In 2023 we published a pair of articles reviewing EPO case law in the field of antibodies, addressing questions of sufficiency of disclosure and inventive step. This article provides an update, a year on.
Arguably the most consequential decision of the past year relates to patent claims which define an antibody by its binding to a discontinuous epitope. Our previous review highlighted the EPO’s approach to such claims as an area to watch, and the question has now been considered by the Boards of Appeal in T 435/20. Interesting decisions on antibody-specific questions have also issued in relation to inventive step and added matter.
Together with the recent update to the EPO Guidelines on antibody inventions, this provides important guidance to all those interested in the field of antibody innovation.
Developments Relating to Sufficiency of Disclosure
As set out in the EPO Guidelines, one of the ways in which the EPO permits patent applicants/proprietors to define a new antibody is by specifying the epitope to which it binds. However, where an antibody is defined by a functional property, such as its epitope, the patent must enable the “skilled person” to reliably generate and identify antibodies having the specified property without undue burden. While it has generally been accepted as routine to produce antibodies binding a defined linear epitope, there has been uncertainty as to the acceptability of defining an antibody as binding a particular discontinuous (conformational) epitope.
This question was considered in T 435/20, relating to a patent for antibodies against IL‑23. IL‑23 is a heterodimeric cytokine comprising the subunits IL‑23p19 and IL‑23p40. The patent in question was directed to an antibody that bound human IL‑23p19 at an epitope comprising residues 82-95 and 133-140. The patent provided a single antibody which bound the specified epitope (“7G10”), and included various functional analyses of this antibody, including determination of its epitope by X-ray crystallography, but its method of production was not described.
The Board of Appeal agreed with the opponents that on the facts of the case, there was an undue burden for the skilled person to raise an antibody binding the specified discontinuous epitope. The particular objections of the Board were that: (i) the patent did not teach what antigen should be used; (ii) the patent did not teach how antibodies raised against IL‑23 could be screened to identify those claimed; and (iii) even if the skilled person was able to identify an appropriate antigen and screening method, there was no evidence that antibodies binding the claimed epitope could be generated reliably enough to guarantee that the skilled person could successfully obtain one. Rather, this appeared to be a matter of chance. The patent was therefore revoked for lack of sufficiency.
Although the underlying patent was revoked, the Board’s decision provides some guidance to applicants as to how to improve their chances of obtaining a European patent for an antibody defined by its binding to a discontinuous epitope (and conversely to opponents, as to the kinds of deficiencies which could lead to such patents being revoked).
Firstly, it is important for applicants to describe how they obtained one or more specific antibodies which bind the claimed epitope, including describing how the antibodies were raised and any screening performed to identify them from a broader pool. Secondly, applicants may need to show that antibodies with the claimed epitope are obtained with high enough frequency that they can be reliably generated.
Further guidance on this topic can be expected when the appeal T 0326/22, relating to EP 2812443, is decided. This patent is directed to antibodies against CD47 which bind a discontinuous epitope comprising 14 specified amino acids, and which prevent CD47 from interacting with SIRPα. The patent was upheld at first instance, and the appeal hearing will take place in August 2024.
The rationale of T 435/20 has also been applied in a different situation in T 1345/20. Here, the claims related to an antibody for an “unusual” non-peptidic target molecule, rather than a non-contiguous epitope. The patent related to the use of Lyso-Gb1 as a biomarker for Gaucher’s disease. Lyso-Gb1 comprises a fatty acid chain with a monosaccharide headgroup:
The patent contained a dependent claim specifying that the detection of the biomarker is performed by immunoassay, though the examples of the patent only demonstrated lyso-Gb1 detection by liquid chromatography-mass spectrometry (LC-MS).
The opponent argued that the patent lacked sufficiency of disclosure because it failed to provide any suitable antibodies for detection of lyso-Gb1, or to show how they could be generated. It was argued that lyso-Gb1 is a challenging molecule to generate highly specific antibodies against since it lacks features commonly recognised as epitopes for recognition by antibodies, particularly polysaccharides and polypeptides.
The proprietor argued that generating antibodies against lyso-Gb1 was no undue burden, since generating antibodies against known targets was considered routine according to established EPO case law. However, the Board cited the conclusion in T 435/20 that this only applied when the skilled person knew (i) which antigens are suitable for raising the desired antibodies and (ii) how a pool of antibodies could be screened to select those claimed. It characterised these conclusions as being “generally applicable to any antigen”. The Board held that there was no indication of whether lyso-Gb1 is a suitable antigen or of how an antibody with the necessary specificity for use in a diagnostic assay (i.e. capable of discriminating between lyso-Gb1 and similar molecules) could be selected. The claim was therefore found not to comply with the requirements for sufficiency.
Conversely, in T 835/21 the Boards applied similar criteria to reach a conclusion that sufficiency could be acknowledged, even in the absence of a reproducible teaching of any specific antibody in the examples. This case confirms the possibility of obtaining broad, functional antibody claims at the EPO, even with relatively little supporting data, where routine techniques can be used to obtain antibodies binding to the antigen/epitope and where the claimed functions can be easily screened for.
The patent in this case was directed to an antibody which binds the “propellor 3” domain of the human protein LRP6, and inhibits Wnt3- and Wnt3a-specific signalling. Notably, the patent did not provide an enabling disclosure of any antibody falling within the claim: while two antibodies having the claimed properties were mentioned in the examples, no sequence or hybridoma deposit information was provided for either. Nor was their generation described. Nonetheless, the Board considered the patent to be sufficiently disclosed. Generation of antibodies against the propellor 3 domain of LRP6 was considered routine, as it is a large domain of 300 amino acids and the patent showed that antibodies binding it could be identified by screening antibodies raised against LRP6 against mutants with the propellor 3 domain deleted. Assays to screen the effect of antibodies on Wnt signalling were known in the art. Accordingly, the Board found that it would not have been an undue burden for the skilled person to identify antibodies falling within the claim.
Developments Relating to Inventive Step
The case law at the EPO relating to inventive step for antibodies is relatively well established: as set out above, development of antibodies against known targets is generally considered routine. Therefore, for an inventive step to be acknowledged, an antibody against a known target must generally have a surprising technical effect, generally an unpredictable improvement over prior art antibodies in terms of e.g. biological/therapeutic activity, affinity for its target or pharmacokinetics.
These principles were tested in T 1669/19, relating to an antibody against human PCSK9. The antibody was defined by its full variable domains, and was argued by the proprietor to be inventive over prior art anti-PCSK9 antibodies due to its higher affinity for its target and higher potency in terms of inhibiting PCSK9-mediated LDL uptake.
The opponent argued that even if the claimed antibody was superior to those of the prior art, methods for improving the affinity and IC50 of antibodies were well known in the art (e.g. affinity maturation). According to the opponent, the skilled person would have had a reasonable expectation of being able to improve the prior art antibodies using such routine techniques.
Had the Board accepted these arguments, this would have significantly limited the possibility of obtaining European patents for antibodies against known targets. Thankfully for antibody innovators, the Board rejected these arguments. Although the Board acknowledged that routine techniques were available for optimisation of antibody affinity/potency, the degree of improvement seen in the claimed antibody compared to those of the prior art was considered unpredictable, and the antibody was considered inventive.
Developments Relating to Added Matter
Unusually, this year saw a pair of cases addressing an antibody-specific issue relating to added subject matter, namely, does the disclosure of the full-length sequence of an antibody chain implicitly disclose the sequence of the mature chain?
This question was considered in T 1435/20 and T 1515/20, which related to two cases in the same family, both of which had been refused by the Examining Division. The cases related to an antibody against complement protein C5. The original application as filed contained an unfortunate error, in that while the mature heavy chain sequence was correctly disclosed, the disclosed light chain sequence included the leader peptide. The mature light chain sequence was not disclosed.
The applicant sought to address this by referring in the claims to “a heavy chain of SEQ ID NO: 2 and a light chain of residues 23-236 of SEQ ID NO: 4”, i.e. excluding the 22 amino acid light chain leader sequence. There was no explicit basis for specifying residues 23-236 of SEQ ID NO: 4, but the applicant argued that it was the correction of an obvious error. At the EPO, correction of obvious errors does not require basis in the application as filed. However, the test for correction is a strict one and requires it to be obvious both (i) that there is an error, and (ii) what the correction should be.
The Board did not consider the inclusion of the immature light chain sequence an obvious error: while it was common general knowledge that antibody chains are encoded with leader sequences, on the facts of the case they did not consider that it would have been obvious that the light chain sequence presented in the application comprised the leader sequence. As a result, the Board found that it was not obvious that an error had been made, and no correction was possible.
The take home message, of course, is that it is important to ensure that the sequences are defined correctly in the application at the time of filing, including removal of any sequences that will not be present in the mature antibody.
Summary
The last year has seen some important decisions relating to the patenting of antibodies at the EPO, including decisions on sufficiency which have helped to illustrate when obtaining (further) antibodies within the scope of a claim may or may not be considered routine. We anticipate more important decisions over the coming year and will look forward to providing further updates!
In the meantime, please do get in touch if you need any assistance in respect of an antibody patent prosecution or opposition.
Ed handles patent work across the life sciences/biotech field, primarily in relation to therapeutics. He is very experienced in drafting patent applications and prosecuting them around the world, and also manages opposition and appeal work at the European Patent Office.
Email: edward.couchman@mewburn.com
