The Barrier Around Antibody Inventions at the European Patent Office

Acknowledgments

Tamaris Bucher is 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.

The Landscape Today

An antibody defined by its sequence in a patent application discloses chemical structural information to the public. As such, it is important to understand the landscape for patenting new chemical compounds. Examiners at the EPO assess the inventive step of claims to chemical compounds by analysing structural similarity as set out in the Case Law of the Boards of Appeal of the EPO (CLBA) I.D.9.9.2: ‘Structural Similarity’. This approach is sometimes referred to as the ‘doctrine of structural non-obviousness’ although this terminology is not itself mentioned in this section of the CLBA.

In practice, this assessment begins with an examination of the closeness of the claimed structure to a structure in the prior art which is indicated to have the same or similar usefulness for solving the underlying technical problem. This is followed by an assessment of whether the skilled person would have known, from their common general knowledge or from specific disclosure(s) at the effective filing date, whether the structural differences from the prior art structure were so small that they would have no essential bearing on the property that is important for solving the technical problem.Case Law of the BoA I.D.9.9.2 (July 2022, Tenth edition): ‘Structural similarity’: see for example T 852/91 In other words, the similarity to a structure having the same or similar usefulness is first assessed, and then the predictability of the impact of the structural difference(s) on that usefulness is assessed. Although the analysis is under the heading in the CLBA I.D.9.9.2: as ‘Structural similarity’ the essence of the analysis is the predictability of the structure to function relationship.

Antibodies are chemical compounds and, therefore, it would be legitimate to expect that for an antibody defined by specific amino acid sequence information, i.e. by a chemical structure, these well-established principles should apply.

However, with the changes to the Guidelines for Examination in the EPO (GL) in 2021, the EPO introduced a new section which outlined a specialised standard for assessing the inventive step of antibody claims, even those defined by sequence information. The EPO has a commendable policy of continual reassessment of the Guidelines with cycles of feedback and revision. As such, it is time for an appraisal of whether this new section is serving the needs of the EPO, applicants, and the public in general.

It will be demonstrated below that this section in the Guidelines concerning the inventive step of antibodies has created an artificial barrier between claims to low molecular weight chemical compounds and claims to compounds that include the word ‘antibody’. This mindset had previously been practised in relation to some types of antibody claims, but the phrasing in the Guidelines formalises and extends this artificial barrier between these fields of chemistry. This inconsistent and unclear approach to the different fields, and the purported basis for it, must certainly be questioned. Patents have a crucial role in incentivising the research and development of biologics, a field of tremendous public health and economic significance. Therefore, the following dissection of the Guidelines and case law is necessarily rigorous, to determine whether the standards now being imposed have stayed aligned with the innovative technology being developed, and to question why antibodies, in particular, are being singled out to meet a higher patentability hurdle than other chemical fields. There is ultimately a constructive intent in the analysis, to bring consistency within the field of chemical inventions and to identify a solution as to how this can be achieved.

EPO Guidelines Surrounding the Inventive Step of Antibodies

The new section of the Guidelines concerning the inventive step analysis of antibodies is Part G.II.6.2 (2024 version):

6.2 Inventive step of antibodies

The subject-matter of a claim defining a novel antibody binding to a known antigen does not involve an inventive step unless a surprising technical effect is shown in the application or unless there was no reasonable expectation of success of obtaining antibodies having the required properties (cf G‑VII, 13). Examples of surprising technical effects include an unexpected improvement over prior-art antibodies in one or more properties, such as therapeutic activity, stability or immunogenicity, or an unexpected property not exhibited by prior-art antibodies. (Emphasis in bold and italics added)

At the end of Part G-II.6.2, two further exceptions are identified which could permit the recognition of an inventive step:

Nevertheless, antibodies can be inventive if the application overcomes technical difficulties in generating or manufacturing the claimed antibodies. A novel type of functional antibody format may also be considered inventive.

The choice of the wording in the opening paragraph (highlighted in bold above), through its negative and unequivocal phrasing (“does not”, rather than “may not”), established for the first time a negative premise that an antibody is prima facie non-inventive subject matter. This may not have been intentional, but the choice of wording means that the applicant must overturn this negative premise by showing that the claimed antibodies can meet the specified exceptions ( namely: a surprising technical effect, no reasonable expectation of success of obtaining antibodies having the required properties, difficulties in generating or manufacturing the claimed antibody, or a novel type of functional antibody format).

Notably, Art. 56 EPC does not start with a presumption that needs to be overturned – i.e. that claimed subject matter prima facie does not have an inventive step. Nor is this presumption applied in other technical fields, whether chemical or non-chemical.

Therefore, it is important to consider whether this negative premise is justified in the case law for antibody inventions. This is especially critical because the Guidelines, while they do not constitute legal provisions themselves,“It should be noted also that the Guidelines do not constitute legal provisions.”  Guidelines, General Part.3: General Remarks. have the effect of creating de facto standards: examiners are expected to act in accordance with them.“As a general rule, parties may expect the EPO to act in accordance with the Guidelines until such time as they – or the relevant legal provisions – are amended.” See Guidelines, General Part.3: General Remarks.

The Guidelines suggest an explanation for the negative premise and cite as their support two Technical Board of Appeal decisions:

Inventive step is not acknowledged solely on the basis that an antibody is structurally different from the prior-art antibodies. Arriving at alternative antibodies exclusively by applying techniques known in the art is considered to be obvious to the skilled person. The 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 (see T 605/14, section 24; T 187/04, section 11). (Italics added)

Thus, they seek to justify the negative premise with the further assertions that:

1. making alternative antibodies using known techniques is obvious (highlighted by the italicised statement above) and
2. unpredictability of the structure (on the level of the amino acid sequence) is not enough to acknowledge inventiveness.

The first assertion highlights a fundamental misconception about the challenges in engineering antibody sequences; it shall be addressed in more detail in Part 3 of this paper. The second assertion, which is responsible for excluding from the assessment of antibodies the structure-function approach that is applied to low molecular weight compounds will be examined in detail below.

Foundation of the Barrier

Firstly, the authority of the two cited Technical Boards of Appeal decisions to support the negative premise must be questioned. There has been no Enlarged Board of Appeal decision which specifically addresses the inventive step analysis of antibodies. Moreover, the compilation of case law in the CLBA contains no Technical Board of Appeal decision which includes the statement in the Guidelines that “a novel antibody binding to a known antigen does not involve an inventive step [unless the specified exceptions apply]”No cases reciting this premise were found upon a keyword search of the Case Law of the BoA (July 2022, Tenth Edition). T 605/14 is cited in Section I.D.4.4.2 but this is in support of the concept of the reformulation of the subjective technical problem into an objective technical problem in the use of the problem-solution approach. It is not, however, cited in the case law book in connection with the negative premise in the Guidelines that antibodies do not involve an inventive step, unless the specified exceptions apply, nor with assertions 1) and 2) above..

In this regard, it should not be forgotten that the two cited decisions are not actually binding on subsequent decisions. While the Technical Boards are required to consider an interpretation or explanation of the EPC that is provided in earlier Technical Board decisions – and the grounds for any deviation must be givenRules of Procedure of the Boards of Appeal: Article 20(1). – they are not in fact bound by any earlier Technical Board decision. Moreover, a statement in one case is often only applicable to the facts of that case and, therefore, not generally applicable as a broad legal principle. Consequently, without a higher instance ruling which specifically establishes the binding and broad premise that a novel antibody binding to a known antigen does not involve an inventive step unless the specified exceptions apply, this premise in the Guidelines is simply not justified from a legal perspective.

Furthermore, a detailed examination of the reasoning in the two cases, T 187/04 and T 605/14, reveals that they do not themselves provide a solid support for the direction that the examiners are given in assertions 1) and 2) above.

T 187/04

T 187/04 concerned a claimEnglish translation of Claim 1 of the main request “1. An antibody against a fusion polypeptide comprising a histidine portion, wherein the antibody is directed against the histidine portion and the latter comprises 6-18 consecutive histidine residues.” to an antibody that was directed against a fusion polypeptide comprising a poly-histidine peptide marker. Claim 1 was interpreted to relate to antibodies which exclusively bind within the marker.T 187/04 Reasons 2.1- 2.3 and 7: Claim 1 was interpreted as encompassing on the one hand a polyclonal preparation including antibodies which bind to different epitopes all exclusively within the histidine portion of the fusion polypeptide, and on the other hand, a monoclonal antibody preparation with antibodies which exclusively bind to one specific epitope within the histidine portion. The patent included examples to both a polyclonal preparation and a monoclonal antibody disclosed by reference to a deposit of a hybridoma.The patent makes a shorthand reference to the antibody being deposited, but the deposit receipt indicates that a microorganism was deposited (i.e. the deposit was of a hybridoma that would express the antibody). The amino acid sequence of the monoclonal antibody was not provided.

In relation to the generation of the claimed antibodies, the Board opined that such antibodies would have been routine to produce because (i) techniques to isolate monospecific antibodies from polyclonal sera were well developed at the filing date; (ii) the production of monoclonal antibodies by the Köhler-Milstein [hybridoma] method was known for 20 years; and (iii) no difficulties in the production of the claimed antibodies were disclosed in the patent. On that basis, the Board decided that the claimed subject matter was not inventive.T 187/04 Reasons 7-8.

The Board also rejected a narrower claim“1. Monoclonal antibody against a fusion polypeptide comprising a histidine portion, characterised in that the antibody is deposited with the DSM under ACC 2207.” (English translation). to a monoclonal antibody defined by the deposit. The reasoning applied to this claim (Reason 11 of T 187/04) is that which now finds itself in the Guidelines. Specifically, the Board commented that because it had already decided that the class of antibodies to which the claimed monoclonal antibody belonged was not inventive, this individual antibody was likewise not inventive, it merely being a species of the class as a whole. The Board also made the statement:

“The mere fact that a monoclonal antibody was provided in an individualised form for the first time cannot justify inventive step” (English translation and emphasis in bold added).T 187/04 Reason 11. “… Die Tatsache allein, dass ein monoklonaler Antikörper in individualisierter Form das erste Mal bereitgestellt wurde, kann die erfinderische Tätigkeit nicht begründen.“

Here it must be emphasised that the Board’s decision was made in the context of the earlier reasoning that the manufacture of monoclonal antibodies could be routinely accomplished in practiceT 187/04 Reason 7.2. and in relation to a claim which lacked any recitation of chemical structure. It must therefore be questioned how this particular appeal case could support the statement in the Guidelines – now applied to antibodies defined by their sequence – that “[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.” A discussion of structural considerations was completely absent from T 187/04.

In this regard, the monoclonal antibody in T 187/04 was defined by reference to a deposited hybridoma, which could be used to produce the antibody itself. Such deposits were and still are, a mechanism available for biological material to fulfil the requirements of Art. 83. While the deposited cell line, if correctly cultured, should have been able to produce an antibody – and that antibody inherently would have a sequence – the structural sequence information of the antibody itself was not disclosed or claimed. Moreover, a later Board decided that the deposit of a hybridoma does not itself convey any technical information about the structure of an antibody, including its amino acid sequence.T 32/17 Reason 12. See also the Case Law of the Boards of Appeal (July 2022, Tenth Edition) I.C.5.2.7. In sum, the statement in T 187/04 was not made in the context of a claim that recited a specific chemical structure, and nor could have been: in the relevant patent application, the sequence information of the antibody was not available.

Now for a principle to be broadly applicable to subsequent cases, it is surely a minimum requirement that the principle itself must have undergone a thorough analysis in the original case and have been part of the ratio decidendi. Moreover, what is surely decisive to whether the statement in T 187/04 is applicable to subsequent cases in which an antibody is claimed by its specific amino acid sequence is not whether the antibody claimed in T 187/04 (produced by the deposited hybridoma) inherently had a structure, but whether that structural information was both disclosed and claimed. In the absence of the sequence information from the claim, the unpredictability of the impact of structural changes on the desirable function was not, and could not, have been considered.

In summary, it is a stretch to cite Reason 11 of T 187/04, as the Guidelines have done, as a purported justification for the assertion in the Guidelines – now applied to antibodies defined by their sequence – that “[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”. The statements in T 187/04 could not possibly have been intended to prevent the long-standing principles espoused for chemical compounds (CLBA, Section I.D.9.9.2: ‘Structural Similarity’) from being applied to antibodies: such an important derivation would certainly have deserved more consideration, in the context of a claim which actually specified a structure.

T 605/14

T 605/14 concerns an anti-angiopoietin-2 (“anti-Ang-2”) antibody, which was useful for diagnostic and therapeutic purposes. Claim 1 as granted“1. An antibody or antigen binding fragment thereof comprising a variable light chain having a sequence defined by SEQ ID NO:81 and a variable heavy chain having a sequence defined by SEQ ID NO:79.” related to an antibody defined by its heavy and light chain variable region sequences (those of the specific antibody “mAb 3.19.3”). The closest prior art was another anti-Ang-2 antibody (“Ab 536”) also described by its sequence. The Board concluded that the data in the application did not show an improvement for the claimed antibody over the prior art antibody and thus formulated the objective technical problem as the provision of an alternative anti-Ang-2 antibody which could be used in the therapy of pathological angiogenesis. It then asserted that the skilled person, when faced with this problem, would have obtained such alternative antibodies by applying routine methods.T 605/14, Reason 23. [A further general discussion about the ‘routine methods analysis’ will be presented in Part 3 of this paper].

T 605/14 is cited in the Guidelines with particular reference to Reason 24, which says:

“That the structure of the thus obtained antibodies, i.e. their amino acid sequences, is not predictable, is not a reason for considering these antibodies as not obvious.”

While the Guidelines have paraphrased this statement from T 605/14, they fail to appreciate that the statement itself is unsupported in the decision. In particular, there is no justification in the subsequent sentence of Reason 24, which reads:

“The skilled person, aiming at providing an antibody with a particular functional property, knows that such a functional property is the result of the antibody’s amino acid sequence, but knows also that such a functional property may be embodied by various different amino acid sequences.” (emphasis added)

At best, this sentence in Reason 24 is simply acknowledging that antibodies having a particular functional property can have different primary sequences – but it does not say any more than that. In particular, it does not address the question of whether a specific amino acid sequence, if it were to have been contemplated in advance, would have been expected to exhibit the desirable functional property of a therapeutically effective level of binding to Ang-2.

Finally, this sentence in Reason 24 of T 605/14 does not consider or explain why the long-standing principles espoused in the case law for low molecular weight chemical inventions (CLBA I.D.9.9.2: ‘Structural Similarity’) should not apply to an antibody defined by sequence information (i.e. by a specific molecular structure). It does not therefore support the proposition in the Guidelines that antibody inventions should be treated differently from other chemical inventions. Moreover, such a deviation would surely have deserved a more extensive analysis before examiners are directed away from considering structure-function unpredictability in an obviousness analysis.

It is noted that both T 187/04 and T 605/14 cite earlier Board of Appeal decisions as support for their reasoning.T 187/04 cites T 645/02, T 906/91; T 605/14 cites T 735/00, which in turn cites T 645/02, as well as T 512/94. Each of the earlier decisions they cite concern a monoclonal antibody described by the deposit of hybridoma cell line, without any disclosure of the sequence of the antibody. Hence, none of these earlier cases, as mentioned for T 187/04, deliberated on claims to antibodies defined by their amino acid sequence, let alone exclude from the inventive step analysis the assessment of the (un)predictability of the impact of sequence changes on the desirable functional properties.

In conclusion, the broad, negative premise in the Guidelines, that a novel antibody binding to a known antigen does not involve an inventive step unless certain limited exceptions apply, is not justified by the case law. However, this statement has created an artificial barrier between antibodies on the one hand, and low molecular weight compounds on the other, by deeming that the former subject matter is prima facie unpatentable, unless exceptions are proven. In this regard, it should be noted that i) both of the Board of Appeal decisions that are cited in the Guidelines are not binding on subsequent decisions; and ii) neither of those decisions in any way explains why the principles espoused in the CLBA, Section I.D.9.9.2: ‘Structural Similarity’ should not also apply to antibodies. A consistent path for handling antibody based chemical inventions like low molecular weight compounds should therefore not be prevented by the Guidelines. The artificial barrier between these two chemical fields should be removed.

PART TWO

In Part 1 of this paper, the EPO Guidelines relating to the inventive step of antibodies, and the Board of Appeal decisions that were used to support the negative premise in those Guidelines that a novel antibody binding to a known antigen does not involve an inventive step unless limited exceptions apply, were dissected in detail. In Part 2 of this paper, two of the exceptions to that premise are investigated: the existence of a ‘surprising technical effect’ and the claiming of ‘a novel type of functional antibody format’. From these exceptions it can be seen that the assessment of structural similarity and the unpredictability of the impact of structural differences on function is in fact evident in the inventive step analysis. The question that remains, therefore, is why the application of that principle is confined to those exceptions, rather than being applicable more generally. For example, why do the Guidelines encourage the assessment of structural considerations at the level of the overall antibody format, yet discourage that approach at the level of the amino acid sequence?

Surprising technical effect

The first exception in the Guidelines to the negative premise concerns the existence of a ‘surprising technical effect’. In practice, examiners will tend to focus on this exception when assessing a claim to an antibody defined by its sequence.

At the outset, it is important to recall that there is no requirement under Art. 56 EPC for any claimed subject matter to be an improvement over the prior art. This is clearly stated in the Case Law of the Boards of Appeal (CLBA): “[f]or an inventive step to be present, it is not necessary to show improvement – substantial or gradual – over the prior art … An earlier solution to a given technical problem does not preclude later attempts to solve the same problem in another, non-obvious way.”CLBA (July 2022, Tenth edition) I.D.4.5. The early antibody case law must therefore be examined to understand how the requirement to show a ‘surprising technical effect’ has arisen to such prominence.

The decisions in the Guidelines, T 187/04 and T 605/14 (via T 735/00), each cite T 645/02 as the basis for recognising the presence of an inventive step in a claim to an antibody when a surprising or unexpected property is demonstrated.T 187/04 Reason 12; T 605/14 Reason 25, which cites T 735/00 Reason 26. As such, it is important to look at this earlier decision.

T 645/02 concerned the provision of a monoclonal antibody that was described and claimed by the deposit of a hybridoma. However, as pointed out by Ingham and Smyth in their earlier paper,Ingham S., 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, Pages 154 –164, see page 156. the original text in T 645/02 did not actually refer to a “surprising technical effect”. To the contrary, a closer reading of the decision, and in particular of Reason 7,T 645/02, Reason 7: “In this case, it is not the theoretical possibility of producing a particular monoclonal antibody by using a known method, but the actual provision of an antibody with precisely defined properties not disclosed in the prior art that has elements of surprise, that justifies the recognition of an inventive step.” (English translation) shows that the element of surprise was in the actual provision of an antibody with desirable properties, rather than being present in those properties per se. This interpretation is grammatically correct. The phrase as written is:

“…sondern die tatsächliche Bereitstellung eines Antikörpers mit genau definierten, im Stand der Technik nicht offenbarten Eigenschaften, die Elemente der Überraschung aufweist, welche die Anerkennung einer erfinderischen Tätigkeit rechtfertigt.“

The confusion arises as the relative pronoun “die” before the words “Elemente der Überraschung” (elements of surprise) could potentially refer back to (i) the word ‘provision’ (“Bereitstellung”); or (ii) the word ‘properties’ (“Eigenschaften”). However, the grammatically correct interpretation is option (i). If it were option (ii), then the verb in the clause would have been in the plural (“aufweisen”, rather than “aufweist”). This difference is of crucial importance.

An analysis of the original German text confirms that this was not only what was explicitly written, but also what is consistent with the underlying reasoning. Thus, preceding this statement in Reason 7 of the decision is an acknowledgement by the responsible Board that the classical [Köhler-Milstein] fusion method for producing monoclonal antibodies did not give any indication of the chances of success of isolating a particular monoclonal antibody which had the precisely defined properties. Thus, the paragraph as a whole conforms to the notion that it is the provision of a particular antibody with the properties in question that was considered to be surprising: the properties themselves did not per se have to be surprising or unexpected from the state of the art.

In T 645/02, the sequence information of the deposited antibody was not disclosed in the patent. As a result, the decision contains no analysis of a sequence limited antibody claim, and nor does it investigate the more granular issue of the unpredictability of the impact of sequence changes on the properties of an antibody. The Board did recognise that, in comparison to the prior art antibody, the claimed antibody had a different isotype (which facilitated immunoprecipitation by one antibody and detection by another), a higher inhibition of the ligand-receptor interaction and a higher affinity for its target.T 645/02 Reason 5. The use of the particular immunogen for the production of the claimed antibody was also considered to be not obvious. It is noteworthy, however, there was no discussion in the decision of whether it was surprising or not that such differences or improvements were possible, e.g. whether it was in any way unexpected that a known property, such as binding affinity for the target, could be improved beyond the level that was shown in the prior art, or whether it was thought to have reached its maximum potential. Hence, what was considered to be surprising, and the source of the unpredictability, was not actually the possibility of obtaining an improvement. It was rather the manner in which that improvement was achieved.

Now in order for there to be a surprise in how a particular technical effect is achieved, there must surely be a recognition of the unpredictability in the underlying structure-function relationship. In particular, there must be a tacit acceptance that: (a) not all antibody structures will exhibit the effect; and (b) it is not predictable a priori which particular antibody structures will exhibit the effect. Thus, when considering the element of surprise in the provision of an antibody with desirable properties, it is the unpredictability in the structure-activity relationship of an antibody-based invention that is actually being recognised.

Unfortunately, the decision in T 735/00, when citing T 645/02, abbreviated the analysis of the earlier Board of Appeal ( T 645/02) to the shorthand of identifying “unexpected properties.”T 735/00 Reason 26: “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).” (Bold emphasis added) Likewise, decision T 187/04 referred to “precisely defined properties, not known in the prior art, that exhibit elements of surprise.”T 187/04 Reason 12: “In der Entscheidung T 645/02 (16. Juli 2003) wurde beispielsweise die erfinderische Tätigkeit für einen gegen ein bekanntes Protein gerichteten, hinterlegten Antikörper anerkannt, weil er genau definierte, im Stand der Technik nicht bekannte Eigenschaften hatte, die Elemente der Überraschung **aufwiesen **. (Bold emphasis added to highlight the plural form “aufwiesen”, which attaches to the word “**Eigenschaften **” (properties)).

Now it was certainly not wrong to have acknowledged the position that an inventive step would have been present if a ‘surprising technical effect’ had been observed. However, the shorthand phrase ‘surprising technical effect’ is an oversimplification, which leads to a wrong emphasis on what is ‘surprising’ and moreover a misunderstanding of the underlying principle because the effect is divorced from its structural basis.More recently, T 505/19 (Reason 29) when citing T 645/02, correctly tied the ‘element of surprise’ to the provision of the antibody. Thus, the unpredictability in the structure-function relationship should in fact be recognised at whatever level it exists, and in the inability to identify a priori which of the enormous diversity of antibody structures are able to achieve the effect.

Novel type of functional antibody format

As noted above, the Guidelines also recognise that “a novel type of functional antibody format” may be inventive. The term “format” is typically used to refer to the domain types and arrangements that constitute the overall architecture of an antibody. This exception seems to recognise that antibody-derived compounds are often now based on an incredible array of diverse synthetic formats (e.g. scFv-Fc, VHH-scFv, DVD-Ig, diabodies, minibodies, bi- and multi-specifics). These often comprise multiple domains in various configurations.

At the level of granularity of the overall architecture, the Guidelines thus acknowledge that structural considerations can be used in the analysis of the inventive step of antibodies. The focus is on the provision of a novel format as a means for achieving a particular functional property: what is therefore being recognised is the inability of the skilled person to have conceived of or predicted the claimed structure as a means for achieving the function (i.e. structure-function unpredictability).

It is certainly not wrong to recognise an inventive step when a ‘novel type of functional antibody format is present’.

However, a clear inconsistency in the Guidelines is revealed: while examiners are allowed to make an initial, brief structural comparison with the prior art compounds at a high level of resolution (and to find an inventive step on the basis of overall architecture), they are contrarily instructed that an “[i]Inventive step is not acknowledged solely on the basis that an antibody is structurally different from the prior-art antibodies… The 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.”

Thus, examiners are permitted to consider the level of the overall architecture of an antibody, but they are discouraged from zooming in any further to the level of the amino acid sequence. Moreover, they are encouraged to assess how domains (which may have a variable region) are positioned overall, but not to assess the structure within a variable region, because this structure is typically provided as an amino acid sequence. Rather astonishingly, therefore the Guidelines seek to direct examiners away from considering the CDRs and VH/VL regions, i.e. those elements that are most directly involved in binding to the antigen and achieving the intended function. These structural elements, defining the paratope, have a much greater influence than the overall format on many of these functional effects: the paratope establishes the site on the target antigen at which the antibody binds and the strength of the binding interaction. This in turn has multiple effects on related properties such as selectivity, efficacy, toxicity, and the determination of whether the antibody exhibits e.g. an agonistic or antagonistic behaviour. Thus, the CDRs may form a small percentage of the overall compound, but they certainly should not be treated as minor structural differences, and therefore ignored in a structure-function analysis.

Also becoming increasingly common in the practice at the EPO is a suggestion that the minimum ‘binding unit’ of an antibody is more than just the CDRs. Examiners may require the full VH/VL sequence to be recited in the claims on the basis that a desired activity can be influenced by the framework sequences. In taking this approach, there is an underlying assumption that specific framework sequences are not arbitrary variants but have an important role to play. Conversely, however, the specific sequence information that is embodied by the CDRs and VH/VL is discounted when assessing inventiveness. Thus, even if those sequences are not a predictable solution to the underlying technical problem - that is still not available as a reason for finding the claimed antibody inventive – another perplexing inconsistency in the currently practised approach.

Developing the discussion further, the following analysis will contrast the approach to structural considerations at the level of antibody format or amino acid sequence. A claimed antibody may fall into one of the following scenarios:

1. the claimed [structure] is useful for a particular purpose and is markedly dissimilar from a prior art [structure] that is used for the same or similar purpose.

2. the claimed [structure] is useful for a particular purpose and is substantially identical to a prior art [structure] that is used for the same or similar purpose, except for the presence of a small number of [structural] changes.

3. the claimed [structure] is useful for a particular purpose and is substantially identical to a prior art [structure] that is used for the same or similar purpose, except for a small number of [structural] changes, wherein it could not have been predicted that those specific changes would result in the attainment of a further/improved technical effect.

What seems to be practised by some EPO examiners is that if the term [structure] in the above scenarios is a novel format for an antibody, then an inventive step may be found in at least scenarios i) and iii). However, if the term [structure] is a novel amino acid sequence of the CDRs or VH/VL domains, then scenario i) is completely ignored. Typically, the presence of CDRs that are markedly different from those of a prior art antibody are treated as only being minor differences in the overall structure (because they constitute a small percentage of the total number of amino acids within the antibody). As such, scenario ii) is considered to be the default setting for a novel antibody, this often in turn resulting in a finding of obviousness. Only if the sequence differences can be shown to result in a technical property that is per se unexpected (scenario iii)) can an inventive step be recognised.

Here it is submitted that the inventive step analysis should in fact begin in scenario i), rather than this scenario being completely disregarded: examiners should consider if there is marked dissimilarity from the closest prior art antibody within the CDRs or variable regions and, if so, an inventive step can be recognised in the absence of a technical effect – or an improvement therein – that is per se surprising or unexpected: as noted in the CLBA, “[f]or an inventive step to be present, it is not necessary to show improvement – substantial or gradual – over the prior art … An earlier solution to a given technical problem does not preclude later attempts to solve the same problem in another, non-obvious way.”

In this context, it should certainly be appreciated that there is a great deal of unpredictability in the structure-activity relationship within the variable region. Knowing the desired function or even the desired epitope, does not predict which particular amino acid sequences will be useful as a paratope. Moreover, there is an incredibly vast number of potential permutations in the primary structure (amino acid sequence) of antibody variable domains, in particular on account of the hypervariable regions, i.e. the CDRs. Even the names of these regions (“variable” and “hypervariable”) reflects this diversity. Yet only a small fraction of the total number of permutations will be able to bind to a particular target antigen and exhibit the required functional property: even a single amino acid difference in an otherwise identical sequence may entirely disrupt its properties.Rudikoff S., Giusti A. M., Cook W. D., Scharff M. D. “Single amino acid substitution altering antigen-binding specificity.” Proc Natl Acad Sci USA. 1982 Mar;79(6):1979-83. A change of a glutamic acid to an alanine at position 35 in the first CDR of the heavy chain resulted in loss in binding activity.

This unpredictability in the variable region, has actually been recognised in the antibody case law, e.g. in decision T 67/11 in the context of framework sequences, as discussed below. Why then can it not be recognised in the CDRs?

Unpredictability of the effect of structural changes in the variable region on functional properties

In decision T 67/11, the unpredictability of the effect of structural changes on functional properties was explicitly taken into consideration in an inventive step analysis of a claim that was directed to a humanised antibody which recognised the NGcGM3 ganglioside. The humanised antibody included specific mutations within the variable region as compared to the donor mouse antibody that was known from the prior art. The problem was considered to be the provision of an antibody with a similar affinity to the NGcGM3 ganglioside as that of the known mouse antibody, but which was less immunogenic when administered to human patients.

T 67/11 provides a detailed analysis of the predictability of the effect of structural changes on functional properties, in particular in Reasons 19-25:

20 …it still needs to be determined whether the prior art provides clear instructions to the skilled person to obtain the specific mutations recited in the claims, i.e. whether such mutations were obvious possibilities for the person skilled in the art faced with the problem defined …

21.1 Studies in the prior art in respect of the contribution of the variable domains of the heavy and light antibody chains on the specificity of antibody-antigen binding warn the skilled person that “a single amino acid change may seriously disrupt site structure and in some instances abolish binding” …

21.3 …[there was a] general warning in the prior art that minor structural differences involve significant functional consequences which the skilled person could not have predicted.

22. …the prior art failed to provide the skilled person with any pointer towards any specific set of mutations which would do the job.

24. Accordingly, the expectation of success is inversely related to the difficulty of predicting the successful resolution of a technical problem and is inherently low in cases like the present one where minor changes (point mutations) bring about major differences on the final technical effect and where there is no guidance from the prior art concerning the impact of such possible changes… the board considers that the skilled person had no means to predict the impact of each possible mutation in the FR of the antibody on the final properties of the antibody or antibody fragment. In these technical circumstances, it is the attainment of an antibody (or antibody fragment) having indeed the desired characteristics which is considered surprising, not the theoretical possibility of achieving one. (Emphasis added)

Notably in this case, in line with T 645/02, the Board explicitly recognised that it was the provision of a particular antibody having the desired functional characteristics that was considered to be surprising. In contrast to that earlier decision, however, structural sequence information was disclosed in the application and so the unpredictability of changes within the variable region sequences was considered and acknowledged in the inventive step analysis.

The reasoning in T 67/11 was followed in a more recent decision, T 1171/18. In that later decision, a narrow sequence claim to an anti-biotin antibody, defined by its heavy and light chain variable region sequences, was refused by the examining decision.Claim 1 of the main request “1. An antibody comprising a VH sequence of SEQ ID NO: 12 and a VL sequence of SEQ ID NO: 16.” This sequence was humanised, without significant affinity loss, from a known murine donor antibody. On appeal, the Board recognised the inventiveness of this claim. It further commented that the reasoning in T 67/11 should be considered to be relevant not only when the affinity of the humanised antibody is retained or improved as compared to the donor.T 1171/18 Reason 32. Specifically, even though there was a slight affinity loss after humanisation for the anti-biotin antibody, the subject matter was deemed inventive. Importantly, in coming to its decision, the Board acknowledged the unpredictability inherent in the generation of functional sequences: it observed that it was not known which of the known humanisation techniques to follow, nor which back or forward mutations to choose in order to obtain an antibody with substantially the same affinity and that nor could the skilled person reasonably expect that such an antibody could be obtained at all with any of the known humanisation methods.T 1171/18 Reason 33: “Consequently, taking into account the disclosure in documents D10 to D16 and the fact that no humanised anti-biotin antibody was known in the art, it was not evident to the skilled person which of the known humanisation techniques should be followed and, when following the standard method of CDR grafting, which back- or forward mutations were necessary to obtain a humanised anti-biotin antibody exhibiting about a 2.5-[fold] loss in affinity compared to the parental murine antibody, nor could the skilled person reasonably expect that such an antibody could be obtained at all with any of the known humanisation methods.”

In both T 67/11 and T 1178/18, the conclusions were reached on the basis that the skilled person would not have been able to predict a priori the effect on the binding affinity of the structural differences which distinguished the claimed antibodies from those of the prior art.

Moreover, both cases considered the unpredictability of the structure-function relationship at the level of the amino acid sequences of the variable region. As noted by Rigby in her paper in relation to T 67/11, “[o]nce it is accepted that the characteristics of an antibody are determined by its sequence, and that it cannot be predicted how a change in sequence may affect an antibody’s properties, the assessment of obviousness does not seem to be too distant from the structural non-obviousness approach”.Rigby B. “Is the European Patent Office’s approach to assessing obviousness of antibody inventions consistent?” Expert Opin Ther Pat. 2014 Dec;24(12):1283-5, see page 1285.

In summary, the current EPO Guidelines, which have the effect of directing examiners away from assessing structural considerations at the amino acid sequence level when evaluating antibodies, are simply not justified, not least for the reason that there are appeal decisions which explicitly recognise the unpredictability of the effect of structural changes, within the variable region, on the functional properties of an antibody. These decisions (e.g. T 67/11 and T 1171/18) are certainly more relevant to variable region sequence defined claims than the early hybridoma cases (such as T 187/04), which related to applications from which amino acid sequence information was absent.

At a minimum, therefore, if antibodies are to be singled out for special consideration, then the principles established in decisions such as T 67/11 and T 1171/18 warrant recognition in the Guidelines and, moreover, the approach of structural non-obviousness should be considered in the assessment of the inventive step of at least humanised antibodies. With that being said, creating another exception for humanised antibodies within the overall approach to the inventive step of antibodies is arguably bad practice and, moreover, there is simply no reason to limit the approach of considering the structure-function relationship to this particular subset of antibodies.

Part 2 of this paper has shown that consideration of the unpredictability of the structure-function relationship is evident in the inventive step analysis of antibody cases in the “novel type of antibody format” exception and in the humanisation cases of T 67/11 and T 1171/18. This principle also implicitly underlies the “surprising technical effect” exception (particularly when that term is correctly interpreted using the original wording from T 645/02). However, the fact that the Guidelines apply this principle to certain circumstances only, e.g. to the overall architecture of an antibody but not to its variable region sequences which are primarily responsible for the function of the antibody, is a bewildering inconsistency that needs to be remedied. Part 3 of this paper will consider how a correct focus can be brought to the problem-solution approach to assessing the inventive step of antibodies that are defined by their CDR and VH/VL sequences, i.e. how the practice applied to antibody inventions can be aligned with that applied to low molecular weight compounds.

Part 3 will be published in the next edition of the EPI/Information in 2025.

Comments