T 1394/21 (VISTA and PD-L1-antagonists/DARTMOUTH COLLEGE) 25-04-2023

1. Admittance of documents D83 to D87, D87a, D88, D88a, D89, D89a and D90

1.1 Documents D83 and D84

The argument that antibody 13F3 was not sufficiently disclosed had already been raised in point 1.1.2. of the notice of opposition, so documents D83 and D84 could and should already have been submitted in the opposition proceedings.

Thus, documents D83 and D84 were not admitted into the proceedings (Article 12(6) RPBA).

1.2 Document D85

The board did not consider it necessary to decide on the admittance of document D85 and the submissions based on it, the reason being that the limited therapeutic window or effect mentioned in document D85 would still fall within the scope of claim 1 of the main request. Therefore, neither the data nor the arguments based on them could have changed or influenced the board's assessment of sufficiency.

1.3 Documents D86 to D87, D87a, D88, D88a, D89, D89a and D90

Apart from the fact that documents D86 to D87, D87a, D88, D88a, D89, D89a and D90 do not address the central issue of the lack of positive controls in the D29 experimental report, these documents, which are intended to provide information as to who carried out the experiments of document D29, could and should already have been filed during the opposition proceedings and were therefore not admitted into the proceedings (Article 12(6) RPBA).

2. Admittance of lines of argument made for the first time during oral proceedings in the context of document D70

The objection as to sufficiency of disclosure regarding the claimed medical use had already been raised with the notice of opposition. Document D70 had already been cited in the appellants' reply to the notice of opposition. Furthermore, the arguments under discussion relating to documents D70 and D29 had already been put forward by the appellants during the opposition proceedings (see point 14.2.5 of the decision under appeal). They were reiterated in the statement of grounds of appeal.

During the oral proceedings before the board, the respondent presented a new line of argument based on previously uncited passages of document D70.

Although the board had stated, in its communication under Article 15(1) RPBA, that the data in document D70 confirmed the effect of targeting VISTA and PD-L1, this cannot be construed as an invitation to provide a new line of argument based on document D70. Thus, no exceptional circumstances existed.

Consequently, the board decided that newly cited passages in document D70 and the corresponding arguments made in the context of document D29 were not to be taken into account (Article 13(2) RPBA).

3. Disclosure of the invention - main request

3.1 Anti-VISTA antibody 13F3 and anti-PD-L1 antibody MIH5

3.1.1 The opposition division held that antibody 13F3 was not disclosed in an enabling manner.

The opposition division decided on the other hand that the specific antagonistic anti-PD-L1 antibody MIH5 used in Example 26 had been known from document D54 and, thus, was enabled.

3.1.2 The application as filed does not disclose the specific antagonistic anti-VISTA antibody 13F3 in an enabling way. The board agrees with the opposition division and the respondent that document D13, relied on by the appellants, also fails to provide further details on the 13F3 antibody such as its amino acid sequence, a deposit number of the corresponding hybridoma made according to the Budapest Treaty (see Rule 31 EPC) or proof/a source of commercial availability.

The putative in-house designation 13F3 may be used for different antibodies and does not allow an antibody to be clearly identified.

3.1.3 Similarly, the application as filed does not disclose the specific antagonistic anti-PD-L1 antibody MIH5 in an enabling way. No specific sequence, hybridoma deposit accession number pursuant to Rule 31 EPC, commercial source or any prior art document disclosing the MIH5 antibody has been provided. The application as filed does not cite document D54.

There is no evidence whatsoever that the MIH5 of the application as filed could be the same as the MIH5 of document D54.

Yet even if the application as filed did refer to document D54, said document would not be enabling on its own either since it does not further characterise the antibody designated MIH5.

3.1.4 Although neither antibody 13F3 nor MIH5 is sufficiently disclosed in the application as filed, the board agrees with the appellants that a skilled person would absolutely understand that the two specific antibodies are example antagonistic antibodies to VISTA and PD-L1, respectively, and that other antagonistic anti-VISTA and anti-PD-L1 antibodies may be used in combination in order to perform the invention, provided that the skilled person knew how to generate such antibodies.

3.2 Antagonistic anti-VISTA and anti-PD-L1 antibodies

3.2.1 The opposition division did not deny that the skilled person could generate anti-VISTA antibodies per se without undue burden on the basis of general, commonly known antibody production methods which had been exemplified in paragraphs [0414] to [0419] of the patent (see [0498] to [0505] of the application as filed).

The opposition division, however, was of the opinion that none of the examples in the application as filed/patent provided the detailed technical information and criteria for convincingly identifying antagonistic anti-VISTA antibodies.

Figure 40B showed that only certain types of T cells could be affected and, if they had been affected at all, it had not been in a convincing manner. The tumour size reduction experiments with murine tumour models were difficult and unreliable to work with and certainly did not constitute a meaningful way of screening and selecting/identifying antibodies.

Document D13, to which the application as filed refers, did not provide the technical details and criteria for identifying antagonistic anti-VISTA antibodies either.

Therefore, identifying antagonistic anti-VISTA antibodies constituted an undue burden.

3.2.2 The board cannot agree. It is undisputed that paragraphs [0498] to [0505], [0621], [0706] and [0722] of the application as filed provide the information needed for producing anti-VISTA antibodies in general, i.e. comprising agonistic and antagonistic antibodies.

Example 26 of the application as filed (see paragraphs [0777] and [0778]) refers to the use of MLR assays for identifying antagonistic anti-VISTA antibodies. Paragraphs [0165], [0713] and [0736], and Figures 25A to D of the application as filed describe how to perform such MLR in vitro assays and the effect to be expected for antagonistic anti-VISTA antibodies.

According to the description in paragraph [0180] of the application as filed, Figure 40B provides the effects of immobilised VISTA-Ig and PD-L1-Ig, i.e. the soluble receptors and not the antibodies as (erroneously) indicated in the figure caption for the X-axis, together with anti-CD3/CD28 on the proliferation of CD4+ and CD8+ T cells (see Example 26, paragraph [0774]). Thus, contrary to the opposition division's assessment, this figure cannot show that the assay is unsuitable for identifying antagonistic antibodies.

MLR assays measuring T cell activation were well known in the art and already used for identifying antagonistic antibodies targeting immune checkpoint proteins such as PD-L1 (see documents D4, D17, D47 to D50, D65 or D79) or CTLA-4 (see documents D80 to D82). The application as filed also identifies VISTA as a putative immune checkpoint protein ligand (see e.g. paragraphs [0002], [0013] and [0014]), thus providing a link to the known methods for identifying antagonists to well-known and characterised checkpoint proteins (such as CTLA-4 and PD-L1).

Example 26 of the application as filed also states that VISTA-inhibitory antibodies identified in in vitro assays are further tested in vivo for promoting anti-tumour responses in animal models of cancer (see paragraph [0780]) "e.g. as further described in the preceding examples", i.e. the tumour models of Examples 7 and 8 (see also Figures 20A to D).

The opposition division argued (see point 14.3.2, paragraph 7 of the decision under appeal) that the appellants had indicated in their letter of 6 October 2020 (see page 68, penultimate paragraph) that the B16F10 tumour model was complex and unpredictable. Example 26 (see page 218, fifth and sixth penultimate lines) of the application as filed also states that "[d]ue to its poor immunogenicity, B16F10 also represents a very challenging murine tumour system for immune-interventions against cancer".

It is, however, not clear on which basis the opposition division concluded that "[a]s regards the tumor size reduction experiments, the murine tumor models are difficult and unreliable to work with and certainly do not constitute a meaningful way to screen and select/identify antibodies", i.e. has evidently extrapolated this statement to all in vivo tumour models used in the application as filed, such as Example 8.

The board agrees with the appellants that no evidence or specific technical reason has been provided as to why the skilled person could not repeat the tests of Examples 7 and 8 with another anti-VISTA antibody that has been preselected in vitro to have efficacy in blocking T cell responses induced by VISTA.

As an alternative, the skilled person could use the EAE model used in Example 20 (see paragraphs [0717] and [0737] of the application as filed) and in prior art document D13 for the in vivo evaluation of candidate antagonistic anti-VISTA antibodies.

In the board's view, screening a pool of anti-VISTA antibodies for antagonistic properties using the MLR assays described in the application as filed and further selecting therapeutically interesting clones on the basis of the described in vivo models can be considered routine and does not involve an undue burden for a person skilled in the art in the field of immunology.

Thus, on the basis of the teaching in the application as filed and in line with what is taught in the relevant prior art, such as document D13, a skilled person in the field of immunology would understand that an anti-VISTA antibody that is a "VISTA antagonist" is an antibody that blocks (inhibits or neutralises) the suppressive effects of VISTA on T cell responses. An antagonistic anti-VISTA antibody can accordingly be identified on that basis using the methods described in the application as filed.

The board does not agree with the respondent's assertion that the skilled person needed information on the number of antibodies to be tested before identifying a single antagonistic antibody, since this is an issue inherent to all screening methods.

As argued by the appellants, the skilled person would know that blocking antagonistic anti-PD-L1 antibodies would be suitable positive controls for selecting antagonistic anti-VISTA antibodies which act by deblocking the VISTA-mediated suppression of T cell proliferation (see also [0777] of the application as filed).

3.2.3 As argued by the opposition division, antagonistic anti-PD-L1 antibodies in general and their antagonistic properties, i.e. the properties to be screened for when trying to identify such PD-L1 antagonists, were known in the art, see e.g. documents D4, D17, D47 to D50, D54, D65 or D79. Thus, providing such antagonistic anti-PD-L1 antibodies was not an undue burden and was within the skilled person's capabilities.

3.2.4 Therefore, in the board's opinion, a person skilled in the art was able to provide antagonistic anti-VISTA antibodies on the basis of the teaching in the application as filed and anti-PD-L1 antibodies on the basis of what was known to the skilled person from the relevant prior art.

3.3 Suitability of the claimed antibodies for the claimed therapeutic effect

3.3.1 The independent claims have been drawn up as purpose-related product claims in accordance with Article 54(5) EPC. The subject-matter is directed to an antagonistic anti-VISTA antibody and an antagonistic anti-PD-L1 antibody for use in treating cancer.

According to established case law of the boards of appeal of the EPO, attaining the claimed therapeutic effect is a functional technical feature of claims directed to medical uses.

Thus, under Article 83 EPC, the application as filed must render it credible that the claimed antagonistic anti-VISTA antibody and antagonistic anti-PD-L1 antibody are suitable for treating cancer unless this is already known to the skilled person at the filing date (see G 2/21, especially Reasons 74 and 77).

3.3.2 The appellants argued that the skilled person was able to repeat Example 26 on the basis of their knowledge and the teaching of the application as filed, such as Example 8, which described the effect of an example antagonistic anti-VISTA antibody on the regression of different tumours.

3.3.3 The respondent pointed to the data in the application as filed and to document D29 - an attempt by the respondent to rework Example 26 of the application as filed - as showing that neither an antagonistic anti-VISTA antibody or antagonistic anti-PD-L1 antibody alone nor a combination of them had any therapeutic effect on the growth of B16F10 tumour cells in vivo.

3.3.4 The board agrees with the opposition division that the validity of the negative results for the experiments performed in document D29 cannot be assessed as there are no controls to show that the anti-VISTA and anti-PD-L1 antibodies used in them were functional at all.

3.3.5 Moreover, the board has doubts about the reproducibility of the data provided in Example 26 of the application as filed. Apart from the lack of enablement of the antibodies 13F3 and MIH5 used (as discussed in points 3.1.2 and 3.1.3 above), essential parameters such as the amount of anti-VISTA and anti-PD-L1 antibodies applied, the drug administration route or the number of B16F10 tumour cells inoculated are missing in Example 26, nor is there any relevant reference to other examples in the document.

3.3.6 However, none of the claims requires a synergistic effect of the claimed treatment with an antagonistic anti-VISTA antibody and an antagonistic anti-PD-L1 antibody.

What matters is whether the skilled person would have reasonably expected, on the basis of the application as filed, the common general knowledge and the prior art, that an antagonistic anti-VISTA antibody together with an antagonistic anti-PD-L1 antibody would be suitable for treating cancer.

The application as filed shows that VISTA can suppress T cell-mediated responses (Example 3) and that an antagonistic VISTA-specific antibody enhances cell-mediated immune responses (Example 7).

Example 8 (see also Figures 20A to D) shows that treatment with an antagonistic anti-VISTA antibody reduced tumour growth in four different tumour models in which mice were inoculated subcutaneously with (A) MB49 (bladder carcinoma), (B) MCA105 (fibrosarcoma) or (C) EG7 (thymoma) tumour cells, or (D) intraperitoneally with ID8 (ovarian cancer) tumour cells, and treated with 300 myg antibody every other day beginning one day after inoculating B6 mice with the respective tumour cells, i.e. day +1. Thus, the application as filed/patent credibly shows that an antagonistic anti-VISTA antibody has an anti-cancer therapeutic effect in vivo.

This is consistent with the teaching of prior art document D13, which discloses that VISTA negatively regulates T cell responses and shows that an anti-VISTA antagonistic antibody interferes with VISTA-induced suppression of T cell responses by VISTA-expressing antigen-presenting cells in vitro (see abstract, Figure 10).

As argued by both parties, the anti-cancer effects of anti-PD-L1 antibodies were part of the common general knowledge represented by e.g. document D46 (as also cited in the application as filed in paragraph [0018]) and discussed in document D4 or D17.

Thus, on the basis of the evidence in the application as filed, and having regard to the common general knowledge and the prior art, it was credible and predictable for the skilled person at the filing date that an antagonistic anti-VISTA antibody in combination with an antagonistic anti-PD-L1 antibody was suitable for treating cancer by activating anti-cancer immunity in a subject.

3.4 Document D85 discloses that mice developed strong immune responses against the antagonistic anti-VISTA 13F3 antibody, presumably leading to fast clearance of 13F3. In fact, after a week of continuous treatment no 13F3 antibody could be detected any longer in the serum 24 hours after injection of the antibody (see page 1934, right-hand column, second half of first full paragraph).

The respondent cited this document in support of its argument that there was insufficient disclosure for the claimed medical use. However, the scientific report D85 goes on to explicitly state that "[d]espite the apparent immunogenicity and short half-life of 13F3 in vivo, 13F3 treatment significantly suppressed tumor growth in the B16OVA model" (see page 1934, right-hand column, second full paragraph). Thus, this document does not support the respondent's objection.

3.5 In view of the above considerations, the board is of the view that the application as filed discloses the claimed invention in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art as per Article 83 EPC.

4. Remittal to the department of first instance - Article 111(1) EPC

The appellants requested that the case be remitted back to the opposition division for consideration of the grounds of opposition under Article 100(a) EPC. The decision under appeal did not deal with objections under novelty (Article 54 EPC) and inventive step (Article 56 EPC). Thus, there are special reasons to remit the case to the opposition division for further prosecution, as requested by the appellants (Article 11 RPBA).