board compared the claimed invention to NKp30 CAR T cells.

The endorsement of the approach of starting from the nearest CAR, not from a monoclonal antibody, as seen in T 0203/22, may provide a useful angle for applicants defending CAR claims faced with inventive step objections based on a corresponding antibody in combination with assertions that common general knowledge of the structure of CARs would give the skilled person a reasonable expectation of therapeutic success in a CAR context.

In T0419/16 we are reminded that considering inventive step from one closest prior art document may be efficient, it is not appropriate if there are alternative different routes by which the antibodies of the invention may be reached. In this case, all possible starting points must be considered.

The claims at issue in T 0810/22 concerned specific formulations of an anti-PD1 antibody, (pembrolizumab), for use in therapy. The liquid formulation claimed corresponds to the product marketed as Keytruda^® – the top selling pharmaceutical product in 2024 with sales of 29.5 billion USD. The prior art disclosed therapeutic uses of pembrolizumab and generally proposed formulations of the antibodies as aqueous solutions or suspensions, but did not disclose any specific formulation suitable for human use. The proprietor asserted that the claimed formulation was associated with a technical effect of a high degree of stability required for Phase III clinical trials and marketing. The Board agreed, formulating the objective technical problem as the provision of a formulation of pembrolizumab that can be marketed for human therapeutic use. Despite disclosures of approaches for formulation of biopharmaceuticals in other prior art documents, the board held that it would have required hindsight to arrive at the specific claimed formulation from among other suitable formulations. Notably the Board considered that the claimed formulation was the result of an optimisation process for late clinical development and marketing rather than the outcome of a quick routine search for a formulation sufficiently stable for early clinical trials which, by chance, appeared to have an enhanced effect. The Board considered that such an optimisation process required an amount of time and effort that went beyond routine work. Thus, the formulation claims were deemed to be inventive over the prior art. This decision is a useful endorsement of claims covering the commercial formulation of an antibody product, which can provide a useful additional (albeit narrow) layer of protection for innovators.

T 0219/22 concerned a SerpinF2-binding antibody “for use in preventing or treating brain haemorrhage or brain edema resulting from cerebral ischaemia caused by thrombotic ischaemic stroke”. The closest prior art linked a reduction in circulating SerpinF2 levels to a lower risk of intracerebral haemorrhage in acute ischaemic stroke. However, the prior art studies were performed with rats rather than humans, used microplasmin to lower SerpinF2 levels rather than an antibody. The distinguishing features of the claim were therefore its specification of treatment in human patients and the use of an antibody instead of microplasmin. Critically, the board noted that the prior art suggested antibodies as alternatives to microplasmin for depleting SerpinF2. Thus, the board held that a skilled person, seeking alternative methods to reduce SerpinF2, would find it obvious from the prior art that a SerpinF2-neutralising antibody could be effective. The board rejected the significance of the rat/human difference, finding that the use of the rat model was ultimately aimed at finding solutions for human patients. As a result, the patent was revoked for lack of inventive step.

(iv) Sufficiency

T435/20 (also discussed above) considered a claim directed to an antibody defined by the epitope to which it binds. The epitope was defined by two non-contiguous amino acid sequences, i.e. a discontinuous or conformational epitope.  While the board acknowledged that raising and screening antibodies was routine, they stated that this nevertheless require knowledge of the antigens suitable of raising antibodies having the desired properties and knowledge of the screening process for their selection. The primary sequences of the claimed conformational epitope are unsuitable for raising the claimed antibodies or screening them. In the absence of details of how the exemplified antibody was actually produced, the application was held insufficient to enable the claimed invention.

This approach to “suitability” was also seen in T 0203/22 where the board held that the mere assertion that functional properties of the antibody enable the therapeutic effect is not sufficient. They stated that the application as filed must provide clear and direct disclosure demonstrating that the functionally defined antibody is indeed suitable for achieving the claimed therapeutic effect. The functional-therapeutic link must be derivable from the application as filed.