Review Of EPO Antibody Decisions In 2024

Statistics

We identified 43 decisions published between 31 January 2024 and 31 December 2024 which refer to the term "antibody". We excluded 21 decisions from further consideration due to not being directly concerned with antibody inventions. Of the remaining 22 decisions, 18 concern appeals from Opposition Division decisions, and 4 concern appeals from Examining Division decisions. Board 3.3.04 retained its position as the most prolific "antibody Board" issuing 15 of the 22 decisions (Board 3.3.04 issued 11 of 18 considered in our 2023 review). 4 decisions were issued by Board 3.3.08 (5/18 in 2023), 2 by Board 3.3.07 (2/18 in 2023) and 1 by Board 3.3.10 (none in 2023). We discerned no obvious differences in practice between those Boards.

Repeating the trend from previous years, a significant majority of the 22 decisions were decided primarily in relation to inventive step (Article 56 EPC) or sufficiency (Article 83 EPC), or both. There is often considerable overlap between Article 56 EPC and Article 83 EPC for antibody inventions. We found the EPO's assessment of both Articles to be consistent across all of the relevant decisions. Outside of Article 56 EPC and Article 83 EPC, one decision (T 1815/22) was primarily concerned with clarity (Article 84 EPC) – see further discussion in the next section.

The 22 decisions can also be divided into four general categories of invention:

• Broad / 1st Generation: "anti-target" antibody inventions typically relating to the identification of the target and/or a link to a previously unappreciated medical use. Includes antibodies which are defined by binding to a specific epitope sequence ("anti-epitope" claims). Also includes antibody portions of more complex molecules when defined with broad "anti-target" language. May have little or no disclosure of individual antibodies that bind to the target.
• Specific / 2nd Generation: inventions relating to an individual antibody or antibodies, defined by CDR or variable region sequences, and directed to a well-characterised target for which prior art antibodies (often therapeutic antibodies) are known. The EPO assumes that all antibodies specific for a known target are obvious unless an unexpected effect is plausibly demonstrated.
• Downstream / 3rd Generation: inventions relating to advances which typically arise during downstream development of a pharmaceutical, such as combinations with other therapeutics, alternative medical uses, patient subgroups, dosing regimens, compositions and partner diagnostics. The target binding properties of the antibodies are often known.
• Non-target / General: inventions that are not concerned with individual antibodies or targets, but rather with platform technologies such as new formats or methods of manufacture.

The largest category in 2024 (14 decisions – 6 granted/maintained, 8 revoked/refused) is "Downstream/3^rd Generation" inventions. This follows a trend we anticipated last year. The number of decisions in this category may reflect the growing maturity of the antibody field, including the emergence of biosimilars. Producers of biosimilars are likely to be interested in the (usually) later-expiring patents in this category.

The next largest category (5 decisions – 4 granted/maintained, 1 revoked/refused) is "Broad/1^st Generation" inventions. This is an interesting development. Although there are now relatively few new targets and so relatively few "simple" antibody inventions in this category, it is clear that broad "anti-target" inventions continue to be pursued in other contexts, such as parts of more complex molecules and previously unappreciated medical uses.

The remaining 3 decisions are split between the "Non-target/General" category (2 decisions – 1 granted/maintained, 1 revoked/refused) and the "Specific/2nd Generation" (1 decision – 1 granted/maintained) category. We do not see any particular new trends in either category. It remains the case that there are large numbers of "specific" antibody inventions prosecuted at the EPO but relatively few Board of Appeal decisions in this category. We suspect this is because the narrow scope of the claims is less likely to be subject to opposition.

Detailed discussion of individual decisions by category

Broad/1^st Generation Decisions

T 0203/22 was an appeal against a decision of the opposition division to maintain the patent as amended. The decision is interesting at a general level because it shows that in the context of a more complex molecule (here a chimeric antigen receptor or a bispecific antibody) the EPO is still willing to accept a definition of antibody binding domain which uses broad "anti-target" language. Much of the discussion in the decision concerned whether a specific embodiment (described as a bispecific T cell engager – BiTE) having anti-B7-H6 and anti-CD3 scFv binding arms was inventive. The Board held that the skilled person would not have been motivated to combine the anti-B7-H6 antibodies of the closest prior art with any anti-CD3 antibodies or fragments to arrive at the BiTE as claimed.

T 1019/22 concerned a bispecific polypeptide comprising first and second immunoglobulin single variable domains (ISVs), each targeting a different specific antigen on the same cell, with the first ISV having lower affinity for its target than the second ISV (defined by specific KD ranges). Binding of the second ISV to its target enhances the binding of the first ISV to its target, the function of which is then inhibited. The Board held that the closest prior art did not provide any clear pointer towards using binding domains with different affinities for targeting and inhibition and that multiple selections from the art were required to arrive at the specific combination of features as claimed. As with the previous decision, the Board saw no problem with the use of broad "anti-target" language in the definition of each ISV arm in the bispecific molecule. There is though an additional dispute as to whether the parties had been fully heard with respect to an alternative closest prior art document. The case has been remitted to the opposition division for further consideration.

T 0419/16 concerns an IL33 antagonist antibody for use as a medicament (i.e. a first medical use claim). Although the scope of the claim is limited to the antibody for use as a medicament, the broad "anti-target" language encompasses any suitable antibody for such use. The closest prior art concerned a murine receptor for which the ligand was subsequently shown to be IL33, but there was no prior recognition that IL33 was the ligand nor that anti-IL33 could be useful in therapy. Accordingly, the Board acknowledged inventive step for anti-IL33 in general for any medical use, because there was no prior recognition of such utility. It is noteworthy also that the application includes data showing a medical effect for only a single, undefined mouse anti-IL33 antibody.

T 1731/22 concerned a monoclonal antibody capable of binding to a stabilized soluble α‑synuclein oligomer (SASO). The proprietor sought to argue that use of SASO stabilised with 4-hydroxy-2-nonenal and/or 4-oxo-2-nonenal as the antigen leads to the generation of different antibodies as compared to other forms of SASO. The closest prior art disclosed SASO stabilized on a mica surface. The Board held that the skilled person would turn to 4-hydroxy-2-nonenal and/or 4-oxo-2-nonenal as alternative stabilisers, and that it was common general knowledge that stabilised structures are advantageous for raising antibodies. The claims were considered obvious and the patent was revoked.

T 0326/22 concerned an anti-CD47 antibody which binds to a discontinuous epitope comprising fourteen specified residues. It can be challenging to raise antibodies to discontinuous epitopes, and so claims attempting to define antibodies by binding to such epitopes routinely face objections of lack of sufficiency. In the present case, the claims were found to be sufficient because the patent describes a complete process for generating and functionally characterising the antibodies, including assays to screen and select antibodies with the claimed properties and methods for structurally assessing the antibody-epitope complex. The Board found there was no reason to doubt that the skilled person could reliably arrive at further antibodies binding to the claimed epitope. The claimed antibody was further defined as preventing CD47 from interacting with SIRPα and not causing a significant level of agglutination of cells after administration. The closest prior art disclosed a full length anti-CD47 antibody for blocking the interaction with SIRPα, but there was no indication that the antibody could bind the claimed discontinuous epitope, and nothing to suggest that targeting said epitope could achieve the effect of reduced cell agglutination. Therefore, the claims were found to be inventive.

Specific/2nd Generation Decisions

T 1515/20 is an appeal from an examining division decision concerning claims to a specific anti-C5 antibody (eculizumab) defined by its heavy and light chain sequences. Taken in isolation, the decision is (arguably) a routine consideration of a claim to a specific antibody per se, albeit with narrower claims than usual because the heavy and light chain sequences define both variable and constant regions. This was necessary because the argument for inventive step relied upon the presence of an unusual engineered heavy chain constant region, which was demonstrated to reduce undesirable immune responses. The Board held that this was not predictable from the art and hence the claims were inventive. However, the Board also decided on sufficiency that the retention of a signal peptide in the light chain sequence did not render the claims insufficient because the signal peptide would not be expected to interfere with target binding. This is interesting, because the lack of disclosure of a mature light chain sequence was more problematic in two related decisions considered in the next section.

Downstream/3^rd Generation Decisions

T 1087/19 concerned the parent patent of the application in T 1515/20. Unlike the divisional above, the claims at issue were medical use claims, which recite a specific pharmaceutical composition of eculizumab (recited by name) for treating paroxysmal nocturnal hemoglobinuria. The Board found the claims to be insufficiently disclosed because the skilled person would not have been able to obtain the mature antibody based on the content of the application as filed. This problem arose because, as noted above, the light chain sequence disclosed in the patent included a signal peptide at the N terminus. Neither the mature light chain sequence nor the cleavage site for removal of the signal peptide were disclosed. A request to correct the error was denied following the EPO's strict two-part test (both the existence of the error and the correction must be objectively recognisable). Other attempts to claim the mature sequence were considered to add matter.

In T 1435/20 the Board considered similar issues in the context of an appeal from a decision of the examining division to reject a further divisional application. The Board reached similar conclusions to those of T 1087/19 in respect of medical use claims that referred to eculizumab by sequence rather than by name. In short, attempts to define the mature light chain sequence were considered to add matter, and claims reciting the immature sequence including the signal peptide were held to be insufficient, in part because there was no evidence of a therapeutic effect for the antibody including this sequence.

T 1345/20 concerned an in vitro method of diagnosing Gaucher disease by immunoassay. The decision is interesting for the Board's consideration of sufficiency with respect to the antibody used in the assay. The patent referred to immunoassays for detecting lyso-Gb1, but the Examples only demonstrated lyso-Gb1 detection using liquid chromatography-mass spectrometry. The Board disagreed with the proprietor that it was routine to generate antibodies against lyso-Gb1 and referred to T 0435/20 (discussed in our 2023 review). This decision found that it is only routine to raise an antibody for an unconventional target if the antigen suitable for raising the antibodies and the process for selecting them are both known. Here, there was no indication of whether lyso-Gb1 is a suitable antigen, or how an antibody with the necessary specificity could be selected, so the generation of antibodies was considered to be an undue burden, and the patent was revoked for lack of sufficiency.

T 0294/20 concerned anti-PD1 and anti-PD-L1 antibodies that reduce the activity or expression of PD1 for treating nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL). The patent proposed that exhaustion of CD8 T cells in the tumour microenvironment in NLPHL can be reversed by PD1 blockade. However, whether T cell exhaustion is relevant for treating NLPHL is not considered in the patent, nor is it derivable from the prior art. The Board held that the skilled person would not have concluded that the effect of PD-1 blockade on exhausted T cells was relevant to the treatment of NLPHL. In line with the principles of G 2/21, the suitability of a product for the claimed therapeutic use must at least be derivable from application, unless this is already known to the skilled person at the priority date. In the absence of suitable experimental evidence, the patent was revoked for lack of sufficiency. T 0737/20 was a related case, concerning anti-PD1, anti-PD-L2 and anti-PD-L2 antibodies for treating a defined set of persistent viral infections. Here, the claims were revoked for lack of inventive step, because the closest prior art demonstrated that PD-L1 blockade can result in the restoration of CD8 T cell function during chronic viral infection.

The most relevant aspect of T 0293/19 concerned an IgM preparation obtainable by a process defined in an earlier claim, and having a particular proteolytic activity. The prior art taught that proteolytic activity of an immunoglobulin preparation should be as low as possible. The Board held that the proteolytic activity defined in claim 1 was not linked to any special technical effect, over and above what is already known in the art. Therefore, it would have been obvious to the skilled person to attempt to reduce the proteolytic activity when looking to design an improved IgM preparation, even though the product could not have been prepared by the methods available in the prior art. The proprietor limited their claims to the method (excluding the preparation per se) in order to overcome the issue.

T 1815/22 was primarily concerned with clarity (Article 84 EPC). The claims in this case were directed to a composition comprising an individual antibody (mepolizumab) in particular % proportions of oxidised variants and deamidated variants. Rejected as unclear by the examining division, at the Board of Appeal the applicant successfully argued that such definitions were not uncommon in the field and that methods of measuring % variants were known, such that it was unnecessary to specify a method in the claims. It is encouraging that the EPO did not apply any unusual "antibody-specific" criteria when assessing this case.

T 1927/22 concerned a pharmaceutical composition comprising an anti-PCSK9 antibody for use in reducing lipoprotein A (Lp(a)) levels in a patient subgroup defined by a particular serum level of Lp(a). The Board accepted that the specified Lp(a) level was sufficient to establish novelty over prior art uses of anti-PCSK9 antibodies. It was then debated whether a statement in the prior art that it was "conceivable" that PCSK9 inhibitors could decrease Lp(a) levels would have led the skilled person to test this hypothesis with a reasonable expectation of success. Although a link between anti-PCSK9 and other lipoproteins such as LDL-R was established, the Board concluded that there was doubt in the art regarding any link between PCSK9 inhibitors and Lp(a) levels. As such there was no reasonable expectation of success and the claims were found inventive.

T 0025/23 was an appeal filed by the proprietor against a decision by the opposition division to revoke the patent for lack of sufficiency. The claims concerned bevacizumab for use in a method of treating a patient diagnosed with one of a group of platinum-resistant cancers. The outcome of the case depended on whether the claimed medical use is sufficiently disclosed based on the data in the patent. The Board found that the skilled person would have regarded the claimed cancers as a single group. This matched the clinical trial protocol in the Examples, which did not attempt to differentiate between cancers. There was nothing to suggest that the results of the trial needed to be stratified according to each cancer – rather the results were considered applicable to the group of cancers as a whole. Therefore, the skilled person would have no reason to doubt that the results reported are applicable to patients diagnosed with each of the claimed cancers. Accordingly, the Board decided that the claims were sufficiently disclosed.

T 0219/22 related to a SerpinF2-binding antibody for use in a method of preventing or treating brain haemorrhage or brain edema resulting from cerebral ischaemia due to thrombotic ischaemic stroke. The closest prior art teaches that lowering the level of circulating SerpinF2 can reduce the risk of intracerebral haemorrhage in treatment of acute ischaemic stroke. The key differences are first that the claim specifies that the treatment is of a human patient, whereas the prior art only concerns rats, and second that the claim refers to an antibody, whereas the prior art uses microplasmin to lower SerpinF2 levels. However, the Board found that the prior art suggests using an antibody as an alternative to microplasmin to lower circulating SerpinF2. Thus, the skilled person seeking alternative means to deplete SerpinF2 would have realised from the prior art that a SerpinF2-neutralising antibody would be effective. The patent was therefore revoked for lack of inventive step.

T 1639/21 related to a combination of an antagonistic anti-PD1 antibody and a mRNA cancer vaccine. The closest prior art concerned improving the effectiveness of a mRNA vaccine by combining it with an anti-CTLA4 antibody. The case therefore hinged on whether the replacement of an anti-CTLA4 antibody with an anti-PD1 antibody was obvious. The Board agreed that there was a synergistic effect when the vaccine is combined with anti-PD1, but they also found that such synergy is to be expected when replacing one checkpoint inhibitor with another, not least since one of the cited documents showed a similar synergistic effect when anti-CTLA4 was combined with a mRNA vaccine. This case serves as a useful reminder that synergy is not a guarantee of inventive step.

T 1103/22 related to an antibody for use in treating pathological thrombosis in a defined patient subgroup, where the antibody was further defined by functional features: (a) it binds Factor XI; (b) in a specific region of Factor XI ("at the active site located in the light chain region"); and (c) it inhibits the activity of Factor XIa in specific a chromogenic assay. The decision turned on feature (b) with the Board finding the patent insufficient. In effect, the Board treated feature (b) as a functional definition of a specific epitope. They held that there was insufficient information in the application as filed or the common general knowledge regarding the epitope to permit the skilled person to identify antibodies binding to it. The Board once again referred to T 0435/20 (see also above and our 2023 review) as establishing that antibodies against some epitopes (in particular non-linear ones) are not routine to make, and noted also the established case law that insufficient disclosure cannot be remedied by post-published evidence.

T 0466/23 concerned specific formulations of adalimumab. Although there was a single prior art document directed to a similar purpose of reducing the number of components in a formulation and improving stability, the Board also found a number of other documents to be plausible starting points. Therefore, the Board assessed inventive step starting from a number of different closest prior art documents. Notwithstanding, the Board found that none of the prior art would have provided the skilled person with a pointer to use the specific concentrations of components required by the claim. Thus, inventive step was acknowledged and the patent was upheld.

T 0810/22 concerned stable formulations of anti-PD1 antibodies, such as pembrolizumab, for use in therapy. The specific liquid formulation as claimed is also known as Keytruda.The closest prior art discussed the therapeutic uses of anti-PD1 antibodies, including pembrolizumab, but not their preparation as a stable formulation suitable for human use. The technical effect of the claimed formulation is the high degree of stability that is necessary for Phase III trials and marketing. As such, the objective technical problem was formulated as the provision of a formulation of pembrolizumab that can be marketed for human therapeutic use. Although other prior art documents provided extensive disclosure of approaches for efficient formulation of biopharmaceuticals, the Board considered that hindsight was required to apply these teachings to pembrolizumab and select the claimed formulation of other suitable formulations. The Board viewed the claimed formulation to be the result of an optimisation process for late clinical development and marketing rather than the outcome of a quick routine search for a formulation with sufficient stability. Thus, the claims were deemed to be inventive over the prior art.

Non-target/General Decisions

T 0168/19 concerned a method of engineering an immunobinder with improved functional properties by introducing specific mutations to the VH/VL regions. The main difference from the closest prior art was the specific positions and nature of the mutations. The objective technical problem was formulated as a method for introducing mutations into an immunobinder which has an increased likelihood of improving functional properties selected from solubility, stability, non-aggregation, expression and refolding yield. However, after reviewing the patent, the Board found that this problem was not solved by the claims, and thus the problem was reformulated to be less ambitious: the provision of alternative mutations for improving the functional properties of immunobinders. The solution was held to be obvious as the prior art already disclosed many of the indicated mutation positions and suggests screening for improved functional properties. Therefore, the skilled person would have arbitrarily chosen any of the possible residues to solve the problem.

T 1664/22 was an appeal against the decision of the examining division to refuse the application. The claims concerned a method of preparing a pH-dependent conditionally active antibody that is able to preferentially deliver a desired level of activity to a treatment location in the body where an aberrant pH exists, with the aim of reducing side effects of antibody therapies. Importantly, the claimed method included two additional selection criteria for arriving at the modified antibodies. Other than these selection criteria, the closest prior art disclosed a similar method and aimed to solve the same problem of reducing side effects while maintaining or increasing therapeutic activity. The objective technical problem was formulated as the provision of a further method of production and screening of conditionally active antibodies. The Board found that comparing the activity of a mutant antibody to the activity of its parent amounts to an obvious screening step whenever mutagenesis is used in order to produce antibodies with increased or decreased activity in given conditions. Thus, it was obvious to add further screening steps to the closest prior art, to arrive at the claimed invention and the application was refused.