Patenting antibodies – the latest developments from the EPO Boards of Appeal:  Part 2 – inventive step

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

In the first part of this two-part series of articles, we reviewed EPO case law relating to how antibodies can and should be defined in European patent claims, with a focus on sufficiency of disclosure (enablement). In this article we turn to EPO case law relating to the inventive step of antibodies.

The general approach of the EPO to inventive step of antibodies is summarised in the Guidelines for Examination. These set out that a new antibody against a known target is usually considered obvious, unless a surprising technical effect is shown for the antibody or unless there was no reasonable expectation of success in obtaining antibodies with the required properties (G-II-5.6.2).

The Board of Appeal decisions that we have reviewed are broadly consistent with this general approach. They provide useful guidance for applicants on what kind of technical effect may be necessary for an antibody to be considered inventive in various situations.

Inventive step of antibodies defined by their epitope

For an antibody defined by the epitope to which it binds, it is usually required both that the claimed antibody has an unexpected activity, and that the unexpected activity is brought about by binding to the specified epitope. These principles have been applied consistently in the cases we have reviewed.

Where binding of an antibody to a particular epitope is associated with a new and unexpected biological effect, claims to an antibody which recognises that epitope are likely to be considered inventive.

This is exemplified by T 1964/18, which related to an antibody binding NKG2A at the same epitope as a specified reference antibody (Z270) and not binding NKG2C or NKG2E. NKG2A is an NK cell inhibitory receptor while NKG2C and NKG2E are NK cell activatory receptors. The reference antibody Z270 specifically bound NKG2A and blocked its activation, without blocking NKG2C or NKG2E, thereby promoting activation of NK cells in the presence of an NKG2 receptor ligand. This specificity was argued to be a property of its epitope, which was unique to NKG2A.

The antibody considered as the closest prior art blocked all of NKG2A, NKG2C and NKG2E, and the patentee argued that no antibody which specifically blocked NKG2A was previously known. The Board held that there was no pointer in the prior art to generating an antibody specific for NKG2A, nor any direction as to how this could be achieved, and that therefore the claims were inventive.

Conversely, T 1872/16 and T 813/19 both provide examples of cases where the technical effect associated with the epitope bound was considered predictable from the prior art and thus not capable of providing inventive step.

In T 1872/16, the claims were directed to an antibody defined by two characteristics: (i) binding an epitope on hIL-13 corresponding to residues 97-108; and (ii) the ability to modulate the interaction between hIL-13 and its receptor (hIL-13R).

The prior art disclosed an antibody capable of binding hIL‑13 and modulating the interaction of hIL‑13 and hIL‑13R. Residues 103 and 107 of hIL‑13 were identified as essential for binding of the antibody to hIL‑13, along with certain other hIL‑13 residues outside of positions 97-108. The claimed antibody had no technical effect not achieved by the prior art antibody, but the patentee argued that the epitope recognised by the prior art antibody differed to that specified in the patent, in that it contained residues outside of positions 97-108 of hIL‑13. They also argued that the prior art antibody had particularly desirable properties (IL-13 neutralisation) which meant that the skilled person would not have modified the prior art to generate an anti-hIL13 antibody which recognised a different epitope.

The Board disagreed, pointing out that the provision of an alternative antibody was the problem with which the skilled person was faced. The Board found that it would have been obvious for the skilled person to generate an antibody which bound the claimed epitope, since residues 103 and 107 were identified by the prior art as critical to the binding of a neutralising anti-hIL-13 antibody to its antigen.

Similar principles underly the decision in T 813/19. This case related to a patent directed to an antibody against activin A (and use of such an antibody for treatment of cachexia), defined as binding a cysteine knot region of human activin A and inhibiting binding of activin A to its receptor. The prior art disclosed that activin A signalling through its receptor was responsible for cachexia symptoms in mice, while the region of activin A recited in the claims was known to include amino acid residues responsible for binding of activin A to its receptor. The Board concluded that it would have been obvious for the skilled person to produce an antibody that blocked binding of activin A to its receptor as this would have been expected to be effective in cachexia treatment, and that for this purpose it would have been obvious to target the region of activin A specified in the claims because this section included residues involved in binding of activin A to its receptor.

These two decisions show that if binding of an antibody to a particular epitope can be expected to provide the biological effect described in the patent, then claims to an antibody which binds that epitope will generally be considered obvious.

A further decision, T 2258/15, illustrates the importance of demonstrating that the technical effect relied on for inventive step is causally related to the epitope binding.

This decision related to a patent directed to an antibody against human protein tyrosine kinase 7 (PTK7). The claimed antibody was defined as binding the same epitope on PTK7 as one of two specified reference antibodies (referred to as 4D5 and 7C8).

The patentee argued that the two reference antibodies bound epitopes within a single “epitope group”, and that antibodies which bound the specified epitopes were superior to prior art antibodies against PTK7 due to improved internalisation properties. However, the Board found that there was no evidence that the reference antibodies 4D5 and 7C8 bound related epitopes (indeed, the epitope of 4D5 was not determined in the patent at all), and that there was no evidence in the patent that binding to the epitopes recognised by the reference antibodies was associated with improved internalisation. The Board therefore disregarded the internalisation properties and held that the claimed antibodies were not associated with any new technical effect. As a result, they were considered obvious alternatives to those of the prior art.

Inventive step of antibodies defined by sequence

The EPO does not generally apply the principle of “structural non-obviousness” (that an antibody with a new sequence is inventive, because its specific sequence was not obvious). The approach of the EPO is that once an antibody having a particular function is known, then other antibodies having the same function can generally be obtained without difficulty, and so the specific sequence of a new antibody having a known function cannot confer inventive step. Therefore even for antibodies defined by sequence, arguments for patentability tend to revolve around the new functions or properties of the antibody. The decisions below provide some guidance on what kind of technical effect may be necessary for an antibody to be considered inventive.

The pair of decisions T 1911/17 and T 505/19 show that an effect or difference relied on should be technically meaningful and non-arbitrary.

In T 1911/17, the patent in question was found to lack inventive step because the difference relied on by the proprietor was not considered to amount to a technical advantage. The claims were directed to an antibody against human brain natriuretic peptide (BNP), defined as being produced by hybridoma CNCM 1-3746, or as containing the CDRs of the antibody produced by the specified hybridoma. The antibody produced by the specified hybridoma was referred to as 20G7.

The closest prior art disclosed another anti-BNP antibody referred to as 24C5. Post-filing data demonstrated that 24C5 bound the same epitope as the claimed antibody 20G7, and so the Board held that only the structures of the antibodies were different.

The patentee argued that 20G7 was superior to 24C5 based on sandwich ELISA data in the patent that showed 20G7 to be superior to 24C5 in detecting BNP when used in combination with a specific capture antibody. However, the opponent demonstrated that when using a different capture antibody, 20G7 and 24C5 were similarly capable of detecting BNP. The Board held that in the context of sandwich ELISA, different detection antibodies are inevitably compatible with different capture antibodies. Therefore, the board held that the claimed antibody was not superior to the antibody of the prior art: rather, the two antibodies merely required different assay conditions to be used for BNP detection in a sandwich ELISA. On this basis, the Board found the claims to lack inventive step.

On the other hand, in T 505/19 the applicant was able to persuade the Board that the differences relied on were advantageous and non-arbitrary, and in doing so was able to overcome a previous negative decision by the Examining Division. This case related to the refusal of an application directed to an anti-VCAM-1 antibody for in vivo imaging use in humans. Compared to a prior art anti-VCAM-1 antibody, the claimed antibody was non-neutralising, had a lower affinity and had lower on/off rates. The Examining Division had considered these differences arbitrary and refused the application for lack of inventive step. However, the applicant was able to persuade the Board that these characteristics were advantageous for imaging purposes, and the Board therefore concluded that the claims were inventive.

One question which sometimes comes up during examination at the EPO is whether the applicant is required to show that the technical effect relied on for inventive step is not inherently shared by an antibody of the prior art. EPO examiners sometimes insist that this is necessary, even in cases where the antibody of the claim is defined by its sequence and where there is no reason to believe that the prior art antibody would share the same property. The recent decision T 2127/16 appears to be a potentially useful decision for pushing back against objections of this kind.

T 2127/16 related to an antibody against amyloid‑β-derived diffusible ligand (ADDL). ADDLs are soluble oligomers of amyloid‑β, the levels of which are elevated in Alzheimer’s disease patients, and which the applicant argued to be the most relevant amyloid‑β species in Alzheimer’s disease. The claimed antibody (‘3B3’) was defined by its CDR sequences, and the applicant argued that this antibody was advantageous over prior art antibodies to amyloid‑β because it specifically recognised ADDLs and inhibited binding of ADDLs to primary hippocampal neurons.

The closest prior art was considered to be an earlier antibody (known as 7A2). 7A2 was not structurally characterised or made available by hybridoma deposit, but was taught to have the property of specifically recognising dimeric and trimeric amyloid‑β. There was no evidence either way as to whether the prior art 7A2 antibody had the property of inhibiting binding of ADDLs to neurons (i.e., the property relied on for inventive step). However, the application under consideration showed that relatively few antibodies which specifically bound ADDLs were able to inhibit their binding to hippocampal neurons. On this basis the Board decided there was no reason to assume that the prior art antibodies had such activity. Inventive step was acknowledged.

This decision very helpfully indicates that where an antibody is defined by its sequence, it may be enough to show that the advantageous technical feature is uncommon without proving that it is not provided by an antibody of the prior art. This may be particularly the case when the specific antibody of the prior art was not structurally characterised or made available to the public in any other way.

Inventive step of a humanised antibody

Another decision which may be helpful to applicants is T 1171/18, relating to a humanised anti-biotin antibody.

In this case, the antibody was a humanised version of a known murine antibody. The antibody was claimed based on its full VH and VL sequences, which included (in addition to grafting of the murine CDRs into human framework regions) two mutations in the framework regions relative to the basic human sequences and two mutations in VHCDR2 relative to the murine sequence. The mutations were shown to maintain binding affinity of the humanised antibody at nearly the level of the original murine antibody (only a 2.5-fold reduction in affinity was seen for the humanised antibody relative to the original murine antibody).

The Examining Division at first instance had refused the patent application. The applicant had argued that the level of affinity retention by the claimed antibody was high, as the prior art taught that substantially higher losses in affinity could be expected when humanising an antibody (as high as 30-fold). They further argued that the mutations introduced into the antibody sequences to retain the high level of affinity were not predictable, particularly as no humanised anti-biotin antibody was known in the art. The Examining Division held that humanisation of murine antibodies was routine, and that the skilled person would have expected to be able to obtain a humanised antibody without significant affinity loss. However, the Board agreed with the applicant’s arguments, and overturned the decision of the Examining Division, allowing the application.

This case may be helpful for applicants seeking to obtain patents for humanised antibodies, since it recognises that retaining the affinity of the original antibody is not routine and that even if some small amount of affinity is lost, this may nonetheless represent an unexpected technical effect justifying a finding of inventive step.

Inventive step of a bispecific antibody

A recent decision that stands out as having a seemingly unexpected outcome in view of the EPO Guidelines is T 2589/16. This case related to a bispecific antibody against CD3 and a tumour antigen. The CD3 binding domain was structurally defined and shown to cross-react with human and various New World monkey CD3 proteins. A prior art monospecific antibody (known as SP34) was known which displayed the same cross-reactivity, and which was shown to bind the same epitope as the claimed CD3 binding domain. No advantage relative to SP34 was demonstrated for the CD3 binding domain claimed, but nonetheless the Board found the claimed bispecific antibody inventive.

The Board held that the objective technical problem solved by the invention was the provision of a bispecific CD3- and tumour-binding antibody which binds both human and New World monkey CD3. Faced with this problem, the Board concluded that the skilled person would have generated a bispecific antibody comprising the prior art antibody SP34, but would have had no motivation to modify SP34 to generate an alternative antibody. Notably, the Board held that the skilled person would have been concerned that an antibody against the same target with a different sequence would not have shared the cross-reactivity of SP34.

The Board in this case thus found an antibody with the same CD3-binding properties as the prior art to be inventive, seemingly on the grounds that it would have been unpredictable whether a further antibody against the target would have the same desirable properties. This decision seems out of line with standard EPO practice relating to antibodies against known targets. Possibly, the Board was influenced by the fact that the claims were directed to a bispecific antibody, which provided an additional step away from the monospecific prior art antibody. It will be interesting to see whether this approach is generally followed in relation to multispecific antibodies and other antibody constructs.


The decisions issued over the last 3 years largely reinforce the EPO’s existing approach to assessing the inventive step of antibodies, but provide further guidance on what technical effects may be considered by the Boards to confer an inventive step. It seems that the Boards’ approach to claims to antibody constructs may be less predictable, but as more cases relating to antibody constructs come before the Boards we would hope and expect more guidance as to how such claims will be assessed. We will continue to watch this field and look forward to updating you on future developments in the law.


About the authors

This article was co-authored by Edward Couchman (lead), Alex Galbraith, and Rebecca Tollervey.

Blank Author Circle


Edward Couchman

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




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.edom to Operate projects and in performing due diligence, which have led to the successful completion of high value transactions and investment rounds.