The Barrier Around Antibody Inventions at the European Patent Office

Acknowledgments

Tamaris Bucher is an employee of Novartis Pharma AG. I owe a huge debt of gratitude to Harvey Adams of Carpmaels & Ransford LLP, without whose input, advice and review of the manuscript this paper would not have been possible.

The author is also very grateful to the following people for their input, review and feedback:
Cameron Marshall from Carpmaels & Ransford LLP
And the following current or former colleagues:
John Blankenship, Meabh Brennan, Regis Cebe, Ian Hiscock, Karen Larbig, Laura Leitner, Hans Masselink, Ioannis Papapetridis, Isabelle Schubert, Andrea Strang, Sarah Thompson
The views expressed in the article do not necessarily reflect the opinions of Novartis or Carpmaels & Ransford LLP.

PART THREE

Part 3 of this paper will dissect the problem-solution approach as it is currently applied to antibodies in accordance with the Guidelines. This will not only elucidate how an incorrect focus has arisen, but also demonstrate that, if the objective technical problem is correctly formulated, the predictability and obviousness of an antibody defined by its CDR or VH/VL sequences can be judged more equitably in alignment with the chemical case law more generally. Part 3 will also address the challenges and difficulties in the engineering of commercial antibody products, particularly those designed for pharmaceutical use in vivo. Finally, the future of the antibody field, both with and without the artificial barrier from the Guidelines in place, will be examined.

Problem-solution approach applied to antibodies

At present, EPO examiners would seem to apply a very specialised type of problem-solution approach to any claim that is directed to an antibody per se, even including those in which the CDR or VH/VL sequences are specified. In the normal application of the problem-solution approach, the Guidelines in Part G.VII.5.1 dictate that the closest prior art (“CPA”) should first be established, this being a document which is directed to the same or a similar purpose. It is further noted that “[i]n practice, the closest prior art is generally that which corresponds to a similar use and requires the minimum of structural and functional modifications to arrive at the claimed invention.”

Once the CPA is established for an antibody claim, examiners are instructed by the Guidelines (Part G.II.6.2, 2024 version) to apply the negative premise that “a claim defining a novel antibody binding to a known antigen does not involve an inventive step [unless one of the limited number of exceptions applies]”. The previous edition of the Guidelines, in Section G.II.5.6.2, contained the phrase that “[t]he subject-matter of a claim defining a novel, further antibody binding to a known antigen does not involve an inventive step …”. However, the word “further” was deleted from the 2024 version. Thus, it would now appear that the negative premise concerning the prima facie obviousness of antibodies can even apply when no specific prior art antibody to the known antigen is disclosed in the state of the art. Here, it must again be noted that there is no general requirement in either the EPC or the case law which mandates the presence of a “surprising technical effect” for a particular claimed subject matter to be considered to be inventive. However, EPO examiners typically turn to assessing whether a “surprising technical effect” is present. In addition, and as already discussed in Part 2 of this paper, it is often assumed that it is the technical effect per se, rather than the provision of the means by which the effect is achieved, which must be considered to be unexpected.

Often, examiners look for improvements in a particular property as compared to the closest prior art, thereby requiring the submission of comparative data. The objective technical problem is subsequently formulated as the provision of an antibody to the target X that exhibits the relevant improvement. If the improvement is present over the full scope of the claim, an inventive step can then be acknowledged. On the other hand, if no improvement is identified, the typical practice is to re-define the objective technical problem as the provision of an alternative antibody to the target or for use in the same purpose. A finding of obviousness usually follows.

Here, it must be questioned whether the use of the term “antibody” in this formulation of the objective technical problem has the inadvertent effect of causing EPO examiners to overlook the variability in the CDRs. Does this term implicitly suggest that both the claimed and prior art molecules have such a similar structure that the antibody of the claims should be considered to be nothing more than an obvious minor variant of the prior art molecule, with an entirely predictable function?

For low molecular weight chemical compounds, this issue is not encountered, at least not as frequently. This is perhaps for the reason that the structures are claimed as graphical representations: examiners look to the depicted structures when assessing inventive step and, therefore, they more easily take account of their differences. In contrast, for antibody inventions, where the arrangements of atoms are not specified in the claims but are instead represented by the seemingly generic terms of “SEQ ID No. Z” or “SEQ ID No. ZZ”, the differences between the respective three-dimensional structures are markedly less apparent. As such, the structural differences from the prior art compounds are easy to overlook, and the solution itself is then underappreciated. This is further compounded by the instruction in the Guidelines which discourages the assessment of predictability at the level of the amino acid sequence: “[t]he fact that the structure of an antibody, i.e. its amino acid sequence, is not predictable is not a reason for considering the antibody as non-obvious”.

So how then can it be better recognised that claims to antibodies defined by specific CDR or VH/VL sequences have different structures in the region most involved in binding, i.e in the paratope, thereby to ensure that the lack of predictability in the structure-function relationship is taken into account?

As one option, the objective technical problem could perhaps be formulated as the provision of a specific alternative molecular structure (or sequence) for use in achieving the functional activity Y’ {e.g. a mode of action such as an antagonistic or agonistic activity, driving the internalisation of antigen X or achieving a certain effect on a cell that expresses antigen X}. This recitation of a specific alternative molecular structure (or sequence) may better reflect the fact that the claim is reciting precisely defined sequence information that embodies the solution and that the sequences represented by the SEQ ID Nos. have very different structures.

This adjusted formulation of the objective technical problem is also more consistent with the approach that is used for low molecular weight compounds. In that field, the objective technical problem can often be formulated as the provision of an alternative compound with the desirable functional activity (e.g. inhibition or agonism). Such a formulation does not contain elements of the solution and nor does it imply the presence of a similar structure – which is arguably what happens when using the term “antibody”.

The final step in the problem-solution approach, which follows the formulation of the objective technical problem, is to consider the question of whether or not the claimed solution, starting from the closest prior art embodiment and in light of the objective technical problem, would have been obvious to the skilled person from the state of the art. Here it should be considered whether or not, starting from the closest prior art (e.g. from a particular prior art antibody possessing specific CDRs which bind to the target X), and in light of the objective technical problem (more correctly formulated as the provision of an alternative structure/sequence for achieving the desirable activity Y), it would have been obvious to the skilled person to modify the known structure (at least in its CDR sequences) in such a manner as not only to arrive at the structure that is claimed, but to do so with a reasonable expectation of success, i.e. of achieving the desirable activity.

Otherwise put, the question to be asked is whether the particular claimed antibody with its specific amino acid sequences would have been made by the skilled person (i) by following the prior art teaching; and (ii) with a reasonable expectation of achieving the desirable activity. It is not the question of whether that particular antibody could have been made using unlimited time and resources and irrespective of whether there was an a priori expectation that its specific sequences would solve the underlying problem.

Overall, therefore, a possible approach is that if the binding sequences of a claimed antibody (its CDRs or VH/VL sequences) are dissimilar to the corresponding sequences of the closest prior art antibody that achieves the same function, then this structural dissimilarity should play a significant role in the analysis of whether the specifically claimed antibody is obvious. On the other hand, if the CDRs and VH/VL sequences as well as the overall structure of the claimed antibody are extremely similar to those of the prior art antibody, then it could be argued that as very similar structures might be expected to behave similarly, the claimed antibody should be considered to be obvious. Any such conclusion would, however, depend on the facts of the case. For example, if a further technical effect or improvement was exhibited by the claimed antibody resulting from these small differences in the amino acid sequences, and it was not predictable from the state of the art that these differences would lead to that effect or improvement, then this would be unexpected and it could lead to the finding of the presence of an inventive step. Such an approach would bring the problem-solution approach as applied to antibodies better into line with the approach that is used for low molecular weight chemical compounds, thereby removing the artificial barrier between these two fields of chemistry.

Routine methods analysis

In order to justify a divergence in the inventive step analysis of antibodies on the one hand, and low molecular weight chemical compounds on the other, the Guidelines assert that “[a]rriving at alternative antibodies exclusively by applying techniques known in the art is considered to be obvious to the skilled person.” This is tied to the notion that, according to the EPO, ‘it is routine to make antibodies’. The following discussion will challenge these suppositions.

First, the focus on the generation method needs to be examined, since in a claim to an antibody that is defined by its sequences, it is not the method of making the antibody that is being claimed. Stewart et al.Stewart M., Kent L., Smith A., and Bassinder E. “The Special Inventive Step Standard for Antibodies” EPI Information 2/2011, pages 72-76, see page 72 have noted that “[f]or compounds that are not antibodies, the EPO makes structural comparisons with prior art molecules and does not consider whatever methods might have been used to achieve such compounds.” They also note that “[i]t is unclear when a particular method is considered by the EPO to be routine such that any antibodies discovered by such a method are considered obvious.”Stewart M. et al., op. cit., page 75. Rather astonishingly, it would appear that the Guidelines have now set the standard that, as soon as a method is simply known in the art, the use of that technique renders the resulting structures obvious (at least until one of the exceptions is demonstrated).

Are the methods that are used to develop and engineer effective antibodies sufficiently different from those used for chemical compounds to justify this diverging practice? The view has sometimes been opined that, when producing antibodies, “nature does all the work for you”. This was addressed by Ingham and Smyth with their astute comment that this particular view “departs from any meaningful technical reality.”Ingham S., and Smyth D. “Routine rejection: Is the EPO’s approach to antibody and polymorph claims correct, balanced and justified?” Journal of Intellectual Property Law & Practice, Volume 8, Issue 2, February 2013, pp 154-164, see p158- 159. Ingham and Smyth also commented that the methods used to screen and produce low molecular weight compounds and large molecular weight biological molecules have much in common: “[w]hen a biological target is known, is it really such a different exercise to screen against a compound library to identify a hit, develop a structure-activity relationship (SAR) and arrive at a lead therapeutic molecule, which will likely result in a patent, than to develop a therapeutic antibody similarly engineered through a structure-activity evaluation, which likely will not?” Id., page 158. They answer that question on page 159: “In principle, both fields of endeavour have routine features and features which are individual to the specific case, and in our view it is not possible to categorize the SAR development of a novel therapeutic engineered antibody as any more predictable, than that of small molecule compounds. In both cases, the SAR cannot be determined without experimental investigation, and it cannot be predicted that a certain structure will have a certain function. It does not seem correct that the assumed principles should be so different in each case.”

So how has this situation arisen that there is such a mismatch between views from industry and the standards being applied by the EPO? Why are structurally defined antibody claims being assessed according to the method of generation, when low molecular weight chemical claims are not? Why are all antibodies assumed to be made by routine techniques, and worse why are all antibodies now presumed to be obvious as soon as their production technique is known in the art, unless proven otherwise?

Both the ‘method analysis’ and the prejudicial view that the underlying methods are routine - the so-called ‘routine methods analysis’ as noted by the earlier authors and discussed above – would seem to have arisen from the early case law on deposited hybridomas. As pointed out in Parts 1 and 2 of this paper, however, those early hybridoma cases did not disclose chemical structural information at the level of the amino acid sequence and the assessment of obviousness instead focused on whether or not it was obvious to make any (monoclonal) antibodies: an analysis at the level of the class. Furthermore, since the Köhler-Milstein hybridoma method had been known for some time, the use of that method to produce a monoclonal antibody was considered to be routine in those early decisions.For example, see T 735/00, Reason 26: “In 1989, the priority year of the patent in suit which is thirteen years after the technology to produce monoclonal antibodies had been developed, the preparation of monoclonal antibodies was a matter of routine experiment…The case law in this field acknowledges inventive step if and when there is evidence that a claimed monoclonal antibody prepared by routine methods shows unexpected properties (cf decision T 645/02 of 16 July 2003).”

Here it is submitted that, it is inappropriate to apply this early hybridoma case law and mindset to a claim in which more detailed structural information is provided. The analysis should be commensurate with the solution that is provided, and therefore, a structural approach should be used in accordance with the principles in the CLBA I.D.9.9.2, and not a ‘methods analysis’. Moreover, allowing pronouncements about early methods of production to permeate the assessment of any and all claims to an antibody, seems to have resulted in an unfounded prejudice that all antibody generation work is routine. Consequently, every antibody is considered prima facie obvious by default, including those which have been recombinantly produced and subsequently engineered. Therefore, the more complex problems that are solved when producing the sophisticated antibody products that are produced by today’s technology are severely underestimated in the inventive step analysis.

While commercially relevant diagnostic and therapeutic antibodies may be founded on materials originally deriving from nature, they are typically engineered by scientists through the use of inventive skill to satisfy complex requirements in which the optimisation of one criterion may negatively impact another. Such engineering typically involves sequences in the variable regions and, accordingly, a claim to a defined set of CDRs or VH/VL sequences does not represent a mere arbitrary selection from a process of immunisation. This is true for a diagnostic antibody for in vitro use (which must fulfil e.g. specificity, sensitivity, and stability criteria); it is especially valid for antibodies used in vivo (i.e. as a therapeutic or in vivo diagnostic).

This particular view, that any specific set of CDRs or VH/VL sequences is an arbitrary selection from a host of equally viable alternatives obtainable by routine techniques, may once again be due to what is implied by the term “antibody”, i.e. one of a class of binding molecules having a significant portion of their overall structures in common. On one end of the scale, this class encompasses antibodies merely having a very low affinity for a target, but unable to fulfil any further functional requirements. On the other it encompasses a small proportion of molecules, that have been laboriously selected for their advantageous behaviour and/or highly engineered to fulfil a combination of complex requirements. Yet, the Guidelines have attached a judgement to the entire class: any and all antibodies are prima facie non-inventive.

It could be argued that the technical challenges in developing an antibody for diagnostic or therapeutic use are, perhaps to some extent, recognised in the creation of the two specific exceptions to the negative premise in the Guidelines that antibodies are not inventive. These are the exceptions that (a) there were technical difficulties in generating or manufacturing the claimed antibody; and (b) there was no reasonable expectation of success of obtaining any antibody with the desirable properties. However, these important factors, which are inherent to the development of commercially relevant antibodies, should not merely be recognised as specialised exceptions and therefore interpreted narrowly. For a claim to an antibody defined by its specific CDRs or VH/VL sequences, it should not be necessary for an applicant to show multiple failures in their attempts to produce antibodies which exhibit the required properties. Moreover, a lack of reasonable expectation of success should not solely be attached to technical properties or improvements that are unexpected per se. Rather, the lack of predictability in finding or producing antibodies with meaningful affinity and biological activity, particularly those suitable for diagnostic or therapeutic use, should be taken into account, without the underlying properties being novel and surprising per se.

If the principles set out in the CLBA I.D.9.9.2: Structural Similarity are applied to sequence defined antibody claims, then the predictability of the claimed structure for achieving the desired activity would need to be assessed. Consideration of the extent of the structural differences from the closest prior art embodiment – and of whether a claimed antibody which incorporates those differences is a predictable solution to the objective technical problem of achieving that activity – would then have the effect of separating ‘the wheat from the chaff’ with the same level of rigour as is currently applied to low molecular weight compounds. There would be no need for applicant to disprove that the methods that were used to make the antibody were known or routine, and nor would there be an overriding (and it is submitted, legally unjustified) negative premise that any antibody is prima facie not inventive.

This will be discussed in more detail below.

Humanised antibody derived products

Humanised antibody products which are used in vivo as diagnostics and therapeutics are engineered molecules: donor CDR sequences from a non-human species are combined with human antibody framework regions to lower the immunogenicity profile. The affinity of the donor sequence can be negatively impacted or binding completely lost, and so additional mutations may need to be introduced, including in the CDRs, to develop a suitable VH/VL pair.

It should be reiterated here that the formulation of the objective technical problem in respect of such claims should be commensurate with what is actually being provided, i.e. if a more ambitious technical problem than merely providing a binding molecule is being solved, then the formulation of the objective technical problem – and the expectation of success that is inversely related to the complexity of that problem – should be adjusted accordingly. This was recognised in the decisions T 67/11See T 67/11 Reason 24. and T 1171/18, which involved the humanisation of sequences, while maintaining an acceptable affinity profile. It equally applies when other technical properties are adjusted by engineering, even if the techniques that are used to make those adjustments are already known in the art. In other words, if the CDR or VH/VL sequences that are specified in an antibody claim have resulted from targeted engineering, or even from multiple rounds of mutation and screening under different conditions, it is certainly not the case that “nature [has done] all the work for you”. Moreover, since it is not predictable a priori which particular sequences will work – the creation or identification of specific amino acid sequences which exhibit the desired properties should be considered to involve an inventive discovery.

Pharmaceutical products: Formulation of the technical problem

For a specific antibody which not only has a binding affinity in vitro but is suitable for administration to a human patient as a therapeutic molecule, it must surely be acknowledged that there are many additional requirements which need to be satisfied. Such products must be optimised with regard to multiple properties such as at least some of selectivity for the target antigen (to minimise off-target effects), species cross-reactivity (to facilitate a transition from preclinical animal testing to human patients), immunogenicity profile in preclinical tests, stability in solution, solubility, viscosity, aggregation tendency, purity (especially the absence of host cell proteins that may be co-purified with the antibody), absence of undesirable post-translational modifications such as proteolysis, and ADCC and/or CDC activity. All of this must be done while retaining an acceptable level of affinity and biological activity (e.g. antagonistic or agonistic activity against the target antigen X, effect on a cell that expresses X etc). The fact that the individual properties may be recognised as being desirable – or that known techniques may be used in this optimisation – does not negate the fact that attaining a suitable balance is a priori unpredictable.

In this scenario, the objective technical problem should perhaps be formulated as the provision of “a pharmaceutically acceptable antibody for achieving the desirable activity Y” or, in line with the above discussion, as the provision of “a pharmaceutically acceptable structure/sequence for achieving the desirable activity Y”. Since it cannot be predicted a priori which particular combination of amino acids will afford a binding structure that possesses multiple necessary pharmaceutical properties, a claim to a particular antibody that is useful as a pharmaceutical, defined by its CDRs or VH/VL sequences that are dissimilar to those in the art, should in turn be considered to be a non-obvious solution to that objective technical problem.

As noted above, the work of developing a pharmaceutically acceptable antibody is far from routine, even in the situation in which the individual engineering methods are known in the art. A recent overview of the field by Ministro et al. entitled “Therapeutic Antibody Engineering and Selection Strategies” notes in its conclusion that “[d]espite the moderate success, the common technologies for antibody discovery are time-consuming and imply complex methods.”Ministro J. et al. “Therapeutic Antibody Engineering and Selection Strategies” Chapter included in the Adv Biochem Eng Biotechnol. Book series (ABE, volume 171) 2020; pages 55-86, Section 2.3.

Such complexity is reflected by the variety of techniques for identifying binding sequences that has been developed in the art, this for the reason that there is no ‘one-stop shop’ which will always produce a binder (let alone a binder exhibiting the required activity). For example, the list of techniques that the skilled person has to choose from includes inter alia phage display, yeast display, mammalian cell surface display, ribosome display, bacterial display, mRNA display, DNA display, immunisation of rodents, rabbits, camelids or transgenic animals, as well as B-cell based screenings. Each of these techniques has its advantages and disadvantages, and at the outset of the development project, it is certainly not clear which method will be successful: there are seldom any pointers within the state of the art as to e.g. which category of library should be used for a given antigen. By way of example, Ministro et al. note in relation to phage display that “in most of the cases, antigens are not presented in their native conformations, which decreases the likelihood of selecting successful leads for therapeutic uses.”Ministro J. et al., op. cit. section 2.2.1. Also, the technique “often requires reformatting to produce soluble and well-expressed antibodies with properties compatible with efficient manufacturing.” In another example, the use of animal immunisation for eliciting an antibody is complicated by the fact it may be effective in one case, but not in another, depending on the immunological response of the species that is used.Further examples of why a suitable binder may not be produced include the fact that a naïve antibody library may have had those clones with affinity to self-antigens deleted from the repertoire. Additionally, an immune library derived from B cells of immunised donors to a particular target antigen inherently comes with an “unpredictability of the immune response to the immunized antigen.” Ministro J. et al., op. cit. sections 2.1.1 and 2.1.2. The schedule of immunisations – and the vehicle or adjuvant that is used to provide the antigen – are also confounding factors.

As noted above, it is relevant that an antibody for pharmaceutical use must satisfy a certain activity profile. As such, an antibody may need to bind in a specific way, at a particular epitope, or to a particular conformation of the target antigen before the desired functional effect is observed. Selective binding to a particular epitope may also be necessary to prevent off-target side effects e.g. to avoid binding to closely related family members. However, it is often not known beforehand which epitope or conformation will achieve the desired profile, in which case, the targeting of the antibody production method to specific regions of the antigen is difficult.

Finally, any commercial antibody must also fulfil a number of specific biophysical criteria such as conformational, colloidal, chemical, and storage stabilities. The attainment of such properties by simply isolating random antibodies from synthetic libraries, or by producing hybridomas from randomly selected B cells following animal immunisation, is unlikely: extensive optimisation/engineering is usually required.

The high technical requirements that must be met for pharmaceutical antibodies have as a consequence that the generation of specific sequences which fulfil these requirements is a time-consuming, complex process, not least for the reason that the optimisation of one criterion, such as affinity, may negatively impact another, such as stability, and vice versa.

Unfortunately, however, these difficulties in the production of effective pharmaceutical antibodies are often overlooked, most notably for the reason that while positive results are reported in the literature for a given antibody development project, the hundreds to thousands of molecules that were created, screened, characterised, but ultimately discarded as not fulfilling the required characteristics is seldom, if ever, publicised. Most of the promising antibodies that enter the optimisation cycle do not successfully exit it. As such, the statement in the Guidelines in relation to any and all antibodies – that “[a]rriving at alternative antibodies exclusively by applying techniques known in the art is considered to be obvious to the skilled person” – is unjustified. Overall, therefore, a claim to a pharmaceutically useful antibody, defined by its specific CDRs or VH/VL sequences, should not simply be treated as an arbitrary selection from a host of equally viable alternatives obtainable by routine techniques. Recognition should be given to the complexity involved in optimising a combination of characteristics in parallel via multiple processes to generate that antibody and, moreover, to acknowledge the fact that those sequences provide a solution to the objective technical problem that is a priori unpredictable.

It might be asked whether in silico models and deep learning methods have made all the results of antibody engineering techniques predictable. This is certainly not the case. The paper by Ruffolo et al. notes in its conclusion that “[d]espite considerable improvements by deep learning methods for general protein complex prediction, prediction of antibody-antigen binding remains a challenge. Even the recent AlphaFold-Multimer model, which can accurately predict the interactions of many proteins, is still unable to predict how or whether an antibody will bind to a given antigen.”Ruffolo, J.A., Chu, L. S., Mahajan, S.P. et al. “Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies.” Nat Commun 14, 2389 (2023). In addition, the long list of criteria besides antigen binding that a therapeutic or diagnostic antibody must necessarily fulfil in addition to simply binding in order to be useful, clearly serves to dramatically reduce the predictability of identifying a structure which satisfies these multiple criteria.

Future Landscape with the artificial barrier removed

Currently the standards applied by the EPO to antibody inventions on the one hand and low molecular weight compounds on the other, are inconsistent, separated by an artificial barrier in the Guidelines. The arbitrary nature of the barrier between these fields, which singles out antibodies for special attention, can perhaps be illustrated by the following thought experiment: consider a claim in the following terms, which defines a specific therapeutic antibody using language more aligned with low molecular weight compounds, i.e. by referencing a (pharmaceutical) product, omitting the term “antibody”, and mentioning only structural elements while removing the technical effects to the formulation of the objective technical problem. Would this type of claim, by aligning its language with the wording of claims to low molecular weight compounds, be considered more favourably than the traditional type of “antibody” claim?

A (pharmaceutical) product comprising a complex comprising a first polypeptide of SEQ ID. X1 and a second polypeptide of SEQ ID. X2.

[X1: VH or HC, X2: VL or LC]

If any such claim might be treated more favourably than a claim to an “antibody”, then a problem must exist. Thus, whether or not the word “antibody” is included in the claim should not direct the examiner along a different, and more restrictive pathway for assessing inventive step.

It may be asked whether, if the EPO were to remove the artificial barrier between antibodies on the one hand, and low molecular weight compounds on the other, thereby to recognise the unpredictability of the effect of structural changes on functional properties in relation to CDRs and/or VH/VL sequences, then would this lead to the grant of additional patents that would be problematic for the antibody field and/or detrimental to the public more generally? This can hardly be the case because any such patents would have relatively narrow claims.

Of note in this context is the established approach of the USPTO in assessing structural factors and recognising the non-obviousness of molecules defined by specific CDRs which are dissimilar to the known CDR sequences that were used for the same or similar purpose. This approach has been applied for many years, without controversy. In fact, it is so uncontroversial that there seems to be no need to dedicate a specific section in the MPEP to the obviousness of antibodies – let alone to single out this particular field of chemistry as being prima facie non-inventive.

Removal of the artificial barrier would therefore harmonise the EPO practice with this important jurisdiction, and also with the way in which other fields of chemical technology are treated. It is important for a patent system to have – and be seen to have – consistent standards across all areas of technology.

Future landscape with the artificial barrier in place

The current EPO Guidelines reluctantly admit the recognition of an inventive step for a particular antibody only by means of one of a limited number of narrowly interpreted exceptions, i.e. if there is a “surprising technical effect”, no reasonable expectation of success of obtaining antibodies having the required properties, technical difficulties in generating or manufacturing such antibodies, or a novel type of functional antibody format. Antibody patents are being granted on the basis of these limited exceptions. Accordingly, an analysis should be made as to whether these exceptions – as currently applied by the EPO – are already sufficient to serve the antibody field going forward.

As discussed in Part 2 of this paper, some EPO examiners presently consider that the existence of a technical effect must itself be surprising in order for the first exception to apply. As such, identifying or developing a particular antibody that is e.g. selective for a given target may not be considered to embody an inventive step, irrespective of the divergence of its CDRs or VH/VL sequences from those of the prior art: it will be argued that antibodies are known to be able to be selective. Here it should be noted that the functional properties that an antibody possesses essentially relate back to a pair of core properties, i.e. binding (including e.g. affinity and selectivity) and integrity (e.g. serum half-life, stability in vitro). Therefore, as the antibody field advances, arguments in support of an inventive step are less likely to be centred on qualitative improvements (i.e. on the discovery of a new property), but instead on quantitative improvements (a change in the degree of one or more known – and therefore unsurprising – properties). Therefore, will this particular exception, as it is currently being interpreted, become less available over time? If the possibility of obtaining an improvement in such a property is considered unsurprising, then antibodies defined by their specific amino acid sequences will be increasingly difficult to patent.

Notable in this regard was the removal from the 2024 Guidelines of the mention of improved affinity within the list of “surprising technical effects” to be considered (this property having been included initially when the Guidelines to antibodies were introduced in 2021). Thus, does the EPO now consider that engineered improvements in binding affinity are the result of routine development? The current list of properties in the antibody section of the Guidelines is, of course, non-exhaustive. It might therefore be argued that the removal of affinity does not in turn mean that affinity will not be considered. However, the removal from the Guidelines of this and other properties from the list of “surprising technical effects” is certainly a disquieting change.The properties of “an improved affinity”, “a reduced toxicity”, and “an unexpected species cross-reactivity” were removed from the list in the 2024 Edition of the Guidelines, G-II-5.6.2.

Finally, the presumption of obviousness and the narrowly interpreted requirement for a “surprising technical effect” is now being applied by some EPO examiners to any claim that recites an immunological sequence, e.g. in CAR-T and TCR products_._ If this trend continues, then the negative premise set forth in the Guidelines, i.e. that novel antibodies to known targets do not involve an inventive step, will be over applied to new creative fields involving sequence-based information. It must therefore be a concern for the future of biologics more generally.

Comments