Review Of EPO Antibody Decisions In 2025

Welcome to the latest edition of J A Kemp's annual review of EPO Board of Appeal decisions concerning antibody inventions. In this update, we examine decisions issued between 31 January 2025 and 31 December 2025. As in previous years, most decisions handed down during this period align with established EPO examination practice for antibody-related inventions, as reflected in the Guidelines for Examination (G-II, 6 – discussed further in our advanced guide). That said, the case law continues to evolve, and the decisions from the past year reveal several noteworthy developments and emerging trends that merit closer attention.

The first trend concerns post-filed data and its increasing use by parties on both sides of any dispute. Almost three years have passed since the Enlarged Board in G 2/21 confirmed that such data may be submitted as evidence of a technical effect relied upon for inventive step. At the time, this was heralded as a positive development for patent proprietors. It is certainly true that recent years have seen such data deployed to reinforce arguments in favour of inventive step and/or sufficiency. However, it is also clear that the opposite applies, in that opponents can submit post-filed data as evidence that an effect relied upon by a proprietor is not (or not fully) substantiated. See, for example, T 2552/22 discussed further below. We expect both patent proprietors and opponents to continue to use more, and more sophisticated, post-filed evidence to support their respective cases as we head into 2026.

The second trend concerns the availability of broad claims at the EPO. The EPO has historically been willing to allow broad "anti-target" language in antibody claims, but this year patent proprietors have achieved little success when arguing for such claims at the Boards of Appeal. The reasons for failure are usually specific to the unique facts of each case. However, following the first trend above, we see opponents attacking such broad claims by deploying post-filed data to show that an alleged technical effect is not enabled across the full scope. We expect this trend to continue in 2026, in parallel with the first trend, given that broader claims are naturally more vulnerable to such attacks than narrower claims. Nonetheless, it is still fair to say that the EPO adopts a more lenient approach to broad "anti-target" antibody claims compared to the US PTO. Broad claims remain at least theoretically possible in Europe. By contrast, in the USA claims will typically only satisfy enablement requirements if they are limited to antibodies defined by full CDR or variable region sequences.

Finally, in a third trend, we have begun to see a growing number of cases before the Boards of Appeal relating to chimeric antigen receptor (CAR) molecules and associated T cell therapies. See, for example, T 1555/23 discussed below. The antigen-binding domains of CARs are typically based on known or optimised antigen-binding domains from known antibodies. As such, we expect the EPO to examine CARs by applying the same principles and criteria that have been developed in antibody jurisprudence at the EPO, although this is not yet stated explicitly in the Guidelines.

Looking to the future, we were interested to see that the revisions to the EPO Guidelines (due to enter into force on 1 April 2026) have removed the current negative presumption in G-II, 6.2 that a novel antibody for a known target lacks an inventive step "unless a surprising technical effect is shown by the application". Instead, the new Guidelines will state in positive terms that inventive step will be acknowledged for such an antibody if a surprising technical effect is shown in the application. Although this adjustment is subtle, it is perhaps encouraging for patent proprietors that the EPO has chosen to re-frame its position in terms of what will be allowed rather than what will be refused.

Below we provide a statistical analysis of the 2025 decisions and discuss some of the individual decisions in more detail.

Statistics

We identified 35 decisions published between 31 January 2025 and 31 December 2025 which refer to the term "antibody". We excluded 20 decisions from further consideration due to not being directly concerned with antibody inventions. Of the remaining 15 decisions, all relate to appeals from first instance Opposition Division decisions. Board 3.3.04 retained its position as the most prolific "antibody Board" issuing 9 of the 15 decisions (Board 3.3.04 issued 15 of 22 decisions considered in our 2024 review). 4 decisions were issued by Board 3.3.08 (4/22 in 2024), 2 by Board 3.3.10 (1/22 in 2024) and none by Board 3.3.07 (2/22 in 2024). We discerned no obvious differences in practice between those Boards, with the possible exception of some comments by Board 3.3.10 in T 0684/23, discussed further below.

Repeating the trend from previous years, a significant majority of the 15 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 generally consistent across all of the relevant decisions, again with the possible exception of T 0684/23, discussed further below.

As with previous years, we consider there to be four general categories of antibody 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, formulations 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 or optimisation.

The largest category in 2025 (8 decisions – 5 maintained, 3 revoked) is "Downstream/3rd Generation" inventions. This follows a trend that we anticipated in previous years. The number of decisions in this category reflects the growing maturity of the antibody field, including the emergence of biosimilars, variants, and related products such as antibody-drug conjugates (ADCs). Producers of biosimilars are likely to be interested in the (usually) later-expiring patents in this category.

The next largest category (4 decisions – 2 maintained, 2 revoked) is "Non-target/General" inventions. This is an interesting development, and likely reflects the growing interest in antibody-adjacent technology, such as strategies to optimise antibody manufacture and stability, and the "conjugate" part of antibody-drug conjugates.

The remaining 3 decisions are all in the "Broad/1st Generation" category (3 decisions – 3 revoked), with no decisions this year in the "Specific/2nd Generation" category. To the extent that there may be a trend here, it is that broad claims are likely to be both more vulnerable to attack and more problematic for third parties than narrower, specific claims. 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. This is likely because the narrow scope of the claims is less likely to be subject to opposition. Nonetheless, this is the first year in which we have observed no decisions in this category.

Detailed discussion of individual decisions by category

Broad "Anti-Target"/1st Generation Decision

The claims in T 0709/23 were directed to any antibody that specifically binds to feline IL-31 for use in treating a pruritic (itchy) or allergic condition in cats, provided the antibody inhibits IL-31 in a cell-based assay. According to EPO case law, medical use inventions meet the requirements for sufficiency if (1) the substance or composition can be made by the skilled person without undue burden and (2) it is credible that the substance or composition actually achieves the claimed therapeutic effect. There must be evidence for a claimed therapeutic use; a mere assertion is not enough. The patent was revoked because the opponent was able to point to post-filed evidence (generated in a later patent application filed by the proprietor) which showed that at least one anti-IL31 antibody that achieved inhibition in the recited cell-based assay did not achieve the required therapeutic effect. The Board found that there was nothing in the patent to help the skilled person overcome this failure. Accordingly they concluded that the skilled person could not reliably obtain antibodies for treating pruritus in cats by following the teaching of the patent, which was revoked for lack sufficiency.

The Board in T 2552/22 came to a similar conclusion to T 0709/23, albeit in the context of inventive step rather than sufficiency. The decision focused on inventive step because, unlike the medical use in T 0709/23, the effect at issue was not recited in the claims of T 2552/22. The latter claims concerned any anti-LAG-3 antibody which binds to a specific epitope sequence on human LAG-3. The closest prior art was a known antibody which binds to an adjacent epitope on the same loop of LAG-3. The proprietor argued that binding to the newly-identified epitope of the claims led to better T cell stimulation, i.e. that it provides an improved antibody relative to the prior art. The opponent submitted post-filed data showing that an antibody within the claims did not possess the technical effect. It was shown to be no better than the prior art antibody at stimulating T cells. The Board agreed with the opponent and held that the purported technical effect could not be achieved over the whole scope of the claim. As such, this technical effect was ignored in the evaluation of inventive step and the objective technical problem was formulated as the provision of merely an alternative anti-LAG-3 antibody. The Board concluded that designing an alternative antibody that recognised an epitope on the same immunogenic region of LAG-3 was obvious and routine. Accordingly the patent was revoked.

T 0684/23 concerned any anti-symmetrical dimethylarginine (anti-SDMA) antibody for detecting free SDMA, a biomarker of renal function, in a sample. The patent also included claims directed to immunoassay devices incorporating such an antibody. The main question was whether the broadly-defined "anti-target" antibody was inventive over the closest prior art that disclosed SDMA as a suitable biomarker in high-performance liquid chromatography (HPLC) assays. The technical problem solved by the claimed antibody was defined as providing a simpler, more cost-effective method to detect SDMA. The Board considered this to be obvious, since it required merely replacing one technique (HPLC) with a well-established, well-known, alternative technique (immunoassay). This is consistent with the EPO's approach to diagnostic technology in general.

However, perhaps reflecting Board 3.3.10's usual focus on chemical subject matter, they also made a number of statements that appear inconsistent with established antibody practice at the EPO. In particular, although both parties considered it to be important to discuss whether or not it would be difficult to raise an antibody to SDMA with the required properties, the Board explicitly dismissed the relevance of this debate. They noted that: "the claimed antibody needs to be inventive on its own: it cannot be rendered inventive by means of any arguable difficulty in obtaining it." This appears to contradict the following assertion in the current Guidelines and the April 2026 revisions: "antibodies can be inventive if the application overcomes technical difficulties in generating or manufacturing them". This long-established principle is based on a line of case law originating with T 0735/00. It will be interesting to see if any parties now rely on the (apparently divergent) reasoning applied in T 0684/23 to seek a referral to the Enlarged Board in cases where inventive step may turn on difficulties in the manufacture of an antibody. Such cases are though relatively rare and the EPO itself evidently saw no reason to update the 2026 Guidelines in light of T 0684/23.

Downstream/3rd Generation Decisions

The Board in T 0358/22 provided an interesting comment on the test for implicit disclosure for purposes of assessing novelty. The claims at issue concerned a pharmaceutical formulation of the anti-HER2 antibody pertuzumab which includes a disulfide reduced variant (known as an "acidic" variant). The opponent argued this lacked novelty over a prior art document which also disclosed pertuzumab formulations and standard recombinant methods of producing the same in CHO cells. Specifically, the opponent argued that an electropherogram in the prior art implicitly showed that the acidic variant of pertuzumab was present in the formulation, and that this was an inevitable result of the general production method disclosed. The Board first considered the case law, which sets out that a product is considered to be disclosed if it is the inevitable result of a process defined in the prior art. However, said process would need to be adequately and credibly described, including starting substances and reaction conditions. In the present case, the electropherogram in the prior art was not a direct and unambiguous disclosure of the claimed invention. In fact, the prior art document actually stated there was "no evidence of significant product fragments". The Board held that the methods described in the prior art were too general to inevitably lead to the specific variant as claimed, and thus found the claims to be novel.

T 1253/23 represents the latest development in a long-running dispute between Biogen and Polpharma concerning Biogen's therapeutic antibody natalizumab, which is authorised for use in multiple sclerosis (MS). Polpharma have developed a natalizumab biosimilar and so are understandably interested in Biogen's downstream patent estate. The claims of most interest in this case concerned natalizumab for use in treating MS in patients selected based on testing "negative or transiently positive" for a clinically significant immune response to natalizumab, defined by reference to a threshold level of 500 ng/mL anti-natalizumab antibodies detected at two separate timepoints. The closest prior art was a press release disclosing the use of natalizumab in treating multiple sclerosis. The press release contained a "Safety" section which disclosed that patients were tested at intervals for anti-natalizumab antibodies and that such antibodies were detected in approximately 10% of patients at least once during treatment, with 6% of patients remaining persistently positive, i.e. implying the presence also of a "transiently positive" population of 4%. The Board considered it to be obvious from this disclosure to obtain improvement in treatment of multiple sclerosis patients with natalizumab by only treating the negative or transiently positive patients. The specific threshold level of ~500 ng/mL anti-natalizumab antibodies was considered to be obvious based on the typical sensitivity of immunoassays. This reinforces the principle that selecting a numerical threshold for a known biomarker will normally be regarded as routine optimisation and thus not inventive at the EPO, unless there is evidence of a surprising technical effect associated with the threshold.

T 1505/23 concerned an anti-SIRPα antibody capable of blocking the interaction between SIRPα and CD47, for use in the treatment of a hepatocellular carcinoma (HCC) or a melanoma. Inventive step was assessed for each therapeutic use in turn. For melanoma, the closest prior art disclosed the same general principle of treating melanoma by modulating the SIRPα-CD47 axis. Using an anti-SIRPα antibody to block the interaction between SIRPα and CD47 was regarded as obvious solution suggested in the prior art itself, and so the Board found no unexpected advantage associated with the claimed antibodies beyond that implied by the prior art. For HCC, the closest prior art was an abstract which disclosed that anti-SIRPα antibodies could enhance macrophage phagocytosis of HCC cells. The abstract also stated that this was a promising strategy for the treatment of HCC. The Board found that this gave the skilled person a reasonable expectation of success in treating HCC patients with such antibodies. The patentee attempted to argue that their invention represented an improvement because the antibodies exploited a new mechanism of action, but this was dismissed as the specific mechanism was not recited in the claim.

T 2008/22 concerned an in vitro method for diagnosing and monitoring Gaucher's disease by measuring levels of free lyso-Gb1 in a sample over time. As such, this decision is not directly related to antibodies per se but the method encompassed immunoassays as a means of detection of free lyso-Gb1, and interestingly the Board overturned the decision of the opposition division, so this case merits a brief discussion. Since immunoassays are encompassed by the claimed method, it was agreed that for such assays to be sufficiently disclosed, suitable antibodies must be described in the application as filed. However, neither the patent nor the prior art describes how to obtain suitable antibodies, capable of distinguishing free lyso-Gb1 at low concentrations from similar target molecules which might also be present in biological samples. Without suitable epitopes for antibody-recognition required for design of antibodies having sufficiently high specificity and affinity, lyso-Gb1 is a challenging and unconventional target – neither the prior art nor the patent disclosed that lyso-Gb1 is a suitable antigen for generating antibodies. The Board confirmed the accepted principle that whilst it is routine to raise and screen antibodies to a known target, this only applies if the skilled person knows from the disclosure in patent or common general knowledge: (i) which antigens are suitable for raising antibodies having the desired properties and (ii) which screening process should be used to select antibodies without undue burden. This test for assessing enablement of antibodies against unconventional targets is set out in T 0435/20 (discussed in our 2023 review). In the absence of available anti-lyso-Gb1 antibodies, and any means of obtaining them without an undue burden, the Board concluded that the development of such antibodies would require significant research efforts going beyond routine experimentation. As such, the claims were found to be insufficiently disclosed and the patent was revoked.

T 1618/23 concerned methods and compositions for assaying a vitamin D moiety in a sample, specifically using a homogeneous immunoassay with at least two antibodies having defined binding properties. One of the antibodies has a specific binding affinity for the vitamin D moiety whereas the other antibody has a specific binding affinity for the resulting complex. A key issue was how to interpret the term "homogeneous immunoassay" in the context of the claimed methods. The Board decided that "homogeneous" means the assay is performed in a single phase and does not require physical separation (e.g. washing the analyte) before detection. This interpretation was important because it affected whether prior art methods were considered comparable. Novelty was acknowledged because the prior art did not disclose at least two antibodies according to the claimed methods. For inventive step, the closest prior art was a document disclosing a lateral flow immunoassay for detecting vitamin D. However, the Board found that the lateral flow assay in the closest prior art involved phase separation through capillary flow before detection of the assay signal. Accordingly, the closest prior art did not represent a homogeneous assay as construed above, and thus did not render the claimed methods obvious. The Board acknowledged that the claims involved an inventive step and the patent was maintained as granted.

T 1474/23 concerned a method of detecting the presence of neutralising antibodies against the therapeutic antibody sarilumab, in a sample, using a capture reagent comprising biotinylated sarilumab and a detecting reagent comprising labelled soluble IL-6 receptor alpha (IL-6Rα). During opposition, the opponent argued that broader claims in an earlier request lacked an inventive step, but did not provide any new or substantive submissions on appeal specifically addressing why narrower claims to sarilumab/ biotinylated sarilumab/labelled soluble IL-6Rα would also be obvious. The Board found that there was no convincing case that the skilled person would have been led by the prior art to the specific narrower embodiment. This is a useful reminder that an opponent must show that the specific features as claimed would have been obvious to a skilled person in light of prior art; it is not sufficient to repeat arguments used for broader claims. This case also illustrates that specific reagents and labelled constructs (especially when narrowly defined) can be decisive for inventive step if not taught or suggested in the prior art.

T 0781/23 concerned antibodies that neutralise the inhibitory activity of human NKG2A for use in treating cancer in two groups of patients: those who experience disease progression during PD-1 blockade, and those who experience disease progression following PD-1 blockade. The main issue on appeal was whether patent made it credible that neutralising the inhibitory activity of NKG2A is suitable for treating patients who experience disease progression despite PD-1 blockade. The Board reviewed the data in the patent and concluded that NKG2A expression in tumour-infiltrating lymphocytes (TILs) was involved in immune evasion, in tumours resistant to PD-1/PD-L1 blockade. From the common general knowledge, it was known that inhibitors of NKG2A could increase the activity of lymphocytes towards tumour cells and be used to control tumour growth, so the Board found that it was credible that a patient having increased NKG2A expression following PD-1 blockade would benefit from a treatment involving an antibody that neutralises the inhibitory activity of NKG2A. Importantly, the therapeutic rationale relied on a correlation between PD-1 blockade and increased NKG2A expression, but it was not decisive whether increased NKG2A expression was the cause or consequence of resistance to PD-1 blockade. Interestingly, the Board also noted that the presence of some fully responsive animals in the model experiments did not negate the relevance of the model in demonstrating the credibility of the claimed therapy in treating unresponsive patients. Since the appellant failed to raise serious doubts supported by verifiable facts, the Board acknowledged that the claimed invention was sufficiently disclosed.

T 1555/23 is an example of a decision concerning chimeric antigen receptor-modified T cells (CAR T cells). The claims related to any CAR T cell for use as a first-line therapy in a method of treating cancer, wherein the method involves monitoring cytokine levels and administering a cytokine inhibitor as a second-line therapy to manage toxicity. The main issue was whether it was credible that a cytokine inhibitor (specifically an IL-6 inhibitor) can manage toxicities associated with CAR T cell treatment while maintaining a therapeutic effect. Based on the available data, the Board concluded that it was credible. Interestingly, the claims did not specify the target or the sequence of the CAR and (amongst other objections) one opponent sought to argue that this resulted in a lack of sufficiency, because a CAR T specific for one antigen would not be capable of treating a cancer expressing a different antigen. The Board did not agree, finding that the skilled person would be able to select a CAR with appropriate specificity for the tumour to be treated. This is encouraging, in that it is further evidence that the EPO is not dogmatic when considering the scope of claims to antibody-derived molecules such as CARs. As with antibodies per se, the EPO does not insist that the claims to such molecules recite a particular specificity and/or sequence, where these characteristics are not relevant to the inventive contribution of the applicant.

Non-target/General Decisions

The claims in T 1462/22 concerned an screening process for identifying antibodies with pharmacokinetically advantageous properties. The screening method involved finding antibodies with a specific range of pH-dependent binding affinities. At opposition, the opponents argued that the claims lacked sufficient disclosure because the skilled person would not be able to carry out the invention in view of the disclosure in the application as filed. The opposition division agreed with these arguments and found the claims insufficient. In the resulting appeal, the Board noted that claim was for a method and the steps thereof were clear and enabled. The Board specifically noted that opponents' arguments were not relevant because they concerned features which were not included in the claim (e.g. improved antibody recycling). As such, the Board overturned the decision at opposition, finding that the claims were sufficiently disclosed, and the case was remitted for assessment of novelty and inventive step.

T 1913/21 related to methods of preventing and eliminating the formulation of (unwanted) trisulfide bonds in antibodies during their manufacture. Specifically, the claim at issue referred to the use of an inhibitor of cysteine degradation for reducing the formation of trisulfide bonds in proteins, wherein inhibitor is selected from pyruvate [...], wherein the use comprises: culturing cells expressing said proteins in the presence of an effective amount of the inhibitor of cysteine degradation, whereby trisulfide linkage formulation in said proteins is reduced. The decision mainly focuses on claim construction and novelty. A key issue was whether the "use" constituted a second non-medical use. At opposition, the opposition division agreed with the patentee and construed claim 1 as a second non-medical use claim, so the technical effect of the inhibitor in reducing trisulfide linkage formation was considered a limiting functional feature of the claim. On appeal, the Board disagreed, finding that claim should be regarded as a process claim for production of proteins, not a second non-medical use claim. The Board emphasised that where a claim in effect recites physical steps of a process for producing a product, the mere inclusion of an intended outcome/effect does not transform it into a "non-medical use" claim under G 2/88; rather, it must be treated as a regular process claim. Thus, the stated aim ("reducing the formation of trisulfide bonds") did not contribute a functional technical limitation that could render a known compound use novel under G 2/88. Accordingly, the claim was found to lack novelty because the prior art disclosed the same physical steps leading to the same product, irrespective of whether trisulfide reduction was specifically disclosed.

T 0089/23 concerned a method of producing a recombinant antibody in a cell culture medium comprising hypotaurine. The main issue here was the admittance of the main request into the appeal proceedings and there was a lengthy discussion concluding in the Board's acknowledgement that the main request claims prima facie addressed the key objections in a way that justified their admission. The Board then found the opponent's inventive step attack based on document D5 to be unpersuasive, because D5 was not prior art in view of the priority claim. Since the opponent had not relied on any other prior art published before the priority date, the inventive step attack did not succeed and the patent was maintained.

T 1174/22 is of interest because it is an example of a decision concerning antibody-drug conjugates (ADCs). This case is in the "non-target" category, because the claims were directed to any "cell-binding agent" of which any "antibody" was a preferred subtype. The invention was alleged to reside in a particular type of sulfoxide-containing linker used to connect the agent/antibody to a drug, with the definition of the latter including a specific maytansinoid (an established class of chemotherapeutic). The prior art disclosed a seemingly identical maytansinoid+linker joined to an antibody. The patent proprietor's argument for novelty required a sulfoxide group in the prior art molecule to be defined as part of the maytansinoid rather than as part of the linker. The Board dismissed this argument, noting that the title of the prior art document was "Novel maytansinoid derivatives with sulfoxide linker", and the patent was revoked for lack of novelty. Although this decision is very narrow in scope, it illustrates that the EPO is not applying any special criteria to ADCs. Patent proprietors should be free to argue that any component (or any combination of components) in an ADC gives rise to novelty and inventive step.

J A Kemp LLP acts for clients in the USA, Europe and globally, advising on UK and European patent practice and representing them before the European Patent Office, UKIPO and Unified Patent Court. We have in-depth expertise in a wide range of technologies, including Biotech and Life Sciences, Pharmaceuticals, Software and IT, Chemistry, Electronics and Engineering and many others. See our website to find out more.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.