Merck Sharp & Dohme Ltd v Ono Pharmaceutical Co Ltd & Anor

This is a patent case concerning EP (UK) 1 537 878 entitled “Immunopotentiating Compositions”. The patent claims a first priority date of 3^rd July 2002 based on a Japanese filing JP 2002194491. The application was filed on 2^nd July 2003 and the patent was granted on 22^nd September 2010. The patentees are Ono Pharmaceutical Co. Ltd and Professor Tasuku Honjo. Prof Honjo is one of the inventors. Bristol Myers Squibb are the exclusive licensee. I will refer to the patentees and exclusive licensee together as Ono. The patents relate to the use of anti-PD-1 antibodies for the treatment of cancer. Anti-PD-1 antibodies are one type of a revolutionary new kind of cancer immunotherapy. Ono has developed an anti-PD-1 antibody called nivolumab (brand name Opdivo) and Merck has developed an anti-PD-1 antibody called pembrolizumab (brand name Keytruda). Both have obtained clinical approval for the treatment of some cancers and are in clinical trials in relation to many others. An article from the Financial Times on 5^th June 2015 which was in evidence reports a figure of $35bn as the consensus on Wall Street for the annual revenues generated by immunotherapies.

Ono contends that Merck’s product infringes the patent and Merck contends that Ono’s patent is invalid. Ono also states that in this jurisdiction if its infringement claim is successful, given the life-saving nature of this therapy, it will not seek an injunction provided an appropriate royalty is agreed or awarded by the court for future infringement. The fact that Ono adopts this stance does not mean the patent is to be judged by a different standard from any other.

The relevant claims are claims 1 and 3, as follows:

Claim 1 is in Swiss form while claim 3 is a purpose limited product claim (EPC 2000). For the purposes of this case there is no difference between them.

The issues which fall to be decided are:

Wyeth 499 is a citation under s2(3) of the 1977 Act and as such is available for anticipation only and not obviousness.

Merck also cites the paper “Involvement of PD-L1 on tumor cells in the escape from host immune system and tumor immunotherapy by PD-L1 blockade” by Iwai et al (2002) PNAS Vol 99 no. 19 pp12293-12297 published on 17^th September 2002. This Iwai paper was published before the second claimed priority date for the patent. Ono accepts that if the claims are not entitled to the first priority date then they are invalid in the light of Iwai.

There is no distinct issue of infringement. Pembrolizumab is a humanised monoclonal antibody that inhibits the immunosuppressive signal of PD-1 and Merck proposes to market it as a treatment for unresectable/metastatic melanoma. Merck accepts that if the patent is valid, its product falls within the ambit of the relevant claims.

Ono has applied to make a conditional amendment to the claims. The only amendment application live by the closing was to limit claims 1 and 3 to melanoma treatment. Such an amendment is formally allowable. Merck accepts that in the amended form those claims are novel, entitled to the first priority date and not insufficient. Merck maintains its obviousness case against them.

The parties prepared an agreed primer to explain the technical background and agreed aspects of common general knowledge. This was a very useful document. Important parts of it dealt with the adaptive immune system, T-cells, antibodies (polyclonal and monoclonal, structure, function and how they are made), transgenic mice, mouse models and the B7 family of molecules and their receptors.

This case is about a complex area of science. I wish to pay tribute to the parties’ efforts in educating the court and in particular to the time and effort put in to explaining things by all three expert witnesses. To attempt a full and accurate summary of the technical background for inclusion in the judgment would be a disproportionate task. An alternative would be to simply annex the primer but it runs to 68 pages with 113 paragraphs and annexing it does not assist. The approach I will take is to explain technical points as they arise if it seems necessary to do so.

A bibliography of the papers referred to in this judgment is in Annex 1.

Merck called Professor Vassiliki Boussiotis as an expert. Prof Boussiotis is a professor of Medicine at Harvard Medical School and Beth Israel Deaconess Medical Centre and a member of the Dana-Farber/Harvard Cancer Centre. Since 1998 she has run a laboratory focussing on mechanisms and signalling pathways involved in the regulation of activating and inhibitory T-cell responses. She is also a clinical oncologist and treats patients with haematological cancers particularly lymphomas and leukaemias. She is therefore both an immunologist and an oncologist.

Ono called Professor Christopher Rudd as an expert. Prof Rudd is a professor of Molecular Immunology and head of the Cell Signalling Section at the University of Cambridge. He has held previous professorial positions at Harvard Medical School and Imperial College, London. He is an immunologist. A particular focus of his early work relates to T-cell activation and he has played a leading role in uncovering the mechanisms by which co-receptor signals function.

Ono also called Dr Renzo Canetta. He is Vice President of global R&D oncology policy at Bristol Myers Squibb and has worked in clinical oncology since 1974. In these proceedings Dr Canetta was originally called as a fact witness but his witness statements included expressions of his opinions on various topics. Merck objected to this course and Ono decided to designate Dr Canetta as an expert, with Dr Canetta filing a further short statement explaining that he understood his duties to the court as an expert witness and was giving his evidence on that basis. It is a pity that Dr Canetta’s evidence was produced in this way.

Each party criticised the witness(es) called by the other party. The criticisms were extensive and a number of them were serious. I have considered all of them carefully. I found none of them convincing. In my judgment all three of Prof Boussiotis, Prof Rudd and Dr Canetta were thoroughly honest and highly experienced experts in their respective fields who were seeking to explain to the court what they thought about the various issues to the best of their ability.

Perhaps aspects of the way in which Prof Boussiotis’s report was prepared could be criticised but even if those points are well founded, none of them persuaded me that I should discount the evidence of the professor. The way in which Prof Rudd answered the questions put in cross-examination involved him giving quite lengthy answers, to the evident frustration of the cross-examiner. In the course of doing so he mentioned points which Merck contended were new. In fact many of them were developments of points Prof Rudd had made in his report but it is quite true that his oral evidence placed emphasis on aspects which his reports had not and did raise new points of detail. That occurred because the witness was doing his level best to answer the questions put properly and fully. None of this persuaded me to discount Prof Rudd’s opinions. Nor was I persuaded that Prof Rudd’s expertise was not of direct bearing on the issues in this case, or was of lesser relevance than the expertise of Prof Boussiotis. As regards Dr Canetta, I have already mentioned the unfortunate way his evidence was produced.

All witnesses to a greater or lesser extent referred to post-published information and I have taken that into account. There are points of particular detail relating to all three witnesses but they are best dealt with when they arise.

Prof Boussiotis and Prof Rudd expressed their definitions of the skilled person in different words but subject to one point, it was common ground that they were equivalent. I will use Prof Rudd’s formulation. The patent is addressed to a team including a cellular immunologist working on receptors and the function of T-cells in the immune system. The team would include members with experience in producing and modifying antibodies, for example, a person skilled in preparing humanised antibodies or in working with transgenic mice. To this formulation I will add a couple of points mentioned by Prof Boussiotis. The team would have experience of animal cancer models. For certain aspects the team would additionally require the assistance of an oncologist familiar with the treatment of cancer with antibodies and conventional surgical care.

The debate was about signal transduction pathways. These are the mechanisms which operate inside a cell whereby the signal caused by a receptor on the cell surface being activated is acted upon. Prof Rudd’s view was that the knowledge of the skilled team would include knowledge of signal transduction pathways. Prof Boussiotis did not agree. Her evidence was that there was no need to have worked on signal transduction pathways and in support she explained that several leaders in the field at the time had not done so.

Much of Prof Rudd’s work has been concerned with signal transduction and he clearly regarded it as a vital element in the team’s consideration of issues relating to receptor function and T-cell activation. Prof Boussiotis personally had significant expertise in signal transduction but her point about other leaders in the field was that others did not. The debate is one of degree and in my judgment each side’s submission risks putting the matter too high. The team would include individuals with a working knowledge of signal transduction pathways. The skilled team would not ignore signal transduction but equally, while they are important in broad terms, detailed signal transduction pathways are not what the skilled team would be focussed upon.

The law on common general knowledge was summarised by Arnold J in KCI Licensing v Smith & Nephew [2010] EWHC 1487 (Pat) at paragraphs 105-112 which was approved by the Court of Appeal [2010] EWCA Civ 1260 at paragraph 6.

One of the major issues concerns knowledge of contradictory reports relating to the inhibitory nature of the PD-1 pathway. The expression “generally regarded as a good basis for further action” has been used in the case law about common general knowledge and came under scrutiny before me. The phrase comes from General Tire v Firestone [1972] RPC 457 and was used by the Court of Appeal in that case to distinguish from the phrase “accepted without question” used by Luxmoore J in the passage quoted in General Tire from his judgment in British Acoustic Films (53 RPC 221 at 250). The court in General Tire thought that Luxmoore J’s term may have put the matter too high and preferred their phrase.

I do not believe the court in General Tire was seeking to address factual circumstances like those said to arise in this case. In principle the common general knowledge of a skilled person must be capable of including contradictory ideas on a topic, always assuming that information reaches the standard for common general knowledge. The existence of a defined area of doubt and uncertainty does not mean that, in principle, such knowledge is not part of the common general knowledge. An example, referred to by Ono, was in the judgment of Floyd J in Regeneron v Genentech [2012] EWHC 657 (Pat) e.g. at paragraph 67 and the conclusion at paragraph 88 (upheld by the Court of Appeal at [2013] EWCA Civ 93, paragraph 22). Merck submitted the evidence in Regeneron was much stronger than the evidence in this case. The submission about evidence does not alter the point of principle.

Ono submitted, based on Prof Rudd’s evidence, that until the mid-2000s, the field of cancer immunotherapy had been controversial. This requires explanation. The idea of cancer immunotherapy is a very old one. The work of William Coley over a century ago indicated that the human immune system might be involved in fighting at least some cancers. On the other hand, in simplistic terms, if a patient has active cancer, their immune system is self-evidently not stopping the cancer.

The concept of trying to use agents associated with the immune system to attack cancer is old and so too is the idea of trying to trigger the patient’s own immune system to attack their cancer. Whether these ideas work in practice is another matter entirely. It is clear that Ono is correct that the field of cancer immunotherapy had been notably unsuccessful in the past. However Merck’s case is that while that may have been the case beforehand, by the time of the priority date in July 2002 things had changed and the mood of the skilled team was positive and optimistic.

Prof Rudd maintained that in the 1990s many institutes had concluded that immunotherapy would not be a successful approach in treating cancer and that there had been a reduction in research into immunotherapy at that time, citing what happened at the Dana Farber Cancer Institute (at which he worked). He said that it has been the successful post-priority trials of anti-PD-1 and anti-CTLA-4 antibodies, which have dramatically renewed optimism about the potential of immunotherapy. Prof Rudd referred to a paper by Melief et al (2000) which began by stating that the prevailing mood concerning immunotherapy of cancer was “still one of gloom”.

Prof Boussiotis did not agree with Prof Rudd. She explained that immunotherapy had been an important component of cancer research since the 1970s and referred to a number of immunotherapy based cancer treatments which had been investigated and in some cases obtained clinical approval before July 2002. She did not believe the opening sentence of the Melief paper was a fair reflection of the state of mind of the skilled team at the priority date.

Prof Boussiotis did not agree that there had been a reduction in research either generally or specifically at the Dana Farber Cancer Institute. Her evidence was that the label attached to the work was changed, with an increased focus on haematology. I accept that the label attached to the work may have changed but on the question of reduction in research, I prefer Prof Rudd’s evidence about the Dana Farber Cancer Institute on this topic as he was senior in the department at the relevant time and was in a position to know.

Prof Rudd was cross-examined about the Melief paper. It was put to him that the point of the paper was to articulate reasons why, although there had been a negative view in the past (pre-2000) there was now evidence of strong clinical successes and now reasons for optimism. It was also put to him that the overall message of the paper was one of optimism. Prof Rudd maintained that there had been many false starts and that there was a feeling at the time that immunology had done nothing for cancer.

Ono referred to other papers after the priority date which supported Prof Rudd’s view of the position at the time. For example, a citation from the William B Coley 2014 award which was given to Prof Honjo and others refers to the effectiveness of PD-1 checkpoint inhibitor drugs as fuelling “renewed” optimism about the power of immunotherapy to treat cancer. Ono submitted that the optimism needed to be renewed because, until the mid-2000s, there had been mixed success. That is a fair point. Another example is Weber (2014) which, as Prof Boussiotis accepted, refers to the disappointing results of immunotherapy trials in the past.

The most relevant prior art immunotherapy treatments were: (i) the administration of cytokines such as IL2; and (ii) the adoptive transfer of T-cells, which involved taking T-cells out of a patient, modifying the T-cells and re-inserting them to fight the cancer. There were also successful anti-cancer treatments which use antibodies (Rituxan and Herceptin) although they are different in kind to the sorts of treatments in issue here. Dr Canetta’s evidence explained the limited impact immunotherapy had achieved in the treatment of cancer by the priority date and in doing so included the proposed therapies relied on by Prof Boussiotis. Prof Boussiotis accepted Dr Canetta’s comments were fair. Clearly some immunotherapy based cancer treatments had obtained clinical approval by the priority date but I am not satisfied they made an impact which is significant as far as this case is concerned.

Nevertheless there are other papers which are written in a more upbeat tone, good examples being the textbook Rosenberg (2000), the paper Tirapu et al (2002) and a review published in April 2002 (Pardoll (2002)).

Prof Boussiotis herself was clearly enthusiastic and optimistic. However in cross-examination she explained that, while there were enthusiastic immunologists working in the field, there were also a lot of people who were not “believers” at all. By “believers” she meant those who believed that the work of Dr Allison on CTLA-4, and by extension the work on PD-1, was likely to provide a cancer treatment.

Overall, while some workers in the field were enthusiasts, there were many others who were not. I can see that the “mood of gloom” referred to in Melief in 2000 may have been starting to lift by 2002 but I do not believe it had lifted to such an extent that Prof Boussiotis’s view in this respect was representative of the uninventive skilled person. I find that a history of pessimism and failure remained part of the thinking of the skilled person at the priority date.

It was common ground that it was known at the priority date that some cancers were more responsive to immunotherapy than others.

Receptors are protein molecules on the surface of a cell with a part outside the cell to interact with the extra-cellular environment and a part inside the cell to interact with the intra-cellular environment. A ligand is a protein molecule in the extra-cellular environment which interacts with the receptor. The ligand may be a freely circulating molecule or a molecule on the surface of another cell. The binding of an appropriate ligand to an appropriate receptor produces signals inside the cell which lead to whatever function(s) are supposed to happen. In a living system there are cascading chains of interactions involving potentially numerous extra- and intra-cellular molecules, cells, ligands and receptors. The term “pathway” is a general term referring to these cascading chains of interactions. Part of the work of immunologists is to unravel these interactions in the immune system, by seeking to understanding what interacts with what, and what happens when these interactions take place.

Everyone is aware that we have an immune system which fights disease using antibodies which can be adapted to specifically target an infection. That is the basis for vaccination. What is less well known is that antibodies are one of two distinct but related parts of the adaptive immune system which seek to target infections specifically. “Humoral” immunity refers to the first part and “cell-mediated” immunity is the other part. Humoral immunity is based on antibodies while cell-mediated immunity is based on T-cells. T-cells carry special T-cell receptors (TCRs) which specifically target infections.

A lymphocyte is a kind of cell found in the immune system. Lymphocytes are divided into B lymphocytes ("B-cells") and T lymphocytes ("T-cells"). The latter are further sub-divided into T helper (TH), T cytotoxic (TC) (also named cytotoxic T lymphocytes (CTLs)), and regulatory (Treg) cells. The primary role of B-cells is to produce antibodies. TH cells interact with B-cells, an interaction which is essential for the activation of B-cells to produce highly potent antibodies (which is why these T-cells are referred to as "helper" cells). The TH cells also produce cytokines, such as interferons (IFNs) and interleukins (ILs), including IFN gamma, IL-5 and IL-13, which activate other immune cells to attack infectious agents. TC cells destroy host cells which have become infected with viruses or other intracellular pathogens (which is why they are referred to as "cytotoxic", i.e. toxic to cells). The Tregs play a distinct role to control immunity and suppress or limit the untoward responses that could lead to autoimmunity by controlling the B-cell, TH and TC responses.

Annex 2 to this judgment contains Figure 10 from the primer, which summarises different aspects of the immune system. The phagocytosis aspect on the left of the diagram is not relevant. The operation of B-cells on the right hand side is also not important in this case. The relevant part is the operation of T-cells in the middle section.

An antigen is any substance, usually a protein or piece of a protein, that can be recognised by the adaptive arm of the immune system. This substance may derive from a micro-organism, an allergen or a foreign transplanted graft, or can be a component of a vaccine. In addition, in the case of auto-immunity, the antigen can be derived from a self tissue, i.e. a protein derived from the host.

Certain cells, such as monocytes, macrophages and dendritic cells, are known as antigen-presenting cells or APCs. APCs can take up antigen in various ways. Once inside the APC, the antigen is degraded, generally resulting in the formation of peptides. The peptides are bound to molecules known as major histocompatibility complex (MHC) proteins, shuttled to the cell surface of the APCs and "presented" to the T-cells of the adaptive immune system.

The T-cells recognise antigens presented by APCs via TCRs. This is also shown in section 3 of Figure 10 (Annex 2). Each T-cell expresses a particular TCR with a single specificity for an antigen, but the whole population of T-cells will include millions of cells with different TCRs and different antigen specificities that have been created by a genetic diversification mechanism. At the surface of the APC, MHC molecules with a specific antigen peptide bound to them are specifically recognised by the antigen binding site on the TCR on the surface of the T-cell. It is also imperative that the T-cells receive a second, so-called co-stimulatory signal to fully activate the cells. T-cells that are yet to encounter their specific antigen are known as naïve T-cells and may exist for many years in this state. The activated T-cells expand and mature and become the source of memory T-cells poised to react to the specific antigen again should the individual be re-exposed to the pathogen. Once activated, the T-cells undergo a process of differentiation to become activated TH cells or activated TC cells.

When TC cells bind to virally infected cells, their TCRs become engaged leading to the death of the infected cell. TH cells are subdivided into TH1 cells (which activate phagocytosis) and TH2 cells (which activate B-cells, leading to humoral immunity).

Effector T-cells (TH and TC) are relatively short lived (days) but memory T-cells may live for many years and provide protection from reinfection with the same microbe. Memory T-cells, like naïve T-cells, can be readily activated to perform effector functions (e.g. cytokine secretion or cell-killing). Compared to naïve T-cells, memory cells respond more rapidly to antigens and require less or no co-stimulation to be activated.

Merck summarised its case on common general knowledge in a series of 37 propositions, supported in their final form by Prof Boussiotis and commented upon by Prof Rudd. I will refer to them as “CGK points”. It is convenient to set them out and deal with them in turn, grouping them into topics.

The first three CGK points relate to T-cell biology:

The next group of CGK points relates to the rationale for developing cancer treatments by focussing on the T-cell mediated immune response.

The next group of CGK points relate to receptors CD28 and CTLA-4 and their associated ligands B7-1 and B7-2.

The next group of CGK points relate to the PD-1 receptor and its ligands PD-L1 and PD-L2.

The final group are some miscellaneous CGK points.

Inevitably a series of propositions like this will have been crafted with a view to the issues arising in this case but provided that is kept in mind, it is a useful summary of the main areas of common general knowledge relevant to this dispute.

Some of the CGK points bear further explanation. Owing to their homology, proteins are often put into families (see CGK points 15, 25 and 28). The relevant families in this case were the B7 family of ligands and the CD28 family of receptors. The PD-1 receptor is a member of the CD28 receptor family and the ligands PD-L1 and PD-L2 are members of the B7 ligand family. In fact all these molecules are members of a much larger Ig (immunoglobulin) family or “superfamily” but nothing turns on that.

The primer contained a diagram (figure 14) which had been taken from Nishimura & Honjo (2001). The figure is:

Starting at the top, the diagram shows cells in which various ligands in the B7 family are expressed. The molecule PD-L1 was also called B7H1 in the literature. It is second from the right. It is expressed in APCs, B-cells and other tissues. The homology percentages given are relative to PD-L1 for ligands (and PD-1 for receptors). The diagram shows the molecular weights and numbers of amino acids (aa). PD-L1 and PD-L2 are shown to interact with receptor PD-1 by the arrows. The Ys in the intracellular region of PD-1 (that is the region below the thick yellow (grey) line) represent tyrosines in the protein chain which are thought to play a role in signal transduction. Known expression of PD-1 is stated as being on T-cells, B-cells and monocytes (M).

Looking at the left side of the diagram, the interactions between the two ligands B7-1 and B7-2 (aka CD80 and CD86) with two receptors CD28 and CTLA-4 are shown by the multiple arrows. The reason the CD28 and CTLA-4 molecules are each shown with two ovals is because they are dimeric. The other known ligand/receptor pair in this family is the ICOS-L ligand and ICOS receptor.

The row marked “function” contains a plus or a minus. Plus indicates co-stimulation of T-cells and minus indicates down regulation of T-cell activity.

As is mentioned above, it was well known that the recognition of most antigens by the antigen-receptors on T-cells when presented by APCs was insufficient to generate a productive immune response. A further co-stimulatory signal was required to fully activate the T-cells. At the priority date the best characterised co-stimulatory pathway was that of the co-stimulatory receptor CD28 and its ligands B7-1 and B7-2. CD28 increases the response of T-cells both in terms of proliferation and function. In contrast when the same ligands bind to CTLA-4, T-cell activation is inhibited. That is what the plus and minus under CD28 and CTLA-4 refer to.

The discovery of the CD28/CTLA-4/B7 system in the 1990s led to the idea that both positive (co-stimulatory) and negative (inhibitory) co-receptors influence the outcome of the T-cell response. Studies of the effect of anti-CTLA-4 antibodies on tumours in mice had been carried out in the mid 1990s.

Looking again at the right hand side of the diagram above, there is a minus under PD-1. That reflects the point that at the priority date it was known that this pathway had a negative effect, down-regulating T-cells. For example the experiments in PD-1 knockout mice (CGK point 26) published by Nishimura showed that mice lacking the PD-1 gene developed auto-immune diseases. The two different diseases mentioned (lupus like nephritis and dilated cardiomyopathy) arose in different strains of mice. The hypothesis said to explain this phenomenon is that activating PD-1 generates a signal which inhibits T-cells and plays a role in inhibiting T-cells with TCRs for self-antigens. Such self-antigen specific T-cells would, if allowed to be activated, go on and attack cells in the animals own body. Thus in mice with no PD-1 receptor, no such inhibitory signal is generated and their immune system goes ahead and activates these self-antigen specific T-cells, which go ahead and produce an auto-immune disease. This logic is closely related to the prior work done on CTLA-4 (see below).

The molecule PD-L1 was first identified under the name B7H1 in 1999 in a paper Dong et al (1999). The paper Freeman et al (2000) identified the molecule as a ligand of PD-1 which had an inhibitory effect and called it PD-L1. The Latchman paper in March 2001 (the cited prior art) first identified the ligand PD-L2 and reported it to be inhibitory.

Merck’s case, supported by Prof Boussiotis, is that as far as the skilled team were concerned at the priority date, the binding of PD-L1 or PD-L2 to PD-1 generated an inhibitory signal. Ono’s case, supported by Prof Rudd, is that matters are not that simple. Ono does not deny that the pathway was known to be inhibitory but it also points to reports of stimulation of T-cells by PD-L1 and PD-L2. The papers reporting a co-stimulatory signal for PD-L1 are Dong et al (1999) and Tamura (2001). The paper reporting a co-stimulatory signal for PD-L2 is Tseng et al (2001).

Merck’s case is that these co-stimulatory results, which related to the ligands, did not undermine the skilled team’s common general knowledge that the receptor PD-1 was inhibitory. This is important because the claimed invention relates to an antibody directed to the receptor not the ligand(s). Merck submits that the prevailing explanation for the co-stimulatory data was that there was probably another receptor for PD-L1/PD-L2 which was co-stimulatory. It had just not been found by the priority date. So, just as B7-1/B7-2 and CD28/CTLA4 make up a scheme with ligands for two receptors, one co-stimulatory and one inhibitory, so it seemed that PD-L1/PD-L2 and PD-1 may be part of a similar scheme. This prevailing explanation would not put off the skilled team from trying an experiment involving inhibition of PD-1 to turn off the inhibitory signal and treat cancer, just as the same rationale was employed for CTLA-4.

Ono submits that what Merck calls the prevailing explanation was only one of a number of possible explanations which were part of the common general knowledge at the priority date. Merck responds that while these other possible explanations were in the literature, the second receptor hypothesis was the prevalent theory and the others were not part of the common general knowledge.

This is one of the critical issues in the case and a great deal of cross-examination was, appropriately, devoted to it. In the end I was not persuaded that the state of the common general knowledge was in accordance with Merck’s submission, for the following reasons.

First, there were a number of credible articles published shortly before the priority date which published experiments showing co-stimulatory effects using the ligands PD-L1 and PD-L2. Merck characterised the co-stimulatory results as being from “one group” which I think is correct but the group was well respected, the papers were published in peer reviewed and respected journals and other workers were prepared to discuss them without criticism.

Second, important evidence comes from the discussions in the published scientific literature. There is no doubt that the literature universally refers to PD-1 as an inhibitory receptor but the co-stimulatory results associated with the ligands were clearly common general knowledge. There are papers before the priority date which note the discrepancy and suggest they could be explained by the existence of a second co-stimulatory receptor, without mentioning alternative hypotheses. Important examples are the review articles Carreno & Collins (2002) (see p43, top paragraph) and Pardoll (2002) (see Box 2 at p232). This supports Merck’s submission that the second receptor was the prevailing hypothesis or primary suggestion to explain the co-stimulatory effects of the ligands.

However there are other papers which propose alternative hypotheses too. A significant paper is Sharpe & Freeman (2002) which was published in February 2002. It is a review article by two respected authors in the field concerning the B7-CD28 family. The article recognises the inhibitory role of the PD-1 pathway in vivo and the evidence for it. It mentions expression of PD-L1 in placenta (which may suggest a role in immune suppression by the foetus), mentions unpublished observations that PD-L1 is expressed in some tumours and at the end suggests that blockade of PD-1 signals might enhance anti-tumour responses in a similar way to CTLA-4 blockade. However the article ends with the statement that in order to manipulate this pathway effectively, further studies are needed. The reason for this reference to further studies is at least in part because of the co-stimulatory results which have been seen. The article states: “The reasons for the contradictory results of studies with PD-1 ligands are not known.”

The Sharpe & Freeman article discusses the idea of a second parallel receptor which delivers a stimulatory signal as one possible explanation and includes a diagram at figure 1 which includes a possible second receptor and illustrates the analogy with CD28/CTLA-4. However the article also notes the existence of evidence that militates against that possibility and again refers to the need for further studies (p123, top left paragraph).

The article discusses an alternative explanation: that the PD-1 receptor itself, when activated, might be capable of sending either an inhibitory signal or a co-stimulatory signal in different circumstances (p123 LH side). In this context the paper discusses signal transduction pathways and draws a possible analogy between the PD-1 receptor and a receptor called CD150 which has been reported to transduce both positive and negative signals. The intracellular signalling involves molecules SHP-2, SHP-1 and SAP. The authors note that PD-1 has both an ITIM motif and an ITSM motif in its intracellular portion and CD150 also has an ITSM motif, which has been reported to be able to switch signalling. The level of knowledge in relation to signal transduction pathways which I have attributed to the skilled team would allow them to understand this hypothesis as it is articulated in Sharpe & Freeman.

A similar review article published in June 2002 (Greenwald et al (2002)) reviews the same field, including the PD-1 pathway. Its title is “Negative co-receptors on lymphocytes”. Despite its focus on inhibitory signals the article discusses the contradictory (i.e. co-stimulatory) results seen with PD-1 ligands and mentions both the possible second receptor and the alternative analogy with CD150 as potential explanations. No doubt one reason for this is that both Greenwald et al (2002) and Sharpe & Freeman (2002) have Dr Arlene Sharpe as a co-author.

The paper Tamura et al (2001) also addressed this issue and suggested that the different co-stimulation or inhibition effects may be due to another receptor apart from PD-1 or may depend on antigen strength (p1815 RH side bottom).

Merck submitted that the hypothesis based on an intracellular signalling analogy with CD150 was tentative at best, appeared shortly before the priority date, had no evidence to support it and could not be common general knowledge. It is true that there was no evidence to support the CD150 analogy, but there was no evidence to support any of the various hypotheses nor was there clear evidence against any of them. At the priority date they were all hypotheses.

Third, I turn to Prof Boussiotis’s evidence. In her reports Prof Boussiotis stated that it was established that PD-1 was inhibitory and that she was unaware of any contradictory reports relating to signalling with the PD-1 receptor. This evidence was on the footing that the evidence that the ligands PD-L1 and PD-L2 might have a co-stimulatory effect was not evidence that the PD-1 receptor did so. The evidence emphasised the difference between a ligand and a receptor and necessarily rejected the hypothesis that PD-1 could have both positive and negative effects by analogy with the signal transduction behaviour of CD150. On the other hand, the alternative receptor hypotheses would be consistent with Prof Boussiotis’s approach.

However Prof Boussiotis clearly disagreed with both hypotheses. She did not agree with the positive/negative signalling hypothesis but she also found the reasoning in support of the alternative receptor hypothesis to be unconvincing. Prof Boussiotis accepted in cross-examination that the reasons for the contradictory studies with PD-1 ligands were not known. Nevertheless, as Merck pointed out, Prof Boussiotis did maintain her view that it was clear that PD-1 was not a stimulatory receptor.

Merck emphasised that the evidence of a co-stimulatory effect was in vitro and not in vivo data and pointed out that Prof Boussiotis maintained in cross-examination that the data published in the papers which showed a co-stimulatory effect of the ligands did not link that effect to the PD-1 receptor in any way. In cross-examination Prof Boussiotis also referred to the net functional balance of the PD-1 receptor and said that whether the function of the receptor was exclusively inhibitory or also had a stimulatory function, there was no doubt the net functional balance was inhibitory. In this respect I think the professor was exaggerating the level of clarity in the field at the time. Since the explanation for the co-stimulatory results was not known, I do not accept that there can have been no doubt about the net functional balance of the PD-1 receptor in vivo.

Merck suggested that the cross-examination of Prof Boussiotis was deficient in that it ignored many of the papers which mention PD-1 as an inhibitory receptor and only went to isolated parts of the papers she was taken to, thereby failing to focus on the overall message. I do not accept that submission. As for ignoring papers, Merck has been able to make sure that I had my attention drawn to the papers it relies on which were not referred to in the cross-examination. As for isolated parts, Prof Boussiotis was a capable witness and was able to explain her views fully and clearly.

Fourth, I turn to Prof Rudd’s evidence. He agreed that the PD-1 receptor had an inhibitory role and that the contradictory reports did not undermine that inhibitory role, but his view was that the contradictory reports suggested that PD-1 might also have a co-stimulatory role.

Prof Rudd mentioned a further point about one strain of PD-1 knockout mice in Nishimura et al (2001) and in many of his answers in cross-examination on the PD-1 pathway issue he referred to a point on peripheral as opposed to central tolerance. Despite his evidence on this I was not convinced that Prof Rudd’s thinking about peripheral as opposed to central tolerance in the context of PD-1 reflected the thinking of the skilled team at the priority date. Nor will I place weight on a point Prof Rudd raised about depleting suppressor cells.

Although I have not placed weight on some of the points raised by Prof Rudd in cross-examination, overall Prof Rudd did not accept that the second receptor hypothesis was the prevailing hypothesis. His evidence, maintained in cross-examination, was that there were different theories to explain the contradictory results.

Fifth, Merck submitted that the evidence from the PD-1 knockout mice (the papers Nishimura et al (1999), Nishimura et al (2000), Nishimura et al (2001)) amounted to overwhelming evidence that the net in vivo biological effect of PD-1 was inhibitory. In Prof Boussiotis’s opinion the knockout mice studies showed that the loss of the PD-1 receptor resulted in the enhancement of the T-cell response, regardless of what the PD-L1 / PD-L2 ligands do or how they mediate their effects.

Merck submitted that Prof Rudd had no answer to the knockout mice point. I do not accept that submission. Prof Rudd’s answer was the same as the explanation he gave on other occasions. The PD-1 deficient mice showed that PD-1 had an inhibitory role but it could not be excluded that it might also have a co-stimulatory role.

I was not persuaded that the knockout mice papers represented overwhelming evidence in any case. If the knockout mice experiments, which started being published in 1999, effectively ruled out the possibility of the PD-1 receptor having a positive co-stimulatory effect in some circumstances in addition to its clear inhibitory effect, it is surprising that the review papers in 2002 are discussing hypotheses which include the possibility of positive signalling by the PD-1 receptor.

Sixth, Merck referred to the fact that the patent proceeds on the footing that the PD-1 receptor is inhibitory. That is correct but it does not advance the argument.

Seventh, Merck referred to Prof Boussiotis’s evidence that the results reported in Okazaki et al (2001) showed that PD-1 did not associate with SHIP and there was no evidence PD-1 associated with SAP. Without such an association, the analogy with CD150 would not be correct. There is no need to delve into the details of SAP and SHIP or the ITIM and ITSM motif to resolve this point. It boils down to the point I have already mentioned, i.e. that the analogy with CD150 was not experimentally proven. It does not disprove the analogy.

Eighth, although a number of papers published after the priority date were referred to, I am not satisfied that they can be relied on given the speed with which this field was developing at the time.

Overall, I infer the following from the evidence as a whole:

Merck relies on what was known about CTLA-4 in support of its obviousness case. Its case is that there is a close analogy between CTLA-4 and PD-1 (CGK point 27). The scientific facts relating to CTLA-4 on which the analogy is based are summarised in CGK points (11) to (22). In summary Merck’s argument is that CTLA-4 was known to be inhibitory, that CTLA-4 knockout mice suffered a lethal autoimmune disease, that anti-CTLA-4 antibodies were known to inhibit tumour growth in mice, were known to be in clinical trials in humans and, just before the priority date, successful clinical results in phase I studies were published in the Tchekmedyian and Davis abstracts. Merck submits that the first two steps in this chain had been established by the priority date for PD-1 and so the next step, testing anti-PD-1 antibodies to see if they inhibit tumour growth in mice, was obvious.

Both Prof Boussiotis and Prof Rudd were of the view that studies on CTLA-4 informed the direction of research into PD-1 but as I mentioned above, the debate was about the degree to which the work on CTLA-4 informed PD-1.

Taking the first step in the chain of reasoning, I have already addressed what was known about the PD-1 pathway. While CD28 was a co-stimulatory receptor for the same ligands which interacted with CTLA-4, there may or may not have been an alternative co-stimulatory receptor for the same ligands as PD-1.

Turning to the second step, Prof Boussiotis noted that the PD-1 knockout mice also suffered auto-immune disease just as did CTLA-4 knockout mice but Prof Rudd’s view was that the phenotype was materially different. CTLA-4 knockout mice suffered a severe and lethal auto-immune disease. They only lived for a short time. On the other hand PD-1 knockout mice experienced a much milder auto-immune disease. Prof Boussiotis accepted that the difference in phenotypes suggested that CTLA-4 was strongly inhibitory whereas PD-1 was not.

CTLA-4 informed the direction of research into PD-1 but while there are similarities between the two receptors there are also differences. Prof Rudd accepted that the similarities meant that success with blockade of CTLA-4 might apply to PD-1. I think that is a fair characterisation of the position.

The other dispute about CTLA-4 and the common general knowledge was about knowledge of the phase I clinical trials. This knowledge came from two posters presented at the May 2002 meeting of the American Society for Clinical Oncology (ASCO). They are the pleaded Tchekmedyian and Davis abstracts. Phase I studies are conducted in a very small number of patients (17 and 14 respectively) and are concerned with patient safety and pharmacokinetics. Both abstracts report positive clinical results in two patients.

Prof Boussiotis said that this observation of positive clinical results was significant and the results were immediately known to those in the field from the first day they were presented. Prof Boussiotis did not attend the meeting herself but explained she was made aware of the results by colleagues. I accept the results were significant. Although Dr Canetta suggested otherwise in his evidence, I was not persuaded. In an article in 2015 of which Dr Canetta was a co-author, there is a statement describing the phase I results as remarkable (Berman et al 2014).

However while I do not doubt Prof Boussiotis’s sincerity, I am not satisfied that the evidence is sufficient to establish that these clinical results were common general knowledge at the priority date, given the short time between the meeting and the priority date. It is asking too much of the recollection of one witness. The fact that an article in 2014 (Weber (2014)) puts “results of first phase I trial” on a timeline as a milestone in 2002 does not establish that the results in the abstracts were common general knowledge by 3^rd July 2002.

I accept that the existence of clinical trials into anti-CTLA-4 antibodies as a treatment for cancer was common general knowledge but that is a different point. No details were known about them.

Paragraphs [0001] and [0002] describe the technical field and refer to the invention as being the use of an anti-PD-1-antibody for cancer treatment. The background art is referred to at [0003]-[0010]. In particular, [0004] describes CTLA-4 and PD-1 as having been recently discovered as generators of negative (i.e. inhibitory) signals.

Paragraph [0007] describes the studies that had been published on PD-L1 before July 2002, mentioning the expression of PD-L1 on APCs and in tumour cell lines. It states that “PD-L1 is one of these molecules that induce the inhibitory signal by PD-1”. Paragraph [0008] refers to PD-L2 and (referring to the Latchman paper) states that the expression and function of PD-L2 have been reported to be almost the same as PD-L1.

The patent does not refer to the reports that PD-L1/PD-L2 might be co-stimulatory. There is no reason why it should, not least because the patent provides evidence of the utility of PD-1 blockade in treating cancer.

Paragraph [0009] outlines the concept that the inhibitory signals of PD-1 could control immune reactions to auto-antigens. Paragraph [0010] refers to the cited prior art Dana-Farber 557 and makes the point that the document lacks an in vivo cancer model.

The disclosure of the invention begins at [0011]. Paragraphs [0011] and [0012] set out the main finding of the inventors, namely that they have identified that certain substances (anti-PD-1 antibodies) which inhibit the immunosuppressive signal of PD-1 inhibit cancer proliferation.

Paragraphs [0013] to [0085] summarise methods of manufacture and administration of anti-PD-1 antibodies and anti-PD-1 antibody fragments for use in the invention. Throughout this disclosure other agents are mentioned, in particular anti-PD-L1 antibodies, but the claims are clearly limited to anti-PD-1.

The drawings and examples are addressed in paragraphs [0086] to [0129]. There are thirteen examples. The most important in the context of this case are examples 4 and 5.

Example 4 is at paragraph [0101] and figure 4 presents the relevant results, as follows:

The upper boxes compare tumour growth in wild type mice to which: (i) B16 melanoma cells have been transferred (left); and (ii) B16 melanoma cells transfected to express PD-L1 (B16/PD-L1 cells) have been transferred (right). Melanoma is a form of skin cancer. Each line on the graph shows tumour volume in an individual mouse over time. Fig 4 (upper boxes) shows no apparent difference in terms of tumour volume when the B16 or B16/PD-L1 cells are used. The significance of this model is what appears in the lower boxes. These compare tumour growth in transgenic mice expressing PD-1 (left) and PD-1 knockout (PD-1^-/-) mice (right) into which B16/PD-L1 cells have been transferred. Transgenic mice are genetically modified mice which have had their DNA manipulated in some way. Knockout mice are transgenic mice in which an individual gene has been manipulated so that the functional protein normally encoded by that gene cannot be expressed. So the mice on the left over-express PD-1 whereas the mice on the right cannot express PD-1.

These results show that the onset of tumours is earlier and the tumour volume is greater in the transgenic mice which over-express PD-1 than in the knockout mice. Also the tumour growth in the knockout mice (lower right) is less than that seen in the wild type (control) mice (upper right).

The result shown in example 4 is significant as it indicates that blocking PD-1 function inhibits tumour growth.

Example 5 is at paragraphs [0102]-[0103] and the results are in figure 5. Figs 5(B) and 5(C) show:

Example 5 shows the effect of an anti-PD-L1 antibody on transferred myeloma (J558L) cells in syngeneic (genetically identical) mice. Myeloma is a white blood cell cancer.

The first step is to establish that the myeloma cells show significant expression of PD-L1. Those results are shown in a figure 5(A) which is not reproduced in this judgment.

Figure 5(B) then compares tumour growth in syngeneic mice transplanted with J558L tumour cells and (left) with rat Ig as a control and (right) anti-PD-L1 antibodies. Figure 5(B) shows that the anti-PD-L1 antibody leads to reduced tumour volume relative to the control rat Ig.

Figure 5(C) compares tumour growth in wild type mice expressing PD-1 (PD-1^+/+) and in PD-1 deficient mice (there is a mistake in the figure which labels them PD-L1^-/- when it should be PD-1^-/-). Figure 5(C) shows that the tumour cells undergo growth in tumour volume in the PD-1^+/+ mice (on the left), but that the absence of PD-1 (i.e. PD-1^-/- mice) completely prevented or eliminated the tumour. That is why the right hand box is blank. This result is striking and underscores the importance of the inhibition of PD-1 function for tumour elimination. At paragraph [0103] lines 15-17 the patent states:

“The transplanted tumor cells proliferation was completely inhibited in PD-1-deficient mice to which J558 cells had been transplanted (figure. 5(c)). These results present that inhibition of PD-L1 or PD-1 is effective on cancer treatment.”

The results in examples 4 and 5 show that inhibition of PD-1 function reduces, or can even eliminate, tumour growth in a mammal for two different types of cancer.

There are further examples in the patent up to example 13 but they were not focussed upon because they are not present in the first priority document. Following the examples are the claims.

There was an issue as to whether unamended claims 1 and 3 relate to the treatment of any and all cancers in general. The answer is plainly that they do. The skilled reader would understand that the inventors had used the language of the unamended claims to mean that no cancers are excluded, or in other words the use of an anti-PD-1 inhibitory antibody to treat any cancer would be within claim 1. The claim is not limited to particular kinds of cancer and is not, for example, limited to cancers already thought to be immunogenic either at the priority date or later.

This question is of most significance in the context of sufficiency of disclosure, priority and plausibility. With an eye on these arguments, it is worth noting that cancers can be divided up in numerous ways. There is no simple list of types of cancer. For example although in some contexts one can refer simply to lung cancer, in other contexts one distinguishes between non-small cell lung cancer and other kinds of lung cancer. The fact that the skilled reader knows and understands this does not alter the conclusion on construction. The claim is as wide as possible.

Very few drugs work in every single patient to whom they are administered, for a variety of reasons. The fact that no skilled reader of the patent would expect this drug to work in every patient does not alter the point on claim construction either.

Merck pleaded an added matter objection on the basis that if, which was denied, Dana Farber 557 is not novelty destroying on the basis of an assertion about selection from lists which Merck contended was made by Ono, then the claims of the patent must also fall for the same reasons. The point was not mentioned in Merck’s written closing submissions although during Ono’s counsel’s closing speech Merck’s counsel made an observation which suggested it may not have gone away completely. It was mentioned in Merck’s closing speech and I will address it briefly.

The added matter objection is that the application as filed includes wider disclosures than claim 1. There are three points:- a wider range of targets: {PD-L1 or PD-L2 in addition to PD-1}; a wider range of treatments: {infection as well as cancer}; and a wider range of therapeutic substances: {all sorts of chemical compounds as well as antibodies and a broad spectrum of antibodies too}. Merck contends that Ono’s argument over Dana Farber 557 involves an argument that novelty can be derived by selecting from a similar kind of optional disclosure and so Ono cannot have it both ways. The claim cannot be novel over Dana Farber 557 and avoid adding matter.

Considering the added matter objection itself, I reject it. The application as filed contains a disclosure of the treatment of cancer with an inhibitory anti-PD-1 antibody in paragraph [0043] and also in claim 19 dependent on claims 18 and 13. It is true that both the paragraph and the claims also include disclosures of other things, such as an anti-PD-L1 antibody, but the presence of these other things in these paragraphs does not matter because there is an individualised disclosure of an inhibitory anti-PD-1 antibody as a cancer treatment. Furthermore the presence of other disclosures elsewhere in the document, such as to agents which are not antibodies and to treating infection, does not make any difference. To a skilled reader, one of the things disclosed in the application as filed is the idea of using an inhibitory anti-PD-1 antibody to treat cancer. A claim to such a thing does not involve added matter.

I deal with novelty over Dana Farber 557 below. I do not believe the conclusions there are inconsistent with this one.

An important part of Merck’s objections to the validity of the unamended claims are arguments about insufficiency, “Agrevo” obviousness (lack of a technical contribution) and loss of priority. These points are not identical although they all cover very similar territory. If priority from the first priority document is lost it is not in dispute that the claims are invalid in the light of the Iwai paper which was published after the priority date and before the filing date. The Iwai paper is the inventors' own paper and includes (at least) examples 4 and 5 of the first priority document. Ono’s position is that the unamended claims are entitled to priority and are neither insufficient nor Agrevo obvious.

Merck’s submissions on priority are:

Merck’s submissions on insufficiency and Agrevo obviousness are:

Ono’s answers to all these objections together are:

For the law on priority I can refer to the judgment of the Court of Appeal in Medimmune v Novartis [2013] RPC 27 at paragraphs 151-154 which emphasises three points. First, the requirement of Art 87(1) EPC for claiming priority for “the same invention” requires the skilled person to be able to derive the subject matter of the claim directly and unambiguously using common general knowledge from the priority document as a whole. Second the approach is not formulaic but is a matter of technical disclosure, explicit or implicit. Third, the important thing is not the consistory clause or claims of the priority document but whether the disclosure as a whole is enabling and effectively gives the skilled person what is in the claim.

As to enablement, the test is the same as for sufficiency. The relevant principles applicable to Swiss style claims and EPC 2000 claims were set out by the Court of Appeal in Regeneron v Genentech [2013] EWCA Civ 93, at paragraphs 95-103. The question (see paragraph 103, citing the EPO decision T609/02 Salk) is whether the patentee has shown, for example by appropriate experiments, that the product has an effect on a disease process so as to make the claimed therapeutic effect plausible. I held in Hospira v Genentech [2014] EWHC 1094 (Pat) at paragraph 149 that this reasoning, including the requirement for plausibility, was part of the law of priority despite an EPO decision apparently to the contrary. That judgment was affirmed on appeal ([2015] EWCA Civ 57) but the priority point was not argued. Neither side contended before me that the law was otherwise.

For the law on insufficiency, I can refer to the summary given by Kitchin J (as he then was) in paragraph 239 of his judgment at first instance in Human Genome Sciences v Eli Lilly, which was cited with approval by Sir Robin Jacob in the Court of Appeal on the second occasion when that case went to that Court at [2012] EWCA Civ 1185 at paragraph 11:

“239.

The specification must disclose the invention clearly enough and completely enough for it to be performed by a person skilled in the art. The key elements of this requirement which bear on the present case are these:

i)

the first step is to identify the invention and that is to be done by reading and construing the claims;

ii)

in the case of a product claim that means making or otherwise obtaining the product;

iii)

in the case of a process claim, it means working the process;

iv)

sufficiency of the disclosure must be assessed on the basis of the specification as a whole including the description and the claims;

v)

the disclosure is aimed at the skilled person who may use his common general knowledge to supplement the information contained in the specification;

vi)

the specification must be sufficient to allow the invention to be performed over the whole scope of the claim;

vii)

the specification must be sufficient to allow the invention to be so performed without undue burden.”

There are two common sets of circumstances which arise under the umbrella of insufficiency, they can be characterised as “undue burden” cases and “breadth of claim” cases, although they overlap as Kitchin J’s summary at sub-paragraphs (vi) and (vii) recognises.

The question of undue burden is plainly highly fact sensitive. Merck referred me to the Court of Appeal’s judgment in AHP v Novartis [2001] RPC 8 in which the court had to consider the sufficiency of a claim if it was construed as covering not only the compound in issue (rapamycin) but derivatives of that compound. The Court of Appeal held that such a claim would be insufficient despite the trial judge’s finding that the work involved in finding working derivatives would not impose an undue burden because there is a qualitative difference between work required to find out which derivatives work and the application of the teaching of the specification. Leaving the skilled person to ascertain by research what would work was not sufficient (see paragraphs 41-44). That case shows that the quality of any work required may well matter as well as its quantity.

As to breadth of claim, I can refer to paragraphs 100-101 of the judgment of Kitchin LJ in Regeneron v Genentech. There the learned judge emphasised that “it must therefore be possible to make a reasonable prediction that the invention will work with substantially everything falling within the scope of the claim, or, to put it another way, the assertion that the invention will work across the scope of the claim must be plausible or credible.” He also said “If it is possible to make such a prediction then it cannot be said the claim is insufficient simply because the patentee has not demonstrated the invention works in every case.”

Merck also cited a recent decision of the EPO Technical Board of Appeal in T1150/09 in which the board considered the sufficiency of a Swiss style claim to the manufacture of a cancer treatment and held on the facts of that case that the claim was insufficient because it was not possible to predict which cancers can be treated with the invention. Merck submitted the analogy with the present case was plain but that is only true up to a point. It depends on the facts.

The question of whether a patent or priority document makes something plausible has come up in the cases in the context of industrial application, sufficiency, priority, and obviousness. It is also argued to play a part in the novelty case (below). The term is not to be found in the legislation and as Arnold J said in Idenix v Gilead [2014] EWHC 3916 (Pat) as an aspect of patent law, it is a fairly new point.

The question of plausibility was considered by the Supreme Court in Human Genome Sciences v Eli Lilly [2011] UKSC 51 primarily in the context of Art 57 EPC (susceptible of industrial application) and sufficiency. The contrast drawn in that case was between “speculation” on one side and a “plausible” or “reasonably credible” claimed use or “educated guess” on the other, recognising that the line between these things can be difficult to discern (see Lord Neuberger at paragraphs 107 (viii) and paragraphs 120-123 esp. para 123). As Lord Hope put it at paragraph 149, agreeing with Jacob LJ in the court below, the sense that the word plausible conveys is that there must be some real reason for supposing that the statement is true. Lord Hope went on to state that the standard is not any higher than that.

Ono’s submission across the board was to maintain that in vivo test results, such as the mouse tumour model studies in patent examples 4 and 5, were what was needed to make the invention plausible. I put to counsel for Ono that the essence of the court’s judgment in HGS was to set a lower standard for plausibility, after all there were no in vivo test results in the patent in HGS but the invention was there found to be susceptible of industrial application and sufficient based on plausibility. Counsel did not accept that the appropriate standard on the facts of this case was as low as that. The forensic motive for this is clear. While a low standard might work to Ono’s advantage in the context of arguments about priority and sufficiency, in the context of novelty over Dana Farber 557 and Wyeth 499, one of Ono’s submissions is that the lack of such in vivo tumour results in those documents means that they do not deprive the claim of novelty because they did not make the treatment plausible.

Counsel for Ono submitted that the key difference between the facts in HGS and those in this case is in the nature of the claim. The relevant claims in HGS were all product claims of one sort or another, whereas the claims in this case are Swiss style (claim 1) or EPC 2000 (claim 3). Both kinds of claim in the Ono patent include the achievement of a therapeutic effect as a functional technical feature and, to be satisfied, require someone (perhaps the manufacturer or someone else) to possess an appropriate state of mind. The claims in HGS had no such functional technical features.

I accept Ono’s submission at least to this extent. Whenever one is considering plausibility it must be done in the context of the invention determined by properly construing the claim and one must keep in mind the particular legal objection which is under consideration. Moreover it is worth reminding oneself that “plausible” is not a term found in the relevant parts of either the EPC or the 1977 Patents Act. It has proved to be a useful concept in various factual situations but just because that has proved to be true in one case does not mean that everything said in that context applies in a very different context. There is no law of plausibility as such.

I described part of the rationale for a test based on plausibility in the context of the sufficiency of Swiss style claims and EPC 2000 claims in Hospira v Genentech (above) at paragraphs 57-64. What I was trying to explain there related to the fact that such claims have a mental aspect. I will not repeat the passage. I believe the explanation remains sound save to mention that the precise scope of the mental element in such claims has subsequently been considered in detail (see the various recent Warner Lambert v Actavis cases). The scope of the mental element makes no difference to the analysis. It is the existence of such a claimed feature which matters.

It seems to me that the point Ono has identified in the context of Swiss and EPC 2000 claims has most significance when one considers the specificity and the breadth of the material relied on to establish plausibility in whatever context it arises. Sometimes there is a difficulty because the material relied on does not relate with sufficient specificity to the agent claimed (for example: Idenix v Gilead at paragraphs 444-462 and Eli Lilly v Janssen Alzheimer Immunotherapy [2013] EWHC 1737 (Pat) paragraphs 259-271). On the other hand the material relied on must make the particular claimed use, whatever it is, plausible. In HGS it was found to be plausible that the product claimed would have some sort of therapeutic utility. At the level of individual diseases one could not say which might be treated but that did not matter because the claim was not so limited. For a purpose limited medical use claim, more specificity is likely to be required than was necessary in HGS but on the other hand, material which is too narrowly focussed may not support a wide claim. The principle applicable to purpose limited medical use claims must be that the material relied on to establish plausibility must be both sufficiently specific, and have a sufficient breadth of application, to fairly support the claim both in terms of the nature of the agent claimed to have an effect, and in terms of the effect claimed.

I can dispose of the priority objection analogous to the added matter argument first. It is wrong. The skilled reader would derive a disclosure of the idea of using an anti-PD-1 inhibitory antibody to treat cancer from the priority document, clearly and unambiguously. I refer to claims 1, 3, 6, and 8 at p2-3, objects 1, 3, 6 and 8 at p7-8 and the text at p9 ln6-17. As in the application filed which was considered above for added matter, the fact that other things are disclosed too does not take away from the individualised nature of the relevant disclosure.

Next I will consider priority from the point of view of plausibility and breadth of claim. The claim covers cancer generally. Ono’s argument is that the wide claim is enabled across its breadth by the disclosure of the priority document in the sense that the document makes a claim of that scope plausible.

The priority document includes examples 4 and 5. Although these involve using PD-1 knockout mice, the priority document explains that anti-PD-1 antibodies would be expected to have a similar effect (p16 ln7-17). To the extent it is a free standing objection rather than a point to be considered in the round, I reject the submission that priority is lost because there is no disclosure of an anti-PD-1 antibody being generated or tested in the priority document. The success of the knockout mice tumour model experiments along with the teaching in the priority document which I have referred to make it plausible that an anti-PD-1 antibody would have the same effect.

Ono emphasised that the results represent two in vivo examples of the effectiveness of PD-1 blockade in two different types of cancer cells, melanoma and myeloma. In relation to example 4, however Merck pointed out that the murine melanoma B16 cell line which was used had been transfected to express the ligand PD-L1. It contended that the rationale which would be understood from the disclosure of the priority document was that the anti-PD-1 therapy would work against tumours which expressed PD-1 ligands. It also pointed out that the priority document refers to tests on a myeloma cell line J558L to ensure that it expressed PD-L1. In addition to looking at the document as a whole, Merck referred specifically to a paragraph in at p16 ln26 (paragraph [0042] in an alternative translation) about PD-L1 expression. Finally, Merck argued that data relating to the untransfected murine B16 cell line, which did not express PD-L1, had been deliberately excluded from the patent. Prof Boussiotis explained that B16 (untransfected) was well known to be poorly immunogenic.

The paragraph in the priority document which Merck relies on does provide that the anti-cancer effect can be expected in tumour cells which express PD-L1 but the teaching is not limiting. The last sentence in the paragraph (using the alternative translation) expressly provides that the treatment is “more effective” on tumours substantially expressing PD-L1. That is the opposite of a limitation.

Even though the positive experimental results are all in cells which express PD-L1, I do not accept that the priority document only makes treatment of PD-L1 expressing cancer cells plausible. There are two reasons for this. First, the document explains the rationale for performing experiments on cells expressing PD-L1 at p13-14. It is because cytotoxic activity against PD-L1 expressing cells is lower than against cells not expressing PD-L1 and so the recovery in cytotoxic activity due to the substance being investigated can be measured more clearly. In other words the tests have been set up to make the result easier to see. The skilled person would follow that reasoning. Second, although there is plain logic in the proposition that a focus on blocking a receptor naturally leads one to think about where the ligand is expressed, the skilled person knows as a matter of common general knowledge that PD-L1 is present in lymph nodes. Coupled with the data in the priority document, that indicates that the ligand does not have to be expressed on the tumour for anti-PD-1 therapy to be potentially effective. The interaction between the receptor (PD-1) and the ligand could occur at the lymph nodes rather than the tumour, whereby the T-cells could recognise the foreign (tumour) antigen by indirect presentation at the lymph nodes. I accept Prof Rudd’s evidence on this second point. Overall therefore, the skilled person would not expect treatment to be limited to PD-L1 expressing tumour cells. The cancer treatment rendered plausible by the priority document is not so limited.

In reaching this conclusion I have also taken into account Merck’s submission about data excluded from the patent. The argument starts from a comparison between the content of the first priority document and the Iwai paper. The Iwai paper includes results for untransfected B16 cells (in fig 4C). The evidence is that untransfected B16 cells are poorly immunogenic. The results in Iwai show largely comparable tumours in PD ^+/+ mice and PD ^-/- mice. In other words no anti-tumour effect is seen in the knockout mice in that test. These results are not in the priority document or the later filed patent application. Merck invites the inference that they were deliberately excluded, the inventors having decided not to include the data which showed no effect on cells which did not express PD-L1 and were poorly immunogenic. Merck then turns to a report of a meeting between representatives of Ono and academics at Kyoto University. The report names two individuals who are named inventors as having been at the meeting. They are Prof Honjo and Mr Shibayama. The meeting was on 18^th June 2002, two weeks before the first priority date. The work which forms the basis for the invention is discussed, including the idea of using an inhibitor of the PD-1/PD-L interaction as an anticancer agent. The report includes the following sentence (in English translation):

“However it is likely difficult to talk about all of tumor treatments by the PD-1/PD-L1 pathway because PD-L1 is not highly expressed in all of tumor tissues.”

Merck submitted that this report corroborated its case, in that the quoted passage indicates that if Prof Rudd was right, and PD-L1 expression by the tumour was not required in order for PD-1 blockade to be useful, it does not explain why Prof Honjo felt the need to take the non-transfected B16 data out.

I reject this part of Merck’s case. I am not satisfied I can draw any inference on which weight could be placed about the opinions or motives of Prof Honjo or the other inventors. The evidence does not support it. None of them were called as witnesses at trial and there was no reason why they should have been. Prof Honjo is a claimant in this case but that does not mean he should have given evidence. If Merck had wanted to put questions to him there are means by which it could have done. Even if it is the case that the inventor’s subjective views at the time were that the likely utility of the anti-PD-1 treatment they were thinking about would be limited to PD-L1 expressing tumours, that is not relevant to the interpretation of the disclosure of the priority document, which is an objective process.

Merck’s alternative case was that plausibility was limited to cells expressing PD-L1 (or PD-L2) or which were otherwise known to be immunogenic. This was supported by Prof Boussiotis. In this context the term “immunogenic” refers to a tumour which is capable of being recognised by the immune system. In fact this was more or less common ground technically. Since anti-PD-1 antibodies treat the immune system rather than treating cancers directly, for a cancer to be amenable to anti-PD-1 therapy presupposes that the cancer cells can be recognised by the immune system, at least to some extent.

The question is whether this is enough to support claims with the breadth of claim 1. Merck submitted it was not because not all cancers were considered to be immunogenic at the priority date. Ono submitted that claim 1 was supported because a treatment effective to treat any cancer which was responsive to immunotherapy is something which would plausibly treat any and all cancers. All cancers are amenable to an immunological approach. There was no reason to think that the treatment would only be effective in cancers which were, at the priority date, already known to be immunogenic.

Merck maintained that the line had to be drawn at tumours which were, at the relevant date, already known to be or regarded to be immunogenic. It relied on Prof Boussiotis’s evidence that clinical experience did classify tumours into the two classes – immunogenic and non-immunogenic – and on her opinion that it was not possible to extrapolate beyond those tumours where either the tumours expressed the ligand or were known to be immunogenic. Beyond that to expect it to work for all tumours was not plausible.

However in my judgment immunogenicity was (and is) not regarded as a black and white question by the skilled person. There are degrees of immunogenicity and it can change. I find the following facts, which were supported by the testimony of Prof Boussiotis in cross-examination:

I also refer to a point explained in the evidence of Dr Canetta in re-examination, that any malignant tumour has the possibility to proliferate, invade, produce necrosis and therefore present or produce neoantigens and therefore be amenable to an immunological approach. (A neoantigen is a new antigen). This is the reason why, in Dr Canetta’s opinion, any cancer is amenable to immunological treatment.

Taking this evidence together I find that the thinking of a skilled person with their common general knowledge reading the document at the priority date would be as follows. Obviously an anti-PD-1 antibody cannot have an effect on a cancer if it is not recognised by the immune system but the skilled person would not think that the fact that a cancer had been regarded as unresponsive to existing immunotherapies was determinative of whether anti-PD-1 treatment would work and would not regard “known immunogenicity” as a useful category in this context. Given the successful results in the mouse models, it is plausible that anti-PD-1 therapy, which is focussed on the immune system, will have the capacity to treat any type of tumour because all tumours express antigens not found in normal cells.

I find that a skilled person reading the first priority document would be able to make a sound prediction that anti-PD-1 antibodies would work in cancer generally, irrespective of whether the tumour cells expressed PD-L1 or PD-L2. The idea that the anti-PD-1 therapy could treat cancers even if they were not hitherto regarded as immunogenic is a plausible one.

Accordingly based on what I have considered so far, the claims are entitled to priority from the first priority document. I will turn now to insufficiency, bearing in mind that the issues overlap. It is convenient to address two things: the three cancers for which an anti-PD-1 antibody is said to have no significant therapeutic effect, and then the issue of undue burden.

Prof Boussiotis expressed her opinion that published results showed that patients with colorectal cancer, prostate cancer and multiple myeloma did not benefit from PD-1 blockade. Dr Canetta did not agree. An important point which emerged is the difference between different measures of success. In the trials relied on by Prof Boussiotis the measure of objective response used was a 30% tumour shrinkage. This has been a standard measure of success in cancer treatment for a long time. It is true that in some clinical studies, by the 30% tumour shrinkage measure, no such objective response has been seen. However Dr Canetta explained that extended stabilisation of tumours had been found and that was something of undoubted benefit to patients. The significance of extended stabilisation of tumours is that although the tumour may not have shrunk, it seems to have stopped growing and may have a lesser tendency to metastasis. The result is better survival for the patient. Dr Canetta also maintained that anti-PD-1 antibodies (nivolumab and pembrolizumab) were in clinical trials for each of these three cancers.

It is manifest that extended stabilisation of tumours is a benefit to patients. Prof Boussiotis and Merck did not disagree. However Merck submitted that Dr Canetta’s view could not reflect the thinking of a skilled person in 2002. It was based on information produced many years later and after significant clinical experience not only of anti-PD-1 antibodies but also ipilimumab (anti-CTLA-4). The measure of success which a skilled person would use in 2002 was the objective response referred to by Prof Boussiotis (30% tumour shrinkage) and by that measure the drugs do not work for these three cancers.

In my judgment Ono is entitled to rely on the measure of success described by Dr Canetta even though it derives from work done many years after the priority date. Consider a cancer for which anti-PD-1 antibodies have been shown to have no effect by the measure 30% tumour shrinkage but have been shown to have statistically significant extended tumour stabilisation and better cancer survival. If Merck’s submission was correct then the sale of the antibodies indicated for that cancer would not infringe claims 1 or 3. The product would not be for “cancer treatment” because “cancer treatment” had to be defined as a 30% tumour shrinkage. I reject that submission. The claim is broadly worded. The skilled person would not think a particular measure of success was mandatory.

An issue between Prof Boussiotis and Dr Canetta was whether the evidence was sufficient to draw any definitive conclusion based on Dr Canetta’s measure of success today. Taking multiple myeloma first, Prof Boussiotis said that more studies had to be done. Merck referred to a recent letter in the journal Leukaemia which said that research into PD-1 therapy for multiple myeloma had failed. Dr Canetta relied on results of a report from Timmermann Armand and Lesokin (June 2015) which followed from the report relied on by Merck and reported a study of 27 patients, in which one had a complete response and 17 (63%) had stable disease. I find that anti-PD-1 therapy is not ineffective for multiple myeloma.

Turning to colorectal cancer, there is evidence of success in treating a subset of colorectal cancers which was identified in the course of clinical studies of anti-PD-1 antibodies. The subset is patients for whom the cancer has a particular mutation known as mismatch repair deficiency. This group (which Prof Boussiotis called a very, very small sub-group of colorectal cancer patients) has shown a 40% success rate with Keytruda (pembrolizumab) monotherapy. Apart from this sub-group, Merck maintained that anti-PD-1 had not been shown to work at all for colorectal patients. Dr Canetta maintained that clinical trials into colorectal cancer were continuing. That is true as a matter of fact albeit that the trials seem to have been running for an awfully long time and one might think that if success by any measure was apparent, it might have emerged by now.

The position of prostate cancer (strictly castration resistant prostate cancer) is similar to colorectal cancer but without the successful sub-population. Prof Boussiotis maintained there was no evidence of success by any measure and Dr Canetta maintained that clinical trials were continuing. As with colorectal cancer, the trials referred to seem to have been running for a very long time.

Standing back, the evidence today shows that anti-PD-1 therapies plainly work or are worth investigating in a very wide range of cancers. Leaving aside the three cancer types focussed upon as failures, the evidence shows that in addition to the clinical approvals for non-small cell lung cancer, melanoma and renal cancer, anti-PD-1 antibodies are being investigated for the following cancers: small cell lung cancer, bladder, head and neck, oesophageal, gastric, pancreatic, hepatocellular, breast, ovarian, cervical cancer and glioblastoma. They are also being investigated in hematologic malignancies such as acute leukaemia, chronic myeloid leukaemia, Hodgkin’s disease and Non-Hodgkin’s lymphoma. In addition, clinical investigations are on-going in childhood, adolescent and young adult malignancies.

Unsurprisingly the evidence bears out the idea that anti-PD-1 antibodies work better for some cancers than for others. On any view, based on the evidence today, they are not promising treatments for most colorectal cancers and for prostate cancer. It seems to me to be more likely than not that anti-PD-1 monotherapy does not work to treat most colorectal cancers and prostate cancer, and I find that it does not. Where does that leave the unamended claims?

The important points are the following. I am satisfied that to a skilled person reading the patent application when it was filed in 2003 (or the priority document in 2002), the document makes a soundly based and reasonable prediction that the therapy will work to treat cancer. It is a principle of general application. A fair level of generality at which to describe the invention is as a treatment “for cancer”. It is plausible that it is likely to work for cancer in general. Success in this context does not mean success in every patient in all circumstances, no treatment will achieve that. Moreover referring to a cancer by the organ concerned (such as colorectal or lung) does not mean all such cancers are the same, for example in the evidence in this case there is a distinction between “small cell” and “non-small cell” lung cancer; another example is the evidence about the specific subgroup of colorectal cancers. The skilled person would not predict 100% success across the board.

Turning to the evidence of what has happened in practice, it shows that the treatment is indeed effective to treat a range of cancers and may well treat many more. Anti-PD-1 monotherapy probably does not work to treat most colorectal cancers but there are colorectal cancers which it does treat. Anti-PD-1 monotherapy probably does not work to treat prostate cancer. I recognise that both prostate and colorectal cancers are significant diseases which afflict a large number of people. Nevertheless in my judgment neither point undermines the teaching of the patent. The generalisation was a fair one at the time and was supported by the disclosure. It was a major advance. Moreover with all the information available today, it is still a fair generalisation. If anti-PD-1 monotherapy did not work for a large variety of different kinds of cancer then that might be a different matter. But that is not this case. When anti-PD-1 antibodies have been shown to work or to be very much worth testing in such a wide variety of cancers, the generality of the principle disclosed is not undermined by the lack of success in these two instances.

Furthermore we simply do not know if combination therapy involving anti-PD-1 antibodies and some other drugs will achieve success in colorectal or prostate cancers. It is at least worth continuing to investigate them. Merck submitted that success as part of a combination therapy could not count in some way. I do not see why. They are within the claim.

Overall I find that across the scope of claims 1 or 3 the antibodies claimed do work for substantially all cancers.

Merck submitted that the unduly burdensome nature of the research programme required to put the invention into practice is illustrated by the fact that today, almost a decade after the start of clinical studies with nivolumab, research to find an efficacious PD-1 therapy even in the most important and lethal cancers is continuing.

I do not accept that submission, for the following reasons. First, no undue burden for a skilled person is involved in creating a suitable inhibitory anti-PD-1 antibody in the first place. Second, no undue burden is involved in finding an anti-PD-1 antibody which is useful in cancer treatment and which successfully treats many cancers. It is true that these steps all involve very considerable work and very substantial cost but they are not of a quality which undermines the sufficiency of the patent’s disclosure. It is true that clinical research into anti-PD-1 antibodies is continuing but ten years is not a long time in the context of this sort of clinical research. The fact that clinical trials in many cancers are continuing does not count against sufficiency.

As regards colorectal and prostate cancer, if the sufficiency of the claim depended on those two types of cancer being successfully treated then there is force in the point that success could only be achieved with an undue burden, but I have already found that success is not required.

In my judgment the wide scope of claims 1 and 3 is fairly supported by the disclosure in the patent. Those claims are sufficiently disclosed and are commensurate with the technical contribution of the patent. The same is true for the priority document and so the claims are entitled to priority.

The general law of novelty was settled by the House of Lords in Synthon v SmithKline Beecham [2006] RPC 10. For anticipation to be established there must be disclosure of the invention by the prior art and that disclosure must be enabling. The parties agreed that for medical use claims (Swiss style/EPC 2000) there must be an enabling disclosure of the same therapeutic effect in the prior art, since those claims derive their novelty from the intended medical use. For example in Warner Lambert v Actavis [2015] EWCA Civ 556 at paragraphs 120-121 Floyd LJ said this:

“The skilled person would understand that the claim in question owes its novelty to the discovery of the new therapeutic use of the medicament. This emerges from a number of the cases, for example see the passages from Eisai quoted at [26] and [27] in Actavis v Merck. As Jacob LJ said at the end of [27]:

‘the novelty of the process (i.e. use of X in the manufacture of a medicament for Y’ comes from the ‘new therapeutic use’.

Thus the skilled person would understand that the technical subject matter of the claim was concerned with the ultimate end use of the medicament, from which it derived its novelty.”

This reflects the fact that in medical use claims, the new use is a critical feature of the claim. The use provides the novelty to the invention since the substance itself, and its medical use in general are known. As Floyd J noted in Regeneron [2012] EWHC 657 (Pat), at paragraph 99:

“In the European Patent Office the view is taken that, with claims in either form, the actual achievement of the therapeutic effect is a functional technical feature of the claim, as opposed to a mere statement of purpose or intention.”

Thus Ono submitted that plausibility was an aspect of the law of novelty. When the draft judgment was prepared I attributed the same submission to Merck too, but then received notes on that point from the parties. Merck submitted it had not made the submission but Ono submitted that it had. No matter. In my judgment, there is no distinct requirement for plausibility in the law of novelty, over and above disclosure and enablement, but in a proper case plausibility is an aspect of enablement. In order to amount to an enabling disclosure of a medical use claim and thereby deprive the claim of novelty, the prior art has to make the therapeutic effect plausible.

Dana Farber 557 is an international patent application filed on 23^rd August 2000, published on 1^st March 2001 and claiming priority from two US priority filings in 1999. The inventors are Clive Wood and Gordon Freeman. The title of the application is “PD-1, a receptor for B7-4 and uses therefor”. In the document the ligand PD-L1 is referred to as B7-4.

Merck’s case starts by observing that the patent in suit acknowledges in paragraph [0010] that Dana Farber 557 discloses the manufacture of an anti-PD-1 antibody which inhibits the interaction between PD-L1 (called B7-4 in the document) and PD-1 and “discusses that PD-1 may be involved in immunoevasion by tumours”. The patent then states “However, said publication does not demonstrate or explicitly suggest the efficacy of said antibody against tumours in vivo.” Merck contends, supported by Prof Boussiotis, that in fact Dana Farber 557 does disclose the use of an anti-PD-1 antibody to treat a tumour and so on that basis the only issue as far as the novelty of the claims is concerned is plausibility. Merck’s case is that the experimental data in Dana Farber 557, which does include in vivo results albeit not an in vivo tumour model, together with the common general knowledge, is enough to render such a use plausible. The particular experiments relied on are examples 15 and 18. Thus the claims lack novelty.

Ono’s case, supported by Prof Rudd, is that Dana Farber 557 is a long, confusing and contradictory disclosure which repeatedly asserts that PD-L1 could trigger a co-stimulatory or an inhibitory response and contains examples that purport to show this (example 7 and 13 respectively). Dana-Farber 557 does not explain why these contradictory results occur, nor does it teach how to use them in order to provide a method of treatment in a predictable manner. It only provides one example of the application of its teachings in any in vivo disease model, and that is an autoimmune disease model for human multiple sclerosis. Thus Ono submits the document does not make it plausible that blocking PD-1 would be effective in the treatment of any cancer.

I will start with the Dana Farber 557 document itself. The background section briefly discusses how T-cells respond to foreign proteins, and the B7:CD28/CTLA-4 pathway and the significance of manipulating that pathway as something offering great potential to stimulate or suppress immune responses in humans. The summary of the invention section starts with the following:

“The present invention is based, at least in part, on the discovery that PD-1 is a receptor for B7-4 molecules expressed on antigen presenting cells. PD-1 transmits a negative signal to immune cells, similar to CTLA-4. B7-4 molecules are expressed on the surface of antigen presenting cells and provide a co-stimulatory signal to immune cells and can transmit downregulatory signals to immune cells, depending upon the molecule to which they bind. Thus, modulation of PD-1, B7-4, and/or the interaction between B7-4 and PD-1 results in modulation of the immune response.”

This passage refers to the PD-1 receptor as inhibitory and mentions both the possible co-stimulatory and inhibitory roles of PD-L1 (B7-4). The last words in the second sentence “depending upon the molecule to which they bind” seem to be a reference to the possibility of a second receptor for PD-L1. The remainder of the summary section lists out a series of consistory clauses referring to “aspects” and “embodiments” of the invention.

After a list of the figures, the detailed description section is a long section running from p8 to p113. The points arising on that section will be addressed in context. The 18 examples are described from p114 onwards. The examples are followed by the 28 claims and then the 27 figures.

A fair way to approach the disclosure of Dana Farber 557 in more detail is to focus on the examples.

Examples 1 to 5 describe the isolation and characterisation of the nucleic acids encoding for PD-L1. Example 6 examines the ability of COS cells expressing PD-L1 to bind to various T-cell receptors and antibodies.

Example 7 investigates the ability of PD-L1 to co-stimulate T-cell proliferation. Prof Rudd said that these results showed a clear role for PD-Ll in the co-stimulation of T-cells. Prof Boussiotis’s view was that the increase in T-cell proliferation that was observed was due to the experimental conditions used. I doubt that a skilled person would think that the experimental conditions were the explanation for these results. The results relate to the idea, which was part of the common general knowledge, that PD-L1 had a co-stimulatory role and this example would be seen as reflecting that role. It would not be dismissed.

There was an involved argument about whether the result shows a co-stimulatory role of the PD-1 receptor. This point really stands or falls with the arguments about the common general knowledge relating to co-stimulation. The skilled person would see that the authors of Dana Farber 557 were aware of the co-stimulation phenomenon and at least contemplated that a second receptor may be the explanation (at p9 ln22 – p10 ln2) but equally the skilled person would not regard the disclosure of Dana Farber 557 as laying the issue to rest.

There was a point that Prof Boussiotis was named as an inventor of a US patent in which the identical text of example 7 appears and so should not have criticised the result. I did not find that a useful line of questioning.

Example 8 describes the generation of murine anti-PD-L1 antibodies and their use in the detection of PD-Ll on various cell populations, including certain tumour cells. Example 9 describes the generation of fully human anti-PD-L1 antibodies in transgenic mice. Example 10 describes the generation of single chain Fv-like portions of antibodies to PD-1 or PD-L1. In example 11, mouse and human PD-L1 are shown to bind to mouse and human PD-1, but not to CTLA-4, CD28 or ICOS.

Example 12 shows that an increase in the number of COS cells expressing PD-L1 leads to a decrease in cytokine (IL-2) production. There was a minor debate whether the results in the relevant figure (fig 17) support the conclusion but nothing turns on that.

Example 13 investigates the effect of PD-L1 on T-cell proliferation and shows that PD-L1-Ig/anti CD3 coated beads had an inhibitory effect on T-cell activity. These results can be contrasted with those in example 7, when the presence of PD-L1 leads to T-cell proliferation.

Example 14 looks at the relationship between PD-1 and CD28 and provides results consistent with the idea that binding PD-L1 to PD-1 can cause down-regulation of the immune response.

The relationship between PD-1 and CD28 is further considered in example 15, the results of which are shown in figures 20 to 23. The conclusion drawn by the inventors from these experiments is that the PD-1/PD-Ll pathway has capacity to antagonise a strong B7/CD28 signal, at least when antigenic stimulation is weak or limiting.

Example 16 investigates the mechanism of the PD-1/PD-L1 pathway and the effects are due to inhibition of T-cell cycle progression or apoptosis. Prof Rudd’s view was that the results were consistent with the conclusion that one role of the PD-1/PD-L1 pathway is to inhibit T-cell progression.

The results of example 17 indicate that, although the antibody fragments are effective at inhibiting PD-1 binding to PD-L1, the intact antibody is significantly more effective.

So far the examples on Dana Farber 557 relate to experiments conducted in vitro, including in vitro cell culture. The final experiment, example 18, is different. It describes the administration of soluble PD-L1 Fc (a soluble PD-L1 fusion protein) in a mouse model for autoimmune disease known as experimental autoimmune encephalomyelitis (EAE). The results show that that administration of the PD-L1 fusion protein makes the effects of the autoimmune disease worse. There is an increase in the clinical signs of disease, represented by the mean clinical score, and a decrease in survival rates, in the presence of the agent. The example concludes with the statement that the findings “clearly demonstrate the immunoregulatory potential for the [soluble PD-L1 fusion] protein in vivo.”

Finally Dana Farber 557 has a series of claims, which include at claims 19-22 claims which refer to treating a condition which would benefit from upregulation of the immune response, such as a tumor, by administering an agent which inhibits the signalling via PD-1 in an immune cell.

Merck submitted that Dana Farber 557 clearly and unambiguously discloses all the features of claim 1, that is to say it discloses the use of a blocking antibody to PD-1 to up-regulate the immune system so as to treat, amongst other things, cancer. As a whole there is a disclosure from which it can be directly and unambiguously deduced that the antibody will have the claimed therapeutic effect. This is so even though there is no in vivo tumour model. The in vivo EAE model in example 18 demonstrates that blocking PD-1 upregulates the immune response in vivo and the data from example 15 is also a relevant predictor of in vivo activity.

Ono submitted that there was no clear and unambiguous disclosure of an anti-PD-1 antibody to treat cancer at all, because the document was too contradictory, and even if there was, the therapeutic effect was not rendered plausible by the document.

My conclusions on these issues are as follows. I am satisfied that as a matter of disclosure (rather than enablement), Dana Farber 557 discloses the idea of using an anti-PD-1 agent, which could be an anti-PD-1 antibody, for the treatment of cancer. An anti-PD-1 agent is not the only agent disclosed and cancer is not the only disease proposed to be treated but nevertheless there is an individualised disclosure of that combination in Dana Farber 557.

The issue is enablement. Dana Farber 557 is not only long, it hedges its bets. The skilled person reading it with their common general knowledge would sympathise with the authors in the sense that the document really just reflects the state of the reader’s common general knowledge. The fact that both upregulation and down regulation of the immune system are mentioned is not in itself contradictory since, if the co-stimulatory response and the inhibitory response could be controlled separately, it would in theory be possible to take either approach. Merck suggested that all this could be explained on the basis that PD-1 signalling can indeed be either stimulated to treat conditions which would benefit from down regulation of an immune response or inhibited to treat conditions which would benefit from upregulation. However the document does not contain a clear and unambiguous teaching of how to do these things. Merck’s argument involves putting to one side the co-stimulatory effect of PD-L1 as being relevant to the behaviour of the receptor PD-1, on the basis that the second receptor is the explanation for the co-stimulatory effect. I do not accept that the document unambiguously enables that approach.

Moreover, due to the use of a soluble PD-L1 fusion protein, as Prof Boussiotis accepted, it is not possible to tell whether the results in Example 18 are due to interaction only with a PD-1 receptor or with another receptor. They may be the result of the co-stimulatory effect of PD-L1 on another receptor. It is true that Dana Farber 557 includes a teaching that PD-L1 will bind PD-1 and inhibit its negative signalling, but that does not alter the fact that one cannot determine from the results whether that is what is going on in the example.

Ono submitted that the fact the EAE model involved using complete Freund’s adjuvant, which Prof Rudd described as having a massive non-specific activating effect, undermined its relevance to the plausibility of cancer treatment even further. Prof Boussiotis did not agree, on the basis that a clear difference between the control group and the treatment group was observed. I preferred Prof Boussiotis’s evidence on that point. Nor was I persuaded that points on a difference in environment and the deactivation of T-cells in example 18 which Ono took were of significant weight. Nevertheless the fact remains that the EAE model is a model for multiple sclerosis rather than cancer and that is a factor to take into account. It shows that the immune system can be upregulated in vivo but it is not concerned with whether that upregulation can have an effect on cancer.

Dana Farber 557 does not contain a cancer model studied using an agent which can be unambiguously identified as an anti-PD-1 agent. Merck relied on example 15 and on the common general knowledge itself, but I am not satisfied that either of these, or both of them, is enough to make up for the absence of such a model. Overall in my judgment the content of Dana Farber 557 is not specific enough in one way and not broad enough in another. It is not specific enough to render plausible the use of an anti-PD-1 agent for the relevant disease. On the other hand, while the content may be sufficiently broad to render plausible the idea of using an agent which acts on the PD-1 pathway in medicine generally, and in that sense is similar to the situation in HGS; the content is not broad enough in its application or, looked at another way, not specific enough for cancer, to render plausible the use of that agent in the treatment of cancer.

I find that Dana Farber 557 does not amount to an enabling disclosure of the use of an anti-PD-1 agent for the treatment of cancer. Claims 1 and 3 are novel over Dana Farber 557.

Wyeth 499 and Dana Farber 557 are closely related documents with closely related disclosures. Wyeth 499 contains some further experiment results in addition to those in Dana Farber 557. The only in vivo results in Wyeth 499 are the same as Dana Farber 557. The EAE experiment in Dana Farber 557 which is experiment 18 there becomes experiment 22 in Wyeth 499. The significant difference between the two documents is somewhat clearer reference to an anti-PD-1 antibody to be used to treat cancer. The key passage in Wyeth 499 for this purpose is at p85 ln20-26, as follows:

“In another application, upregulation or enhancement of a B7-4 co-stimulatory function is useful in the induction of tumor immunity. Endogenous expression of wild-type B7-4 on tumor cells inhibits the immune response against tumor cells. Accordingly, inhibition of the interaction between B7-4 on tumor cells and PD-1 is useful in inducing tumor immunity. In one embodiment a PD-1 antagonist (e.g. a non-activating antibody against PD-1 or B7-4 or a small molecule PD-1 or B7-4 antagonist) is administered to a subject having or at risk of a tumor.”

Merck submitted this satisfied the disclosure test for novelty and disclosed the subject matter of claim 1. I agree, however since I have accepted that the same disclosure exists in Dana Farber 557, little turns on it. Dana Farber 557 is not an enabling disclosure of the claimed invention. This text does not turn Wyeth 499 into an enabling disclosure if Dana Farber 557 itself is not.

To be valid an invention must involve an inventive step, which means it must not be obvious to a skilled person having regard to the state of the art (s1(1)(b) and s3 of the 1977 Act, Art 56 EPC). The structured approach to the assessment of obviousness was set out by the Court of Appeal in Pozzoli v BDMO [2007] EWCA Civ 588. In Conor v Angiotech [2008] UKHL 49 the House of Lords considered the issue of obviousness. There Lord Hoffmann (with whom the others of their Lordships agreed) approved the following statement of Kitchin J made in Generics v Lundbeck [2007] RPC 32:

A particular point arises in this case about “obvious to try” but I will address that in context.

In relation to plausibility in the context of obviousness, I can refer to the judgment of Floyd LJ in Generics v Yeda [2013] EWCA Civ 925 at paragraph 49. The point is that a technical effect which is not rendered plausible across the breadth of the claim cannot be taken into account in assessing obviousness. On the findings above relating to priority and sufficiency, that problem does not arise.

The Latchman paper was published in March 2001. As its title indicates, the paper discloses that PD-L2 is a second ligand for the receptor PD-1 and inhibits T-cell activation. The paper starts with a background section describing the B7 family, the CD28 and CTLA-4 receptors, PD-1 and PD-L1. Next there is a section entitled “A second B7 homolog that binds PD-1”. This describes how PD-L2 was identified by computer based sequence homology searches, discusses its structural similarity to PD- L1 and explains how it was identified as a ligand for PD-1. The section ends with tissue distribution data showing mRNA expression in human and murine tissue. The expression of PD-L2 is similar to PD-L1 with high expression in placenta and low expression in spleen, lymph nodes and thymus. There is expression in some tumour cell lines but not others.

The next section, entitled “PD-L2:PD-1 interaction inhibits TCR-mediated responses” looks at the role of the PD-L2:PD-1 pathway in T-cell activation. The results of an experiment in which proliferation of anti-CD3 activated T- cells was shown to decrease in the presence of PD-L1 and PD-L2 are shown in figure 4(a). Figure 4(b) presents the results of a different experiment using T-cells from DO11.10 transgenic mice. Those result show that cytokine production is reduced in the presence of PD-L2.

The next section, entitled “PD-L2 and PD-L1 can inhibit TCR-CD28 signals” seeks to compare the interplay between TCR-CD28 stimulatory signals on the one hand and PD-1:PD-L2 or PD-1:PD-L1 signals on the other. The results in figure 5(a) to (c) show inhibition of TCR and CD28 mediated responses by PD-L2:PD-1 and PD-L1:PD-1 interaction. The authors conclude that PD-1 engagement by PD-L2 or PD-L1 can down-regulate TCR-CD28-stimulation of cytokine production. The section ends with result which are said to suggest a mechanism whereby the PD-L:PD-1 pathways can attenuate weak antigen responses.

The final section reporting experimental work is “Mechanism of action of the PD-L-PD-1 pathway”. The results in figure 6(a) and 6(b) relate to T-cell cycle studies. Figure 6(a) suggests that PD-L2 leads to cell cycle arrest. Although the name “PD” (short for Programmed Death) refers to the process of apoptosis or programmed cell death, the results in figure 6(b) indicate that the mechanism by which the PD-L-PD-1 pathway operates in this context does not involve apoptosis. There was an argument about the significance of figure 6(c) but it is not germane.

The Discussion section considers the implications of the work in the paper, describing that the PD-1 knockout mice and the discovery of the ligands PD-L1 have stimulated interest in this immunoinhibitory pathway. Both PD-L1–PD-1 and PD-L2–PD-1 signals inhibit T-cell proliferation. The expression pattern of the ligands suggests that the PD-1 pathway may be critical for peripheral self tolerance and the PD-L–PD-1 pathway may inhibit activation of self-reactive T-cells. The relationship with CTLA-4 is referred to, making the point that they may have complementary roles and that there may be cross-talk or synergy between the two pathways. The final paragraph is as follows:

“Thus, we have described a subfamily of inhibitory molecules within the B7-CD28 family. The PD-L–PD-1 pathway may play a key role in the induction and/or maintenance of peripheral tolerance and autoimmune disease. The expression of PD-1 ligands in peripheral tissues suggests that this pathway may dampen inflammatory responses at these sites. It is worth noting that PD-L1 and PD-L2 mRNA expression are up-regulated in a variety of tumor cell lines. Cell surface expression of PD-L1 was confirmed in three of four human breast cancer lines examined. These findings give impetus to the investigation of whether PD-L expression on tumors attenuates anti-tumor responses as well as the role of the PD-1 pathway in the pathogenesis of human autoimmune disease. Because PD-L1 and PD-L2 can inhibit effector T cell proliferation and cytokine production, the PD-L–PD-1 pathway may be an attractive therapeutic target. Blocking the PD-1 pathway may enhance anti-tumor immunity, whereas stimulating this pathway may be useful for down-regulating ongoing immune responses in transplant rejection and autoimmune and allergic diseases.”

In the context of this case, this paragraph makes some significant points. The paragraph highlights the idea that expression of PD-1 ligands on tumours may attenuate anti-tumour responses, suggests that the PD-1 pathway may be an attractive therapeutic target, and proposes blockade of the PD-1 pathway as a way of enhancing anti-tumour immunity.

With that introduction to the Latchman paper I turn to consider obviousness. The skilled person (or team) and the common general knowledge have been considered already. In terms of identifying differences between the Latchman paper and the claimed invention, there is no description in the paper, explicitly or implicitly, of the use of an anti-PD-1 antibody which inhibits the immunosuppressive signal of PD-1 for the manufacture of a medicament for cancer treatment. There is no need to identify an inventive concept distinct from the words of the claims.

Merck submitted that the invention was obvious over the Latchman paper. Ono submitted that the invention was not obvious.

In paragraphs 13.3 and 13.10 of her first report, Prof Boussiotis expressed her opinion as follows:

“13.3

Together [the findings disclosed in Latchman] indicated that the PD-1/PD-L pathway was one of nature’s important mechanisms for tolerance induction. They would provide immediate interest and impetus for the skilled person to block inhibitory signaling of the PD-L1/PD-L2-PD-1 pathway in order to induce anti-tumor immunity, as proposed in the last paragraph of the discussion of the Latchman paper.

13.4

It would have been obvious for the skilled person to test the effects of blocking PD-1 mediated signaling in an in vivo tumor model such as a mouse model in which a tumor is implanted or in mice which are known to develop tumors spontaneously. I believe the skilled person would have used an antibody against murine PD-1 or one of the murine PD-1 ligands as the blocking agent and would test its effect on tumor growth in that model.

13.5

In my opinion, the skilled person would in the first instance, choose to test an agent that blocked the signaling of PD-1, as opposed to a blocking agent against the ligands, because Latchman had shown that some tumors express PD-L1 and some express PD-L2. A blocking agent against PD-1 would be expected to prevent inhibitory signaling from either PD-L1 or PD-L2 from mediating their inhibitory effects through PD-1 on T cells.

13.6

I consider that in a transplanted tumor model an obvious tumor cell line to use would have been the P815 mastocytoma cell line. The results shown in Figure 3c of Latchman are that murine PD-L1 and PD-L2 mRNA are both expressed by P815 tumor cells; P815 had been extensively studied before and was well known to be immunogenic; and previous studies have documented that immunogenic tumor cells are the ones amenable to immune-based approaches for growth inhibition. Together these reasons made it a good candidate for the test of improving anti-tumor immunity by the blockade of PD-1 mediated inhibition.

13.7

Such mouse models and tumor cell lines were commercially available and easy to purchase by the skilled person in July 2002.

13.8

Methods for generating monoclonal antibodies would have been well known to the skilled person in July 2002 although a significant amount of practical work would have been required to generate an anti-PD-1 monoclonal antibody. However, such work would not have been unusual and would have been well within the norm for developing a therapeutic antibody at that time. For example ….

13.9

Before carrying out the mouse tumor model test, the skilled person, using standard assays known before July 2002, would have confirmed that a generated anti-PD-1 antibody binds to the PD-1 protein, blocks the binding of the PD-1 ligands to PD-1 and increases T cell responses as a result of blocking PD-1 signaling.

13.10

I consider that the generation and testing of an anti-PD-1 antibody in a mouse tumor model would have been logical and obvious for the skilled person to do based on the information in Latchman. For these reasons, I do not believe that this can be regarded as inventive or innovative in the light of what Latchman describes along with what the skilled person would have known from their common general knowledge.

13.11

The experiments in Latchman teach the skilled person that PD-L1 and PD-L2 through PD-1 mediate a potent inhibitory signal on antigen-specific mediated T cell activation, cell cycle progression and cytokine production. Although these experiments were not done in the context of tumors, they provide a strong expectation that blocking PD-1 mediated signaling would enhance anti-tumor immunity. There are a number of additional factors which would have made the skilled person expect a positive outcome:

(1)

The PD-L1/PD-L2 mediated inhibition seen in Latchman was observed in the context of moderate to low intensity TCR signals. Tumors are in general weakly immunogenic, thereby their low intensity TCR signaling would have a high chance of being inhibited by PD-1 and therefore PD-1 blockade would be expected to have a beneficial effect on anti-tumor immunity by enhancing T cell responses.

(2)

Antibodies to CTLA-4, which was considered to be a paradigm for PD-1, had already demonstrated that up-regulation of the immune response as a result of blocking signals which suppress T cell functions provides anti-tumor effects in mouse models. This prior work would have been very encouraging to the skilled person.

(3)

CTLA-4’s ligands, B7-1 and B7-2, were known not to be expressed on tumors. They are expressed on APCs. PD-1 ligands are expressed both on APCs and on tumors themselves suggesting that the PD-1 pathway might have greater relative importance as an immunosuppressive pathway that compromises anti-tumor immunity. Therefore blocking of this pathway would have a very high chance to enhance T cell responses against tumors.”

Thus the obviousness case in brief is that it was obvious to carry out an experiment in which PD-1 is blocked using an anti-PD-1 antibody in a mouse tumour model; a suitable tumour cell line and suitable mouse model were both readily available and an anti-PD-1 antibody could be produced without difficulty. The argument is that the result of the test would be analogous to the experimental results in the first priority document and the results would tell the skilled person the same thing, i.e. that anti-PD-1 antibodies are useful to treat cancer (or, on Merck’s priority/insufficiency case, at least some cancers). Therefore, assuming Ono’s case that the claimed invention is entitled to priority and is sufficiently disclosed, it follows, contends Merck, that the invention is obvious.

Ono, rightly, did not dispute that suitable tumour cell lines, mouse models and anti-PD-1 antibodies could be made or procured if the skilled person wanted to undertake this test. An alternative experimental model which could have been used was PD-1 knockout mice but the argument is the same.

The point on which the issue of obviousness turns is about the expectations of the skilled person. Merck submitted that the skilled person would have a sufficient expectation of success to make the invention obvious; Ono submitted the skilled person would not.

The law in that regard has recently been reviewed by the Court of Appeal in Teva v Leo [2015] EWCA Civ 780. In that case Sir Robin Jacob explained that I had got it wrong at first instance when finding an invention was obvious because the “obvious to try” test requires a higher expectation of success than the one I had applied (judgment paragraph 32 and see also paragraph 21). Ono also referred to two judgments of Floyd J on this topic. First, Ono referred to LEO v Sandoz [2009] EWHC 996 (Pat) at paragraphs 98-100, which in turn refers to Lord Hoffmann’s phrase “a fair expectation of success” used in Conor v Angiotech [2007] RPC 20. Second Ono cited to Omnipharm v Merial [2011] EWHC 3393 at paragraph 92, which emphasises, in the context of obvious to try arguments, that there is one statutory question – was the invention obvious? Finally on the law on expectation of success I refer to the judgment of Kitchin LJ in the Court of Appeal in Regeneron at paragraphs 85-87.

Merck’s major reasons why the skilled person would have a sufficient expectation of success to make the invention obvious are summarised below. They were all maintained and supported by Prof Boussiotis in cross-examination. First, the common general knowledge that PD-1 was an inhibitory receptor. Second, the very encouraging analogy with CTLA-4. Third, the existence of two ligands for PD-1, which made it obvious to target the receptor rather than the ligands. Fourth, targeting the receptor would be even more obvious given that the co-stimulatory results relate to the activity of the ligands. Fifth, the likelihood of PD-1 blockade having anti-tumour activity is enhanced because Latchman describes inhibition of low intensity TCR signals and tumours are weakly immunogenic. Sixth, the data on tissue distribution of expression of the ligands reported in Latchman was encouraging. In fact the PD-1 pathway may have greater importance than CTLA-4 with respect to anti-tumour immunity because PD-1 ligands are known to be expressed on tumours, unlike CTLA-4’s ligands.

The first point does not accord with my findings on common general knowledge. There was more to the skilled person’s common general knowledge about PD-1 than that. As for the second point, the analogy with CTLA-4 was common general knowledge but that cannot be taken too far. I have considered that above. The third point is a good point. The fourth point again does not accord with my findings on common general knowledge. The fifth point is a fair one. The inhibition in Latchman of low intensity TCR signals would justify enhancing the skilled person’s expectations about manipulating the PD-1 pathway as a possible anti-tumour treatment. The factual premise on which the sixth point is based (tissue distribution) is correct. In as much as the ligands have been found to be expressed on tumours, that is a good point in Merck’s favour. On the other hand, the fact the ligands are expressed in placenta, while indicative of a role in self-tolerance, is not a major point.

Ono’s reasons to the contrary, supported by Prof Rudd, were the following. First, at the priority date cancer immunotherapy was still controversial and the treatments had had limited success. Second, a plethora of different immunotherapeutic strategies were being proposed. None were more likely to work than any others. Third, there were numerous contradictory reports in the literature about whether the PD-1/PD-L pathway had a co-stimulatory effect, an inhibitory effect or both and in what circumstances. Fourth, the evidence of both experts was that the results of the in vivo tumour mouse models in examples 4 and 5 of the first priority document was very exciting. Fifth, if the invention was obvious over Latchman it is surprising that no anti-tumour mouse models were reported in Latchman itself.

As to the first point, I have made findings about the state of cancer immunotherapy at the priority date already. On the second point, Prof Boussiotis agreed that there were numerous approaches available to a skilled person. When asked about this in the context of the various approaches mentioned in the Melief article, she agreed that in 2002 she could not have predicted what kind of results the clinical trials would show in patients. I have dealt with the third point under common general knowledge.

As to the fourth point, there was clear evidence that workers were excited by the publication of positive results for CTLA-4 blockade in mouse anti-tumour models. In other words despite the published earlier work on CTLA-4, the positive results in mouse tumour models still caused excitement. Ono submitted the same logic should apply to the results for PD-1 blockade in mouse anti-tumour models shown in examples 4 and 5 of the first priority document. Ono relied on the following passage of the cross-examination when Prof Boussiotis was being questioned about those results:

Ono argued that this evidence was inconsistent with Merck’s obviousness case, on the basis that Merck’s case was that the experimental results were predictable from the prior art (with the common general knowledge) whereas this excitement demonstrated that they were not. Merck did not agree, submitting that if that is what Ono had wished to establish with the witness then the point about predictability should have been put to her at this stage in the cross-examination. Merck argued that Prof Boussiotis had given evidence earlier in her cross-examination that success in the mouse models was indeed predictable from the prior art.

There is no doubt that in her evidence Prof Boussiotis did express her view that the anti-tumour mouse model data in the patent was predictable from the knockout mouse studies and the fact that PD-L1 was expressed on tumours (e.g. T1/141₁₇–149₂₃). However having heard all her evidence I will place less weight on her opinion about predictability than on her evidence about the results being “very, very exciting”. The former, while entirely genuine, was an opinion advanced for the purposes of this case. The latter came across to me as an unguarded and authentic expression of the thinking of someone working in this field.

On the question of obviousness and predictability, Merck also pointed out that Prof Rudd had accepted in cross-examination that on the assumption PD-1 was a negative receptor, an anti-PD-1 antibody would be obvious to use. Prof Rudd did indeed accept that (at T5/857-858) but he maintained he did not agree with the premise (that PD-1 is only generating a negative signal). This takes one back to the common general knowledge issue about the role of the PD-1 pathway and PD-1 receptor.

At another stage in the cross-examination, Prof Rudd gave answers which Merck contended showed that Prof Rudd agreed that there would be a reasonable expectation from the Latchman paper that PD-1 blockade would elicit an effect on tumour cells which expressed PD-L1. This testimony was at T5/786-788. However in re-examination Prof Rudd explained that he had been saying there was that there was a reasonable expectation from the patent, rather than from Latchman. Merck contended that I should not place reliance on the re-examination because: the testimony in cross-examination was clear, the question in re-examination was leading, because of a pause which was said to have occurred in Prof Rudd’s answer in re-examination, and because the only data in the patent referred to in cross-examination was data in Latchman. I do not agree. The cross-examination was not clear. I recall thinking at the time that the questioning shifted very quickly from a focus on the patent to a question focussing on what one learned from the prior art. It was not clear to me at the time that Prof Rudd had understood the question Mr Waugh was asking and during the cross-examination I had made a mental note to consider whether to clarify the witness’s answer later once the questioning was complete. That did not prove necessary because it was covered in re-examination. Given the nature of the cross-examination, the questions in re-examination were fair. I do not derive any assistance from the fact Prof Rudd took time to give his answer nor from the fact that the only patent data referred to in the questions came from the Latchman paragraph. The witness had the patent in mind.

Merck also relied on an answer given by Dr Canetta in cross-examination in which he was asked about what could be predicted from what was known in 2002, including Latchman, and said that it would have been expected that an anti-PD-1 antibody would be effective to treat cancers which express a PD-1 ligand or are responsive to immunotherapy. However Ono explained that Dr Canetta had not been called to deal with the prior art. His role was to answer Merck’s case that anti-PD-1 antibodies had been demonstrated not to be effective in some cancers. Accordingly, Ono submitted, Dr Canetta had not read the patent or the prior art and his answer had to be seen in that context. I accept Ono’s submission. Counsel are entitled to ask witnesses questions on any relevant topic but if Merck’s counsel had wished to elicit Dr Canetta’s view about what was predictable from the prior art, which was not something he had addressed in his written evidence, the witness ought to have been given a chance to consider the prior art properly.

I find that at the priority date the results in examples 4 and 5 of the first priority document would have been very exciting.

The fifth point taken by Ono was the submission that if the invention was obvious it is surprising that it had not been done before and it was surprising that no mouse model data was reported in the Latchman paper itself. It is correct that the first PD-1 knockout mice data and, along with it the thesis that PD-1 negatively regulated the immune response, were indeed published in 1999. That is about three years before the July 2002 priority date. Nevertheless I am not satisfied that I can draw an inference in Ono’s favour on this point which would bear any weight. The evidence is too uncertain and Ono could have called more detailed evidence on this topic had it chosen to. I accept that all the elements necessary to perform a test were available in the sense that they could be obtained commercially or made without undue difficulty, but the evidence did not focus in detail on how much work was actually required. Also the Latchman paper itself contains information which was not published in 1999 and which adds to the obviousness case directed to blockade of the receptor PD-1, such as the existence of a second ligand PD-L2. The fact a step was not taken when the second ligand was unknown does not help decide whether it was obvious once the existence of the molecule as a ligand for PD-1 was published. Plainly there has been extensive research into cancer treatments for years, but for the reasons given above, I will not place substantial weight on the fifth point.

When she was asked about this point Prof Boussiotis said that she had in fact done some work on anti-tumour mouse models but the publication of the example 4 and 5 data (in the Iwai paper) came out first and so she never published any results. Ono submitted this evidence was unsatisfactory in that it had not been mentioned in her reports, there was no disclosure, and no detail about what Prof Boussiotis had actually done. Ono submitted the inference must be that Prof Boussiotis had done very little. Merck argued that if the cross-examiner had wanted to follow up and find out what Prof Boussiotis had done, the questions could have been asked and there was no basis on which to invite adverse inferences against the witness. I have dealt with my views on the witnesses already. I am now addressing the merits of the obviousness argument. I am not satisfied that I know enough about the work Prof Boussiotis undertook to place weight on it either way.

Standing back and considering the obviousness case over the Latchman paper overall, I find the position is as follows. The PD-1 pathway was known to have an inhibitory function. The analogy with CTLA-4 would have given encouragement to the skilled person. The existence of two ligands for PD-1 lends weight to the idea of blockading of the receptor PD-1 instead of the ligands. The low intensity TCR signals point and the fact that the ligands are expressed on tumours both add weight to Merck’s case. The mouse tumour model experiments would not be difficult to perform. A skilled person motivated with a sufficient expectation of success would conduct them without difficulty.

However set against all this are the following. First, the common general knowledge about the PD-1 receptor, the ligands PD-L1 and PD-L2 and their properties favours Ono. The position was not as simple as Merck contends. This is an important point. Second, albeit a point of lesser weight, the general position in the immunotherapy field was not as positive as Merck contends. There was a history of pessimism and failure which remained part of the thinking of the skilled person at the priority date. Third, I have found that the results of the in vivo tumour mouse models in the first priority document were exciting. Taking this along with the other evidence in the case, I infer that the excitement is an indication that there would not have been a fair or reasonable expectation of a positive result in such tests starting from prior art such as the Latchman paper.

A skilled person reading the Latchman paper in the light of their common general knowledge, might conceive of carrying out a mouse tumour model test of PD-1 blockade. However if they did think about performing the test, they would be doing so hoping that the test might succeed, not expecting that it would. There would not be a sufficient expectation that the outcome would be successful to render the invention obvious. I find that the invention involved an inventive step over the Latchman paper.

I have asked myself whether the conclusion would be any different if the results of the Tchekmedyian and Davis abstracts were part of the common general knowledge of a skilled person reading the Latchman paper. It would not. At best they might give a further impetus to conceive of carrying out a mouse tumour model test by following the CTLA-4 approach but they do not make any material difference to the consideration about the expectations of success in a mouse tumour model test of PD-1 blockade.

Although in the obviousness argument both sides focussed on Dana Farber 557 first and then considered the Latchman paper second, I have dealt with them the other way round. That is because having reached the conclusions I reached about the disclosure of Dana Farber 557 in the context of novelty, it seemed to me to be worth considering the Latchman paper for obviousness on its own merits as it was likely to be the stronger of the two. Having reached the above conclusion over the Latchman paper and having made the findings I have made about the content of Dana Farber 557, in my judgment the obviousness case over Dana Farber 577 fails also.

The real issue over Dana Farber 557 is the question of expectation of success, just as it was over the Latchman paper. The major element in Dana Farber 557 which is not in the Latchman paper is the in vivo EAE experiment set out in example 18 (and the results in example 15 about IFN-gamma). I have held that this data does not make the invention plausible and I refer to that analysis above.

I am not satisfied that the skilled person given Dana Farber 557 would have a sufficient expectation of success testing an anti-PD-1 antibody in a mouse tumour model to render the invention obvious. The points in Ono’s favour which I have identified above in the context of the Latchman paper still apply with their full force. I reject the obviousness case over Dana Farber 557.

I can deal with the obviousness case based on these abstracts shortly. They are concerned with phase I results of clinical trials into an anti-CTLA-4 antibody for the treatment of cancer. The studies were in patients with malignant melanoma (Tchekmedyian) and hormone refractory prostate cancer (Davis). The abstracts report some positive clinical results even though they are concerned with safety and pharmacokinetics. Those clinical results would be seen as significant by the skilled team. However since the claimed invention relates to the use of an anti-PD-1 antibody rather than the use of an anti-CTLA-4 antibody I cannot see how the invention could be obvious even given the extensive knowledge of PD-1 which was in the common general knowledge and despite the analogy between CTLA-4 and PD-1. The way forward for a team given either abstract is to develop an anti-CTLA-4 antibody. The argument that a skilled team given the abstracts would change tack and embark on testing anti-PD-1 antibodies instead is tainted with hindsight.

This case is complex but my main reasons for reaching the conclusion that the patent is valid can be summarised briefly. They are as follows:

The patent is valid. I find in favour of Ono.