T 2354/18 (EZH2 inhibitor/EPIZYME) 25-04-2023

1. The appellant was not represented at the oral proceedings as announced previously (see section VIII. above). In accordance with Rule 115(2) EPC and Article 15(3) RPBA 2020, the proceedings were continued in the absence of the appellant, which was considered to be relying only on its written case.

Admittance of documents and a line of argument on added matter (Article 12(4) RPBA 2007)

2. Under Article 25(2) of the Rules of Procedure of the Boards of Appeal (RPBA), the new Article 12(4) to (6) RPBA does not apply to appeals in which the statement of grounds of appeal was filed before 1 January 2020 and any reply to it was filed in due time. Instead, Article 12(4) RPBA 2007 continues to apply. This is the case for this appeal. According to Article 12(4) RPBA 2007, the board has the power to hold inadmissible facts, evidence or requests which could have been presented in the opposition proceedings.

3. With respect to claim 1 of the main request, on appeal the appellant submitted that the feature that a subject was only "suspected" to have cancer did not have a basis in the application as filed. The respondent remarked that this objection was presented for the first time on appeal and that the appellant did not justify why it had been raised only then; however, Article 12(4) RPBA 2007 does not require a party to submit such a justification. This argument is therefore not a sufficient reason in itself for holding the objection inadmissible.

4. Moreover, an objection that the expression "for determining whether a subject who has or is suspected to have cancer selected from lymphoma and melanoma" in claim 1 had no basis in the application had already been raised in the opposition proceedings (see point 8 of the notice of opposition), albeit with a different argument. The objection to the feature "is suspected to have cancer" raised by the appellant on appeal is directed to a part of that expression and is considered prima facie relevant. Hence, the board decided to exercise its discretion under Article 12(4) RPBA 2007 to consider this objection on appeal.

5. The board decided to consider documents D16 and D17, as well as documents D18 and D19 submitted as documents D20 and D21 with the grounds of appeal (Article 12(4) RPBA 2007). In view of the outcome of this case, it is not necessary to provide reasons for this part of the decision.

6. Document D22 was submitted by the respondent in response to an objection based on document D21. As the board decided to admit document D21, it decided to also consider document D22 under Article 12(4) RPBA 2007 for reasons of equal treatment.

7. It was not required to decide on the admittance of document D23 as this document was cited in the context of an objection which is not relevant for the present decision (see point 14. below).

Main request - patent as granted

Amendments (Article 100(c) EPC) - claim 1

8. Claim 1 is directed to a method for determining whether a subject who has or is suspected to have cancer selected from lymphoma and melanoma is a candidate for a given treatment (for the complete wording, see section I.). As a basis for "a subject who (...) is suspected to have cancer", the appellant referred to the second paragraph on page 6, the first full paragraph on page 31, and the second paragraph on page 54 of the application.

9. The second paragraph on page 6 discloses that an aspect of the invention is "a method of identifying a subject as a candidate for treatment with an inhibitor of EZH2" and that the method "comprises the steps of performing an assay to detect a Y641 mutant of EZH2 in a sample from a subject; and identifying a subject expressing a Y641 mutant of EZH2 as a candidate for treatment with an inhibitor of EZH2, wherein the inhibitor inhibits histone methyl-transferase activity of EZH2".

10. Hence, this section of the application refers to "subjects" in general without defining them any further and therefore, as such, does not provide a basis for a subject who is suspected to have cancer.

11. The term "subject" is defined in the application as including any human subject "who has been diagnosed with, has symptoms of, or is at risk of developing a disorder" (see second paragraph on page 54). This passage hence does not refer to a subject who is suspected to have cancer, either. The appellant argued that there was no fundamental difference between patients who were suspected to have cancer and patients who had been diagnosed with cancer because a diagnosis was never 100% accurate; however, this argument cannot be followed because it contradicts the common meaning of a medical diagnosis, which is the identification of the nature and cause of a patient's symptoms. A subject diagnosed with cancer is a subject who has cancer. A subject suspected to have cancer has not yet been diagnosed.

12. The only passage in the application that uses the expression "suspected of" is the first full paragraph on page 31, which discloses that, in one embodiment, a sample from a subject "includes cells suspected to express Y641 mutant of EZH2, e.g., cancer cells". This expression does not concern a subject who is suspected to have cancer and hence does not disclose this feature, either.

13. Consequently, the application as filed does not disclose the method in claim 1 insofar as it relates to determining whether a subject who is "suspected" to have lymphoma and melanoma is a candidate for treatment with an EZH2 inhibitor. Claim 1 of the patent as granted hence contains subject-matter that extends beyond the content of the application as filed and Article 100(c) EPC prejudices the maintenance of the patent as granted.

14. The appellant had also raised an objection under Article 100(c) EPC against claim 5 as granted and the respondent had relied on document D23 in its argument against this objection; however, in view of the conclusions concerning claim 1 and the fact that claim 5 as granted is no longer found in auxiliary request 1 (see section V.), there is no need for the board to decide on this issue.

Auxiliary request 1

15. The appellant did not provide any arguments in response to the respondent's reply to the appeal, and thus has not raised any objections against auxiliary request 1. In view of the deletions made in auxiliary request 1 (see section V.), only the objections raised by the appellant against the main request concerning sufficiency of disclosure and inventive step starting from D1 as the closest prior art are relevant.

Sufficiency of disclosure (Article 83 EPC)

16. The appellant essentially argued that the claims as granted related to subject-matter which was not sufficiently disclosed because i) the method of claim 1 was not capable of accurately identifying patients that were candidates for treatment with an EZH2 inhibitor, and ii) the method of claim 6 (and ultimately claim 1) covered mutant specific EZH2 inhibitors for which there was no enabling teaching in the patent.

17. The patent discloses that enzymatic coupling between unmutated (wild-type) EZH2 and Y641 mutant EZH2 leads to increased trimethylation of lysine 27 of histone H3 (H3K27) compared with cells comprising only wild-type EZH2 and that this results in the malignant phenotype of cells heterozygous for EZH2 Y641 mutants (see Example 6 of the patent). This effect was calculated for Y641F, Y641H, Y641N and Y641S EZH2 mutants based on steady-state kinetic parameters and demonstrated for various cell lines harbouring Y641N or Y641F mutant EZH2 forms (see Figures 3A and 3B of the patent). The presence of any of these EZH2 Y641 mutations in a cancer patient's sample therefore indicates that this patient is a candidate for treatment with an EZH2 inhibitor. This was not contested by the appellant.

18. However, the appellant argued that other EZH2 Y641 mutations existed that were not gain-of-function mutations for the trimethylating activity of EZH2 and that the detection of such a mutation in a subject's sample would not indicate that the subject was a candidate for treatment with an EZH2 inhibitor. The same was true for subjects homozygous for an EZH2 Y641 mutation. In either case, the claimed method could not be carried out.

19. However, the appellant did not convincingly demonstrate that mutations at the EZH2 Y641 residue other than the five point mutations described in the patent and document D1 would indeed occur in lymphoma or melanoma. Loss-of-function mutations could in principle occur when mutating residues in the active site of an enzyme and loss-of-function mutations were also found in the asparagine-histidine-serine motif in methyltransferases (see document D9, page 2451, lines 12 to 16 of the right-hand column); however, these observations on other amino-acid residues are not sufficient for the skilled person to reasonably expect that loss-of-function mutations would also occur at the EZH2 Y641 residue in lymphoma or melanoma.

20. Document D20 discloses that an artificially created EZH2 Y641W mutant enzyme lost its enzymatic activity; however, this does not prove that an EZH2 Y641W mutation would also occur in lymphoma or melanoma. It is noteworthy that an EZH2 Y641W mutation (codon "TGG") could not arise from a single point mutation in the EZH2 Y641 codon "TAC"; however, all five EZH2 Y641 mutations that were found in lymphoma samples arose from such a single point mutation, as is evident from Figure 1C of document D1. The legend of Figure 1C of document D1 explains that "[w]ith one exception, all EZH2 mutations found in FL [follicular lymphoma] and DLBCL [diffuse large B-cell lymphoma] alter this amino acid [Y641]" and goes on to explain that "[t]he mutants identified comprised five of the eight possible nonsynonymous variants of this codon". This observation supports the fact that it was not likely that the artificially created EZH2 Y641W mutation would indeed be found in lymphoma samples.

21. The appellant also pointed to the fact that the five EZH2 Y641 mutations that were found in lymphoma samples varied in their kinetic parameters and substrate specificity; however, since all these EZH2 Y641 mutant forms have increased trimethylation activity compared with the wild-type EZH2 enzyme (see Table 1 of the patent; Figure 1 and page 3013, lines 4 to 16 of the right-hand column of document D7), the variations in kinetic parameters and substrate specificity observed for them cannot serve as proof, either, that EZH2 Y641 loss-of-function mutations would exist in lymphoma or melanoma. This argument by the appellant is therefore not persuasive either.

22. The appellant also did not convincingly demonstrate that lymphoma patients homozygous for a somatic EZH2 Y641 mutation existed. Document D21, a data overview for a particular lymphoid tumour sample harbouring a somatic EZH2 mutation, alleges that this mutation was homozygous (see page 3 of document D21); however, the publication associated with the data described in document D21 discloses that this lymphoma sample was in fact heterozygous (see Figure 1 of document D22). This thus raises doubts on the data in document D21. Consequently, the appellant did not submit any persuasive evidence that homozygous EZH2 Y641 mutations would occur in lymphoma or melanoma.

23. As regards the appellant's argument that an appropriate treatment would not be selected when the patient had a cancer that did not rely on EZH2 gain of function, the board notes that, in the context of the claimed method, it is not relevant that not all lymphomas and melanomas result from an EZH2 gain-of-function mutation. This is the case because the method is based on the detection of an EZH2 Y641 mutant form and thus, if no mutant form is detected, the subject is not a candidate for the treatment with an EZH2 inhibitor, irrespective of the lymphoma type. Therefore, the fact that EZH2 Y641 gain-of-function mutations were only detected in follicular lymphoma or germinal centre B-cell like (GCB) diffuse large B-cell lymphoma (DLBCL) (see Table 3 of document D1) and that other cancer forms even resulted from EZH2 loss of function (see document D17, page 724, right-hand column, lines 15 to 20) do not prevent the skilled person from being able to carry out the claimed method.

24. Finally, the appellant's argument that the invention as defined in claims 1 and 6 as granted (claims 1 and 5 in auxiliary request 1) was not sufficiently disclosed in the patent insofar as it covered mutant-selective EZH2 inhibitors other than the two selective EZH2 inhibitors disclosed in the patent is not persuasive, either. The claimed method identifies subjects in which the H3L27 trimethylation activity of EZH2 is increased as candidates for the treatment with a (mutant-selective) EZH2 inhibitor. This method can be carried out by the skilled person, irrespective of whether or not the EZH2 inhibitor is disclosed in the patent or known in the art.

25. In view of the above considerations, neither of the appellant's arguments for why the invention defined in the claims was not sufficiently disclosed in the application is persuasive. The requirements of Article 83 EPC are therefore met.

Inventive step (Article 56 EPC)

26. In the decision under appeal, document D1 was seen as the most suitable starting point for the assessment of inventive step. In the statement of grounds of appeal, the appellant agreed with this finding (see point (42) of the statement of grounds of appeal). The board sees no reason to differ from this assessment. Document D1 describes somatic mutations at residue Y641 of EZH2 in biopsies taken from follicular lymphoma and DLBCL of germinal-centre origin (see title, page 182, right-hand column, Tables 2 and 3, Figure 1 of document D1). It discloses that EZH2 Y641 mutant forms have reduced enzymatic activity to trimethylate an unmethylated peptide substrate in vitro (see last sentence of the abstract, sentence bridging the left-hand and right-hand columns on page 184, Figure 2 of document D1).

27. Document D1 is neither concerned with the selection of a treatment for patients afflicted with these cancer types nor compares any treatment options for these patients. Therefore, the objective technical problem cannot be formulated as that of providing an improved treatment for lymphoma and melanoma patients, as suggested by the respondent. The board therefore decided to assess inventive step based on the objective technical problem as formulated by the appellant, which was that of identifying a treatment for subjects having the EZH2 Y641 mutations disclosed in document D1.

28. Document D1 discloses that EZH2 Y641 mutant forms have reduced enzymatic activity in vitro and states that "Tyr641-altering mutations of EZH2, and possibly a reduction in H3K27 trimethylation, are involved in the pathogenesis of GBC lymphomas" (see the sentence bridging the left-hand and right-hand columns on page 184 of document D1). From this disclosure, the skilled person would not have considered that patients harbouring an EZH2 Y641 mutation were suitable for the treatment with an EZH2 inhibitor. The claimed subject-matter was therefore not obvious to the skilled person from the experimental data disclosed in document D1.

29. The appellant argued that the skilled person would have understood that the EZH2 Y641 mutant forms disclosed in document D1 were likely to be gain-of-function mutations, despite the experimental evidence in document D1. The board is not persuaded by these arguments for the following reasons.

30. It is true that document D1 discusses that its experimental data do not rule out the possibility that the EZH2 Y641 mutations "may alter the product (or target) specificity of EZH2", as known for the so-called phenylalanine-tyrosine switch site in other SET domain-containing proteins (see lines 2 to 14 in the right-hand column on page 184 of document D1); however, in the same passage, document D1 also discloses that the Y641 residue is distinct from the phenylalanine-tyrosine switch site. Moreover, an altered product or target specificity encompasses options other than an overall gain-of-function activity due to increased H3K27 trimethylation. The skilled person therefore would not have realised from this disclosure in document D1 that the EZH2 Y641 mutations must be either gain-of-function or loss-of-function mutations.

31. Document D15 is a scientific article published as a printed version in the journal "Blood" on 9 December 2010, and, according to the footer on the first page of this article (page 5247 of the journal), it was pre-published online as Blood First Edition paper on 24 August 2010. The respondent's doubts as to whether document D15 belonged to the prior art were thus unfounded. Furthermore, the respondent did not file any evidence in support of its allegation that the online version of document D15 was different from document D15 as submitted in opposition. Document D15 was therefore held to be part of the prior art under Article 54(2) EPC.

32. Document D15 discusses the role of EZH2 in GBC cells and lymphomagenesis. It mentions that the EZH2 Y641 mutations described in document D1 are heterozygous, which "is not incompatible with an overall gain-of-function effect of the mutation" (see page 5254, right-hand column, last paragraph). A similar statement can be found in document D17, which considers that the fact that no EZH2 mutations other than at residue Y641 were detected and that all the Y641 mutations seemed to be heterozygous suggested "a very specific mode of action that may involve a gain of function" (see page 724, right-hand column, first full paragraph of D17).

33. However, these comments in document D15 and D17 are just theoretical speculations. Neither of these documents discloses any data on EZH2 Y641 mutant forms. Document D15 also states in the same paragraph that the EZH2 Y641 mutant forms reduced H3K27 methylation in vitro and that the in vivo impact of these mutations was not known (see page 5254, right-hand column, last paragraph). The conclusion in document D15 that therapeutic targeting of EZH2 might have significant anti-lymphoma effects (ibid.) therefore cannot be understood in the context of EZH2 Y641 mutant forms of which the in vivo impact is decidedly unclear. Instead, it must be seen in the context of the experimental data in document D15 itself, which show that siRNA-mediated downregulation of EZH2 in DLBCL cells results in cell cycle arrest and upregulation of tumour suppressor genes (see page 5254, left-hand column, first paragraph).

34. Consequently, documents D15 and D17 do not contain any teaching that would have necessarily caused the skilled person to question the experimental data on the enzymatic activity of the EZH2 Y641 mutant forms as disclosed in document D1.

35. Document D16 discloses that a Y to F mutation in the catalytic domain of the histone methyltransferase G9a increased trimethylation activity of the mutant enzyme on an H3K9 peptide (see Figure 3 of document D16). This document therefore concerns a different enzyme (G9a) which methylates another substrate (lysine 9 in histone H3). The argument that the skilled person would immediately transfer the teaching on G9a to EZH2 is based on hindsight, which must be avoided. The board is therefore not persuaded that the teaching of document D16 was sufficient to cast serious doubts on the disclosure in document D1 that the Y641 EZH2 mutant protein had reduced enzymatic activity.

36. The appellant's argument that the skilled person, in view of their doubts on the in vivo function of the mutated enzyme, would have tested it in a routine assay with the expectation of finding gain of function, as evident from post-published documents D8 and D9, is also based on hindsight. As discussed above in the context of documents D1 and D15, the in vivo function of the EZH2 Y641 mutant forms was not known on the priority date of the application and there were not just two possibilities (gain of function or loss of function) for how the Y641 mutations might have altered the in vivo function of the enzyme (see points 30. and 33. above). Therefore, even if the skilled person felt the need to further investigate the function of the mutated enzyme, they could not have known in advance which functional tests on which substrates were necessary to elucidate the in vivo function of the enzyme and which outcome was to be expected.

37. Consequently, neither the comments in document D1 on the phenylalanine-tyrosine switch site in SET domain-containing proteins nor the disclosure in any of documents D15, D16 and D17 was sufficient for the skilled person to seriously consider that the EZH2 Y641 mutant forms were gain-of-function mutants, despite the teaching in document D1 pointing to the contrary. The claimed subject-matter was therefore not obvious to the skilled person in view of the teaching in document D1 combined with that in either document D15, D16 or D17.

38. The appellant also argued that the technical problem was not solved over the entire scope of the claim where the EZH2 Y641 mutation was not a gain-of-function mutation, where the mutation was homozygous and where the patient had a cancer that did not rely on an EZH2 gain-of-function mutation; however, since the purpose, i.e. the effect, of the claimed method is expressed in the claim, the appropriate legal provision for dealing with these arguments is sufficiency of disclosure (see the Enlarged Board of Appeal's decision G 1/03 (OJ EPO 2004, 413), point 2.5.2 of the Reasons). These arguments were therefore considered under Article 83 EPC (see points 17. to 23. above).

39. In view of the above considerations, the appellant's arguments on inventive step were not persuasive. The claimed subject-matter involves an inventive step in the sense of Article 56 EPC.