T 0650/18 (Sclerostin neutralising antibodies/AMGEN) 23-11-2021

Admissibility of the appeals

1. The appeal complies with the requirements of Articles 106 to 108 and Rule 99 EPC and is admissible.

Main request

2. The description of the divisional application as filed is the same as that of the parent application. The claims of the divisional application as filed are rearranged compared to the claims of the parent application as filed. Hence, the considerations in relation to Article 76(1) and Article 123(2) EPC are the same. In the following, the board deals with the ground for opposition under Article 100(c) in combination with Article 123(2) EPC.

Amendments (Article 123(2) EPC) - claim 1

3. Claim 1 is drafted as a (second) medical "for use" format in accordance with Article 54(5) EPC. The therapeutically used compounds are monoclonal antibodies characterised in that they (a) bind to human sclerostin with a binding affinity of less than or equal to 1x10**(-10) M as determined by surface plasmon resonance and (b) cross-block the binding of antibody Ab-13 or Ab-14 - both defined by their respective heavy and light chain sequences - to sclerostin and/or are cross-blocked from binding to sclerostin by antibody Ab-13 or Ab-14.

4. In the decision under appeal, the opposition division decided that the subject-matter of claim 1 of auxiliary request 1 (corresponding to claim 1 of the current main request) did not relate to subject-matter extending beyond the application as filed for the following reasons.

The application disclosed a list of 28 antibodies, including Ab-13 and Ab-14, and a further list with six different binding affinity ranges. The binding affinity range referred to in claim 1 of less than or equal to 1x10**(-10) being one of them. Since the extraction of antibodies Ab-13 and Ab-14 from the antibody list was per se allowable, the only question to be answered was whether combining the particular antibodies Ab-13 and Ab-14 and the binding affinity range of less than or equal to 1x10**(-10)M constituted an allowable selection from two lists.

The opposition division answered this question in the affirmative. The combination of characterising features was not a selection of independent features from two lists since the binding affinity was an inherent property of an antibody. The selection of one particular binding affinity from the list of binding affinities to specify that inherent property was an allowable limitation and did not result in new technical information. The opposition division saw the circumstances of the current case as being parallel to the circumstances underlying case T 2134/10.

5. For the following reasons, the board is not persuaded by the reasoning in the decision under appeal.

6. As submitted by the appellant - and not contested by the respondent - the binding affinity of a given antibody is (mainly) determined by its heavy and light chain variable regions.

The antibodies according to claim 1 are, however, not defined by specific heavy and light chain variable region sequences but in terms of a functional feature, i.e. by way of their cross-blocking ability: they cross-block the binding of antibody Ab-13 or Ab-14 to sclerostin and/or are cross-blocked from binding to sclerostin by antibody Ab-13 or Ab-14.

Hence, the antibodies according to claim 1 are a structurally undefined and heterologous group of antibodies, i.e. the antibodies encompassed by the group may have any kind of variable region sequences.

7. There is no explicit disclosure in the application that this functionally defined, structurally heterologous group of antibodies displays binding affinities within a specific range. The skilled person, relying on the basis of common general knowledge that binding affinity is linked to an antibody's structure (see point 6. above), would also not derive this from the disclosure of the application.

8. Therefore, the skilled person would not interpret the disclosure of the list of binding affinity ranges on page 110 in the way the respondent submits, namely that it would be clear to the skilled person, in view of the introductory expression of the paragraph "Antibodies according to the invention may have a binding affinity of...", that each of the binding affinities may be exhibited by any of the antibodies of the invention.

9. Assuming that a particular range of binding affinities was linked to the functionally defined, structurally heterologous group of antibodies, the skilled person, would infer, based on common general knowledge, that this could not be each and every range from the list of six affinity ranges provided in the description.

10. Thus, in relation to the reference antibodies referred to in claim 1, i.e. Ab-13 and Ab-14, the affinity ranges in the list are not equivalent. Moreover, since the group of antibodies is not structurally defined, no particular affinity value or range can be considered inherent.

11. Hence, in contrast to the opposition division, the board takes the view that the binding affinity and the cross-blocking ability with reference to antibodies Ab-13 and Ab-14 are independent features disclosed in two different lists.

12. In such circumstances, it is established case law of the boards (see Case Law of the Boards of Appeal, 9th Edition 2019, II.E.2.6) that for the combination of the two features to be allowable, there should be a disclosure in the application (or it being known from the common general knowledge) pointing to their association, for example, a disclosure in an example.

13. However, in this case, there is no explicit nor implicit disclosure in the application pointing to a combination of a binding affinity of less than or equal to 1x10**(-10) M for an antibody capable of cross-competing with Ab-13 and/or Ab-14.

14. In contrast, in the board's view, if guided by the examples as a potential pointer, the skilled person would have chosen a binding affinity similar to the one of the claimed reference antibodies, Ab-13 and Ab-14, i.e. ~1x10**(-12) M (see examples 10 and 11).

15. The respondent referred to decisions T 1621/16 and T 2134/10 to support its position that the combination was allowable.

15.1 With regard to decision T 1621/16 and in particular point 1.7.2 of the Reasons, the respondent submits that the decision confirmed that multiple selections from converging lists were allowable, and so the selection of the range of less than or equal to 1x10**(-10) M from the converging list of affinities disclosed on page 110 of the application did not add matter.

However, the board in case T 1621/16 explicitly stated that amendments based on multiple selections from lists of converging alternatives do not necessarily meet the requirements of Article 123(2) EPC. At least two conditions have to be met for such a selection to be allowable. One of them being that the combination should be supported by a pointer in the application as filed (see Reasons, point 1.7.3).

Apart from the fact that in this case only one of the two lists from which the selection is made can be considered converging (namely the list of affinity ranges), the above cited prerequisite that the combination should be supported by a pointer is not met for the reasons explained above in points 13. and 14. Hence, decision T 1621/16 does not help the respondent's case.

15.2 In decision T 2134/10, also relied on in the decision under appeal, the appellant in that case submitted that the subject-matter of claim 1(d), a polynucleotide encoding a polypeptide having an amino acid sequence at least 95% identical to the sequence shown in SEQ ID NO:66, resulted from the combination of items from two lists, it not being possible to derive this combination from the application documents as filed (see Reasons, points 9 and 10).

The board in question reasoned that there is a direct and unambiguous disclosure of amino acid sequences displaying a specified degree of sequence identity with each of those displayed in Table 1 and that the restriction to amino acid sequences having 95% identity with SEQ ID NO:66 is thus the result of a limitation to one specific degree of identity from among all the degrees specified in relation to SEQ ID NO:66. The board considered that a "specific degree of sequence identity is not a property that, in combination with a particular molecule selected from table 1, could single out a particular molecule or confer properties to the claimed subject-matter not directly and unambiguously derivable from the application as filed".

However, the circumstances underlying case T 2134/10 and the current case are different.

Firstly, in case T 2134/10, there is a direct nexus disclosed in the application between the sequence IDs and the percentages of sequence identities.

Secondly and perhaps more importantly, qualification of a compound by a particular binding affinity selected from among several binding affinities is different from qualification by a particular sequence identity selected from among several sequence identities. The former qualification may indeed define compounds having properties not directly and unambiguously (explicitly or implicitly) derivable from the application. For example, qualification by a particular binding affinity may result in compounds better suited for medical use than those defined by a different binding affinity.

Hence, decision T 2431/10 does not support the respondent's case either.

16. The board concludes in view of all of the above observations that the application as filed does not directly and unambiguously disclose monoclonal antibodies characterised by the following combination of features: that they (a) bind to human sclerostin with a binding affinity of less than or equal to 1x10**(-10) M as determined by surface plasmon resonance and (b) cross-block the binding of antibody Ab-13 or Ab-14 - both defined by their respective heavy and light chain sequences - to sclerostin and/or are cross-blocked from binding to sclerostin by antibody Ab-13 or Ab-14.

17. Consequently, the subject-matter of claim 1 extends beyond the content of the application as filed (Article 123(2) EPC).