High Court declares patent for once-daily rivaroxaban invalid due to lack of inventive step.

APPROVED [2025] IEHC 357

THE HIGH COURT

COMMERCIAL – INTELLECTUAL PROPERTY & TECHNOLOGY

Record No.: 2024/2369P

BETWEEN:

SANDOZ AG and ROWEX LIMITED Plaintiffs

-and-

BAYER INTELLECTUAL PROPERTY GMBH Defendant

JUDGMENT of Mr Justice Rory Mulcahy delivered on 25 June 2025

Index

Introduction .............................................................................................................................. 3

Core concepts ........................................................................................................................... 5

The Patent ................................................................................................................................. 8

Agreed common general knowledge..................................................................................... 10

Assays .................................................................................................................................. 18

Statutory Framework ............................................................................................................ 20

Obviousness ............................................................................................................................ 21

The prior art ........................................................................................................................... 30

The Blood abstracts ............................................................................................................ 31

Abstract 3010 (“the Kubitza SAD abstract”) ................................................................... 31

NO REDACTION REQUIRED

Abstract 3004 (the MAD abstract) ................................................................................... 32

Abstract 3003 (The Harder abstract) ............................................................................... 32

The posters .......................................................................................................................... 33

The Kubitza Single Ascending Dose (SAD) Poster........................................................... 33 The Kubitza Multiple Ascending Dose (MAD) Poster ...................................................... 35

The Harder Poster ............................................................................................................ 36

The Perzborn Article .......................................................................................................... 40

The arguments ........................................................................................................................ 42

The evidence ........................................................................................................................... 45

The experts .......................................................................................................................... 47

Disputed CGK evidence ........................................................................................................ 59

Wide or narrow therapeutic range ....................................................................................... 61

Reliance on LMWHs as comparator .................................................................................... 62

Value of assays ..................................................................................................................... 63

Hip surgery low risk ............................................................................................................. 70

Half-life dosing..................................................................................................................... 72

Once daily versus twice daily ............................................................................................... 73

Foreign judgments ................................................................................................................. 74

The Board of Appeal of the EPO – 27 October 2021 ........................................................... 77

Norway – Sandoz v Bayer – 9 June 2023 ............................................................................ 79

The Netherlands – Sandoz v Bayer – 1 November 2023...................................................... 79

Sweden – Sandoz v Bayer – 2 January 2024 ....................................................................... 80

Germany – Hexal AG v Bayer – 3 January 2024 ............................................................... 80 The Netherlands – Teva v Bayer – 27 March 2024 ............................................................. 81

Lithuania – Sandoz v Bayer – 26 March 2025 .................................................................... 82

France – Sandoz v Bayer – 28 March 2024 ........................................................................ 82

Switzerland – Sandoz v Bayer – 17 January 2025 .............................................................. 83

UK proceedings – Sandoz v Bayer....................................................................................... 83

Non-EPC countries .............................................................................................................. 86

Assessment of foreign judgments ......................................................................................... 86

Approach to assessment of evidence .................................................................................... 87

Discussion................................................................................................................................ 89

Teaching of prior art .......................................................................................................... 93

The Kubitza posters .......................................................................................................... 93

A Phase II trial? ............................................................................................................... 98

The Harder poster........................................................................................................... 102

Preliminary conclusion ........................................................................................................ 114

Judicial comity ..................................................................................................................... 114

Secondary evidence .............................................................................................................. 115

Summary of conclusionsc .................................................................................................... 130

Introduction

1. This judgment concerns the plaintiffs’ challenge to the validity of Irish patent EP (IE)

1 845 961 (“the Patent”). The Patent is entitled “Treatment of Thromboembolic Disorders with Rivaroxaban”. Rivaroxaban is an anticoagulant drug, used for the prevention or treatment of thromboembolic disorders. The defendant (“Bayer”) owns the patent for rivaroxaban, which it produces under the tradename Xarelto. It has been enormously successful.

2. The Patent the subject of these proceedings is not the patent for the drug, which has been off patent since 1 April 2024, but rather is a patent for the dosing regimen for the drug, being once-daily dosing. The plaintiffs, Sandoz AG and Rowex Limited (hereafter “Sandoz”) are manufacturers of generic drugs with plans to produce, market, and sell a generic version of rivaroxaban.

3. Although Bayer has counterclaimed for infringement of the Patent, Sandoz’s response to the counterclaim is that the patent is invalid. Sandoz has not, in any event, yet sought to market its generic drug. Accordingly, the only issue in these proceedings is the Patent’s validity. Although other grounds for challenging the validity of the Patent were included in Sandoz’s initial claim, the only ground of challenge pursued at the hearing of the action was that the invention the subject matter of the patent was obvious and therefore not patentable.

4. As appears from the discussion below these proceedings concern just one front in the battle which is being fought over the Patent (or equivalent patents) in multiple jurisdictions and before The Boards of Appeal of the European Patent Office (“the TBA”). Sandoz has been the challenger to the Patent in some but not all of those jurisdictions. In some jurisdictions, the Patent has been declared invalid; in others, including before the TBA, its validity has been upheld. The relevance of, and weight to be afforded to, decisions reached in other jurisdictions is in dispute in these proceedings.

5. The Patent will expire on 19 January 2026. In light of the fact that the proceedings could be rendered moot by the passage of time, they were carefully case managed in the Commercial list of the High Court. The proceedings commenced by plenary summons on 9 May 2024 and were entered in the Commercial list on 13 May 2024. They were heard over fourteen days between 29 April 2025 and 23 May 2025. A further brief hearing was held on

18 June 2025 to address a discrete issue as discussed below. Each side called an expert clinician and an expert pharmacologist. All four experts provided a number of comprehensive written reports, and were subject to detailed cross-examination. No witness of fact was called by either side.

6. Pursuant to section 58 of the Patents Act 1992 as amended (“the Act”), a patent may be challenged on the ground that the subject matter of the patent is not patentable under the Act. Section 9 of the Act provides that an invention is patentable if, inter alia, it involves “an inventive step.” Per section 13 of the Act, an invention “shall be considered as involving an inventive step if, having regard to the state of the art, it is not obvious to a person skilled in the art.” Put otherwise, if an invention is obvious to a person skilled in the art, it will not be patentable.

7. In these proceedings, Sandoz claims that a once-daily dosing regimen of rivaroxaban was obvious having regard to the contents of the prior art, i.e. the state of the art as at the date of the Patent. A number of specific items of prior art are identified. Although, notably, all of the relevant prior art was produced by Bayer or parties associated with Bayer, Bayer contends that once-daily dosing was not an obvious dosing regimen to try and implement and that its inclusion in Phase II clinical trials involved an inventive step over the prior art.

8. As appears from the discussion below, development of novel anticoagulants presents particular challenges for drug companies. This is because, as counsel for Bayer pithily explained when questioning a witness “their mode of action is also the mode of toxicity.” (Transcript Day 3, p. 122). Put simply, with too little antithrombotic effect, an anticoagulant may not be effective in the treatment or prevention of thromboembolic disorders. However, too great an effect gives rise to a risk of bleeding. Accordingly, the identification of the appropriate dose for a new anticoagulant drug, and the appropriate frequency of dose is not straightforward and requires greater care than may be the case in other novel drugs.

9. In the prior art at issue in these proceedings, rivaroxaban is identified by the Bayer code, BAY 59-7939. During the hearing, it was frequently referred to simply as BAY 59, which is the descriptor I propose using for the purpose of this judgment.

Core concepts

10. In Merck, Sharp & Dohme v Clonmel Healthcare Limited [2022] IESC 11, at para 6, the Supreme Court observed that:

“The test for patentability requires a novel inventive step, sections 9 and 11 of the Patents Act 1992 and Article 54(1) EPC, meaning one which is not obvious and which does not form part of the state of the art as of the priority date, the day when the application for a patent is first filed, that step being a plausible contribution to a technical issue and thus which is capable of application in industry. National law conforms to the EPC; Glaxo Group Limited v Patents Act [2009] IEHC 277. What is inventive is judged from the point of view not of a reasonable and educated person but a skilled person or, most likely, a skilled team, who is or are acquainted with the prior art…”

11. This observation identifies a number of core concepts which might helpfully be explained before a detailed consideration of the issues in the proceedings.

12. What is obvious is assessed by reference to what would have been obvious to a person skilled in the art at the priority date. This skilled person or the skilled team is a hypothetical legal construct deemed to possess particular characteristics. In Inhale Therapeutic Systems Inc v Quadrant Healthcare plc [2002] RPC 419, Laddie J described them (at paragraph 47):

“The notional skilled person is assumed to have read and understood the contents of the prior art. However that does not mean that all prior art will be considered equally interesting. The notional skilled person is assumed to be interested in the field of technology covered by the patent in suit, but he is not assumed to know or suspect in advance of reading it that any particular piece of prior art has the answer to a problem he faces or is relevant to it. He comes to the prior art without any preconceptions and, in particular, without any expectation that it offers him a solution to any problem he has in mind. Some pieces of prior art will be much more interesting than others. A document directed at solving the particular problem at issue will be seized upon by the skilled addressee. Its very contents may suggest that it is a worthwhile starting point for further development. But the same may not be the case where a document comes, say, from a distant and unrelated field.”

13. The skilled person is not a paragon: they bring with them whatever preconceptions or prejudices a person skilled in their field would be expected to have. In Asahi Medical Co. Ltd v Macropharma (UK) Ltd [2002] EWCA Civ 466, Aldous LJ stated (at paragraph 21:

“I will come later to analyse the judge's reasoning, but must first make it clear that a

decision on obviousness does not require a conclusion as to whether or not the skilled person would be slightly, moderately or particularly interested in any document. The court has to adopt the mantle of the skilled person. That mantle will include the prejudices, preferences and attitudes that such persons had at the priority date.

Thereafter the court has to decide whether the step or steps from the prior art to the invention were obvious. That decision has to be taken without the invention in mind and through the eyes of the skilled person. Of course any prior art document relied on must be deemed to be read properly and in that sense with interest. To conclude otherwise would deprive the public of their right to make anything which is an obvious modification of a published document. By obvious I mean that which would be obvious to the skilled person.”

14. A textbook example of this type of preference, prejudice or attitude was provided in Dyson Appliances Ltd v Hoover Ltd [2002] RPC 22 concerning the invention of a bagless vacuum cleaner:

“[89] … The vacuum-cleaner industry was functionally deaf and blind to any technology which did not involve a replaceable bag. The fact that the handicap was entirely economically determined made it if anything more entrenched. The industrial perception of need was consequently, in the judge's happy coinage, bagridden. It is entirely in accordance with what we know about innovation that this commercial mindset will have played a part in setting the notional skilled addressee's mental horizon, making a true inventor of the individual who was able to lift his eyes above the horizon and see a bag-free machine.”

15. Critically, the skilled team is uninventive (see, Boehringer Ingelheim Pharma GmbH & Co. KG v Norton (Waterford) Ltd [2016] IECA 67 at paragraph 44).

16. Although both parties accepted that, for the purpose of the issues in these proceedings, the skilled team could be confined to the skilled clinician and the skilled pharmacologist, there was some dispute as to the precise division of functions between them.

17. The skilled team is imbued with common general knowledge. In Ranbaxy Laboratories v Warner Lambert Co [2009] 4 IR 584, Clarke J stated (at paragraph 25):

“The knowledge which will be attributed to the notional skilled addressee is the knowledge that any worker in the area concerned would be expected to have as part of their general knowledge. See for example General Tyre v. Firestone Tyre [1972] RPC

457 at 482.”

18. In Re Glaxo Group Ltd [2009] IEHC 277, Charleton J described common general knowledge in the following way (at paragraph 43):

“The common general knowledge is the technical background of the notional man in the art against which the prior art must be considered. This is not limited to material which he has memorised and has at the front of his mind. It includes all that material in the field in which he is working which he knows to exist, which he would refer to as a matter of course if he cannot remember it and which he understands is generally regarded as sufficiently reliable to use as a foundation for further work or to help understand the pleaded prior art. That does not mean that everything on the shelf which is capable of being referred to without difficulty is common general knowledge, nor does it mean that every word in a common textbook is either. In the case of a standard textbook, it is likely that all or most of the main text will be common general knowledge. In many cases, common general knowledge will include, or be reflected, in readily available trade literature which a person skilled in the art would be expected to have at his or her elbow and regard as basic reliable information.”

19. The parties produced an agreed common general knowledge document which, in its

introduction stated that it was being provided to the court only to the extent that it may be of practical assistance and “strictly subject to the evidence to be given at trial by the experts.”

The experts’ evidence largely accorded with what had been agreed – none of the witnesses expressly disavowed anything which the document suggested had been agreed. There were, unsurprisingly, some differences in emphasis during the giving of evidence.

20. The validity of the Patent falls to be assessed by what is known at the priority date. The priority date for the Patent was 31 January 2005. In these proceedings, reliance has been placed on documentation, some of which is contended to be confidential, received from Bayer by way of discovery, which pre-dates the Priority Date but was not in the public domain at that time. In addition, reference has been made to publications which post-date the Priority Date but which refer to the process by which the Patent was developed. This documentation was described as “secondary evidence” in the proceedings and its relevance is discussed below.

21. It might be helpful to summarise briefly some keys aspects of common general knowledge which are not in dispute to aid in understanding the Patent and the arguments made regarding whether it involved an inventive step. First, it is necessary to detail the contents of the Patent.

The Patent

22. The description in the Patent commences as follows:

“The present invention relates to the field of blood coagulation, more specifically it relates to a method of treating a thromboembolic disorder by administering a direct factor Xa inhibitor once-daily in oral dosage form to a patient in need thereof, wherein the factor Xa inhibitor has a plasma concentration half-life indicative of a bid or tid administration interval, e.g. of 10 hours or less.

23. The description continues:

“[0005] The anticoagulants, i.e. substances for inhibiting or preventing blood coagulation, which are known from the prior art have various, often severe disadvantages. Accordingly, in practice, an efficient treatment method or prophylaxis of thromboembolic disorders is very difficult and unsatisfactory.

… [0008] Recently, a novel therapeutic approach for the treatment and prophylaxis of thromboembolic disorders has been described. This novel therapeutic approach aims to inhibit factor Xa…

[0009] In general, oral application is the preferable route of administration of the drug, and a less frequent dose regimen is desirable. In particular, once-daily oral application is preferred due to favourable convenience for the patient and for compliance reasons.

However, this goal is sometimes difficult to achieve depending on the specific behaviour and properties of the drug substance, especially its plasma concentration half-life… … [0012] Surprisingly, it has now been found in patients at frequent medication that oncedaily oral administration of a direct factor Xa inhibitor with a plasma concentration half-life of 10 hours or less demonstrated efficacy when compared to standard therapy and at the same time was as effective as after twice daily (bid) administration.

[0013] The present invention relates to the use of an oral dosage form of a direct factor Xa inhibitor for the manufacture of a medicament for the treatment of a thromboembolic disorder administered once-daily for at least five consecutive days, wherein said inhibitor has a plasma concentration half-life of 10 hours or less when orally administered to a human patient.”

24. The patent describes a single experimental example, a clinical study of BAY 59 with the stated purpose of assessing the safety, tolerability, and efficacy of BAY 59 at different oral doses administered twice or once daily, compared with enoxaparin, a low molecular

weight heparin (LMWH), in the prevention of venous thromboembolism. The study was performed on 642 patients undergoing elective total hip replacement.

25. A table of efficacy results is presented followed by this summary:

“The above data clearly demonstrate the efficacy of od administration of [BAY 59], namely fewer occurrence of composite endpoint events, i.e. fewer cases of DVT, PE or death compared to untreated conditions, and in the range of standard therapy.

Furthermore, the od administration is surprisingly perfect in line with bid administration.”

26. In addition, a table of safety results is presented, again followed by a summary:

“The above data clearly demonstrate the safety of the administration of [BAY 59]. The occurrence of any major bleeding events is low, approximately in the range of standard therapy and again perfectly in line with results from bid administration.”

27. The Patent contains two claims only the first of which is relevant for present purposes:

“1. The use of a rapid release tablet of [BAY 59] for the manufacture of a medicament for the treatment of a thromboembolic disorder administered no more than once-daily for at least five consecutive days, wherein said compound has a plasma concentration half-life of 10 hours or less when orally administered to a human patient.”

Agreed common general knowledge

28. In healthy humans there is a homeostatic balance between procoagulant forces and anticoagulant forces. Under normal physiological conditions, intravascular blood should flow freely and not clot; it is only upon an injury occurring that the coagulation system rapidly and locally responds at the injured blood vessel to generate a clot.

29. Haemostasis is a physiological process that stops bleeding at the site of an injury to a blood vessel. This is done by the creation of a clot, or thrombus, consisting of fibrin, a fibrous protein which builds a scaffold that blocks blood from escaping from the wound.

30. Thrombosis is a pathological clot formulation resulting in the obstruction of blood flow through one or more components of the vascular system, including blood vessels and heart chambers, by a thrombus or clot. A thrombus that breaks loose and travels from one location in the body to another is called an embolus. Thromboembolism is the term used to describe the combination of thrombosis and embolism.

31. As at the priority date arterial and venous thrombosis were major causes of morbidity and mortality.

32. Venous thrombosis is the cause of deep vein thrombosis (DVT) and may lead to pulmonary embolism (PE). PE is a serious complication of DVT and remains an important cause of death in both surgical and medical patients, DVT was a well-known and troublesome complication in patients undergoing major surgery and in hospitalised medical patients at the priority date. Patients having major hip and knee surgery were a particularly high-risk population.

33. Atrial fibrillation is a disease in which the heart beats irregularly. The irregular heartbeat leads to turbulent blood flow, which results in the slowing or pooling of blood within the heart. The resulting stasis of that pooled blood can promote the formation of a clot that is similar in composition to a venous thrombus. This clot can break off and embolise to the brain (causing a stroke) or to elsewhere in the arterial system.

34. Both venous thrombi and thrombi caused by atrial fibrillation can be prevented by anticoagulant drugs. Anticoagulant drugs, producing an antithrombotic clinical effect, were used for the prevention (prophylaxis) and treatment of these conditions at the priority date.

35. Coagulation is the process by which a blood clot is formed. The coagulation cascade can be considered to be a series of activation steps of various enzymes, known as factors, involved in coagulation. The coagulation cascade is represented, in simplified form, by the below diagram.

36. The diagram illustrates two pathways to coagulation, the extrinsic and the intrinsic pathways each of which lead to activation of Factor X to Factor Xa (i.e. Factor X in its activated form). As indicated by the dotted lines in the diagram, some factors have multiple roles and some of the clotting factors act as cofactors.

37. Thrombin, also known as Factor IIa, is the terminal enzyme or factor in the coagulation cascade and is directly responsible for cleaving soluble fibrinogen into insoluble fibrin which is the major matrix protein of a clot.

38. There are two pathways that participate in thrombin generation. The tissue factor pathway (extrinsic) and the contact activation (intrinsic) pathway, which converge at factor

X. Once activated, factor Xa cleaves the pro-enzyme prothrombin (factor II), which produces thrombin (factor IIa).

39. At the priority date, new anticoagulant drugs targeting various enzymes in the coagulation cascade were being investigated. Given the central role of thrombin in the development of a thrombus, many strategies for preventing and treating thromboembolic events focused on inhibiting thrombin generation or blocking its activity. In this regard, Professor Wilkins observed in his evidence that thrombin was particularly close to clotting on the coagulation cascade. Professor Meibohm agreed that it was “right above fibrin formation” but stressed that it could not have been said at the priority date whether its proximity in a biological sense meant that it would have the desired clinical effect, antithrombosis rather than anticoagulation.

40. In January 2005 there were a number of commonly used anticoagulants in the market, each of which had certain shortcomings.

Warfarin

41. Warfarin, a Vitamin K antagonist, is orally administered and has been approved for medical use since the 1950s. Warfarin operates by blocking the recycling of vitamin K in the body, resulting in a vitamin K deficiency which reduces the production of normally functioning factors. Warfarin has a slow onset of action and has a narrow therapeutic range (discussed below) with the result that patients on warfarin must be monitored regularly.

Heparin

42. Heparins include unfractionated heparin (UFH) and low molecular weight heparin (LMWH). Heparin is an indirect inhibitor which requires antithrombin to produce its anticoagulant effect. Heparin is administered by injection, either by continuous intravenous infusion or subcutaneous injection. Like warfarin, it is usually monitored by a coagulation assay.

Low Molecular Weight Heparins

43. LMWHs, derived from heparin, produce a more predictable anticoagulant effect. Therefore, unlike heparin, the anticoagulant effect of a fixed dose of LMWH can be reasonably predicted, meaning that monitoring is not required and the patient does not need to stay in hospital to facilitate same. Like heparin, LMWH is administered by injection. In some countries, the approved dosing frequency for LMWHs was once-daily, in others twice daily. Enoxaparin was a leading LMWH at the priority date.

44. There were also a number of novel anti-coagulants in development at the priority date.

Ximelagatran

45. Ximelagatran is a thrombin (factor IIa) inhibitor. It was the first new direct oral anticoagulant to be evaluated in clinical trials. As at January 2005, ximelagatran had been evaluated in several phase II and phase III clinical studies and had been shown to be as effective as either LMWHs or warfarin, both active comparators. The success of ximelagatran in initial clinical studies was highly encouraging and supported the premise that the development of orally administered, safe and effective anticoagulants was an achievable goal. Ultimately, ximelagatran was not pursued due to issues with liver toxicity. However, as at 31 January 2005, the results had been highly promising and the release of the drug for clinical use was highly anticipated amongst thrombosis specialists.

Razaxaban (DPC 906)

46. Razaxaban (DPC 906) was an orally active, direct factor Xa inhibitor. At the priority date, razaxaban was being evaluated in phase II clinical testing in patients. It had been evaluated against enoxaparin (30 mg twice daily) in a phase II dose-finding study in patients who had elective knee arthroplasty. It was given at 25 mg, 50 mg, 75 mg and 100 mg doses twice daily. At the lowest dose there was a trend for improved efficacy versus enoxaparin with similar rates of major bleeding. The three higher dose groups were discontinued because of excessive major bleeding.

47. Pharmacodynamics (PD) is a study of the physiological effects of a drug on the body. Pharmacokinetics (PK) is a study of how a drug moves through the body including its absorption, distribution, metabolism and excretion (ADME). In simple terms, pharmacokinetics is the study of the effect of the body on a drug, pharmacodynamics the study of the drug’s effect on the body.

48. A key pharmacokinetic parameter is the half-life of a drug. The half-life of a drug is the time required for the plasma concentration of the drug to reduce by 50%. In seeking to determine the dosing regimen for a new drug in clinical trials, the starting point will be informed by half-life, with dosing typically once every one or two half-lives.

49. This is subject to any additional information known about the drug and its intended effect. There can sometimes be a dissociation between the effect of a drug and its plasma levels, which may mean that the drug can be dosed less frequently than every one or two half-lives if the duration of effect lasts beyond the duration of measurable drug in circulation (assuming an appropriate assay is used).

50. LMWHs provide an example of a drug-type which is dosed less than once every one or two half-lives.

51. The therapeutic range or therapeutic window refers to the range of plasma concentration in which a drug produces the desired therapeutic effect with minimal adverse effects. It may be described in qualitative terms such as wide or narrow. The lower boundary of the therapeutic range is the minimum plasma concentration required to have clinical efficacy, that is, the dose below which the drug will have a subtherapeutic effect. The upper limit is the maximum plasma concentration which does not have a toxic effect, i.e. above which the drug will have a toxic effect. In the case of anticoagulants, the lower limit is defined by the minimum amount of drug in the system necessary to have the desired antithrombotic effect, whereas the upper limit is the maximum amount of drug in the system which does not produce excessive bleeding.

52. Where a drug is administered in repeated doses, if the dosing interval is the same as or shorter than the time required for elimination, the drug will accumulate in the body. The plasma concentration will steadily increase with each dose until it reaches a steady state at which point the rate of input will equal the rate of output and no further accumulation of the drug will occur. At steady state, if the drug continues to be administered, plasma levels of the drug will fluctuate around a mean concentration, the average steady state concentration. The peak drug concentration at steady state is known as the Cmax, the minimum (trough) drug concentration at steady state is known as the Cmin.

53. The following graph from Professor Meibohm’s written report illustrates two dosing regimens at steady state. Each involves the same frequency of dose, but a different overall dose. One achieves steady state within the therapeutic range, the other does not.

Figure 1 from Professor Meibohm’s Expert Report

54. The next graph, also from Professor Meibohm’s report shows two dosing regimens with the same overall dose, but dosed at different frequencies, one dosed three times daily, the other once daily.

Figure 2 from Professor Meibohm’s Expert Report

55. In that regard, it was agreed that it was preferable to ensure as little as possible fluctuation between peak and trough levels where seeking to ensure that the steady state remains within a narrow therapeutic range.

56. The ideal anticoagulant is described in the Note of Agreed CGK in the following terms:

The ideal anticoagulant is one that is effective in the prevention and treatment of thromboembolic events and that can be safely administered once or twice a day, using standardized dosing regimen, preferably in an oral formulation, with no need for routine coagulation monitoring. LMWH and other injectable anticoagulants were inconvenient for long term use and ease of administration, particularly in the home setting. Once-daily dosing was more desirable than twice-daily dosing from a patient convenience perspective, though efficacy and safety of a dosing regimen were the overriding considerations.

57. In these proceedings, the term ‘efficacy’ was used to refer to the desired antithrombotic clinical effect, whereas ‘safety’ was used in reference to the risk of bleeding.

58. The relative benefits of once versus twice-daily dosing for the ideal anticoagulant was debated during the hearing as discussed below.

59. Drug development typically breaks down into several readily identifiable phases: preclinical; phase I; phase II; and phase III, much of which is subject to mandatory regulatory requirements.

60. Pre-clinical trials involve in vitro and in vivo (animal) testing in order to confirm safety, toxicity and sometimes proof of concept ahead of use in humans.

61. Phase 1 studies are conducted, usually, in a small number of healthy volunteers, typically male, to obtain data on pharmacokinetics, pharmacodynamics, safety and toxicology in healthy subjects. Such trials can provide evidence of the biological impacts of a novel drug, but necessarily cannot determine its clinical impact, since the patients do not have the condition for which the drug is to be administered.

62. Phase II trials are the first trials in patients. They are designed to test for efficacy and safety and to establish the dosing regimen to be used in Phase III trials. Phase II trials are sometimes split into sub-phases. Development of new anticoagulants typically started with studies on the prevention of VTE in patients who are having elective knee or hip replacement surgery.

63. Phase III trials involve much larger cohorts of patients being treated with the dosage and frequency indicated by the phase II trials.

64. It was accepted during the evidence that Phase I and Phase II trials could be iterative, and adapted in light of new information, but that this would not be the case with Phase III trials.

Assays

65. A number of assays and pre-clinical models of thrombosis and bleeding have been developed to measure the effect of anticoagulants on humans. A number of these assays, such as prothrombin time (PT) and activated partial thromboplastin time (APTT), were routine assays performed in a hospital laboratory to screen patients for coagulation disorders and to monitor warfarin (PT) and heparin (APTT) therapy (discussed below).

Prothrombin time or PT

66. PT is a screening test that measures the time necessary to generate fibrin by the extrinsic pathway after activation of factor VII by tissue factor. It measures the integrity of the extrinsic and common pathways and is sensitive to reductions in factors VII, V, X, prothrombin and fibrinogen. It is the test used to monitor the dosage of warfarin and is reported as an international normalized ratio (INR). The INR was developed before the Priority Date to address the serious problem of lack of comparability in PT required to prevent thrombosis. By January 2005, the INR had become a worldwide standard and had largely replaced warfarin monitoring by PT. The therapeutic range in the context of warfarin represents the levels of the PT expressed as an INR that measure above the lowest level required for efficacy and below the highest level where bleeding becomes a problem. This therapeutic range is one that is associated with both optimal efficacy and safety of warfarin.

The INR represents the ratio of the patient’s PT divided by a control PT value using the WHO standard thromboplastin reagent, to normalize the PT value obtained.

Activated partial thromboplastin time or APTT

67. APTT is a screening test that measures the time necessary to generate fibrin by the intrinsic pathway. The test is initiated by adding an activator of factor XII (such as kaolin) to citrated platelet-poor plasma, followed by phospholipid and calcium. These components activate the intrinsic pathway factors to form a fibrin clot. A normal clotting time is 30 to 40 seconds, depending on the reagents used. The test is prolonged by anticoagulants that inhibit any of the steps in the intrinsic or common pathways and is routinely used to monitor the anticoagulant effect of heparin in the plasma. APTT is also sensitive to reduced quantities of factors XII, IX, XI, VII, X, V, prothrombin and fibrinogen (all integral parts of the intrinsic and common pathways).

HepTest

68. The HepTest was developed to measure the levels of unfractionated heparin and LMWH because it is sensitive to inhibitors of factor Xa and thrombin. It can be adapted to measure direct factor Xa inhibitors. It was not routinely used for patient monitoring.

Factor Xa assay

69. Factor Xa assay is performed by adding a measured amount of factor Xa to a plasma sample from a subject who has taken a factor Xa inhibitor. These assays were only available in specialized laboratories, likely for use in the research context rather than routine patient monitoring.

70. In addition to the above, more specialized assays had also been developed by thrombosis researchers before the priority date to assess the relative effects of an anticoagulant on inhibition of thrombin formation in both platelet-poor and platelet-rich plasma. The Skilled Clinician would be able to look up particular assay methodologies in the literature and would understand the general principles underpinning them. These included the endogenous thrombin potential (ETP) assay, the platelet-induced thrombin generation time (PITT) assay and the platelet-induced clotting time (PICT) assays, all of which are referenced in the prior art, but which were not used in clinical practice.

71. None of the assays had been correlated with a clinical effect which achieved safety and efficacy in patients for direct Factor Xa inhibitors.

Statutory Framework

72. Ireland is a signatory to the European Patent Convention (“EPC”) and the Patents Act 1992, as amended (“the Act”), gives effect in national law to the EPC. Part II, Chapter I of the EPC sets out the requirements for patentability. These are reflected in this jurisdiction in the provisions of Part II, Chapter II of the Act. Article 52 of the EPC stipulates that the invention must be new and involve an inventive step.

73. The relevant provisions of the Act are set out in the following paragraphs.

74. Section 57 provides that any person may apply for the revocation of a patent.

57.—(1) Subject to the following provisions of this section and section 58 , any person may apply to the Court or the Controller for revocation of a patent.

75. Section 58 sets out the grounds for revocation. They include that the patent was not patentable under the Act.

58.—An application for revocation of a patent may be made only on the grounds that

(a) the subject-matter of the patent is not patentable under this Act…

76. Section 9 of the Act sets out what is patentable:

9.—(1) An invention shall be patentable under this Part if it is susceptible of industrial application, is new and involves an inventive step.

77. Section 11 defines what is ‘new’ for the purpose of the Act and what is meant by the state of the art:

11.—(1) An invention shall be considered to be new if it does not form part of the state of the art.

(2) The state of the art shall be held to comprise everything made available to the public (whether in the State or elsewhere) by means of a written or oral description, by use, or in any other way, before the date of filing of the patent application.

78. And finally, section 13 explains the concept of an inventive step:

13.—An invention shall be considered as involving an inventive step if, having regard to the state of the art, it is not obvious to a person skilled in the art, but if the state of the art also includes documents within the meaning of section 11 (3) those documents shall not be considered in deciding whether or not there has been an inventive step.

Obviousness

79. In Nichia Corp v Argos [2007] IP & T 943, the Court of Appeal of England and Wales observed as follows:

“[21] … For many years the evidence of the inventor and the efforts (or absence of them) of the objector has been held admissible. Sometimes it is the patentee who prays this in aid, sometimes the objector. Why is this so? I think the answer lies in the ultimate (Windsurfing/Pozzoli step (4)) question: is there an inventive step? The question involves a value judgment which takes into account a variety of factors. Some are purely technical, others not. That is why, for instance, an apparently simple technical step may be held inventive where it can be shown to have led to commercial success—and all the more so if there was a long-felt want. Again the age of the prior art may matter. 'If it was so obvious, why was it not done before?' is a powerful question for the patentee to ask when the cited art is old, and more so when it is old and well-known.

[22] Because obviousness is a multi-factorial question, it is impossible to devise a more detailed sort of question, suitable for all cases. Windsurfing (4) cannot be further refined. The statutory question is the statutory question and none other. Attempts in the past to try to devise a question have all been built around the circumstances of the particular case: see eg Graham J's discussion of the 'Cripps' question' in Olin

Mathieson Chemical Corp v Biorex Laboratories Ltd [1970] RPC 157 at 188–189.

Similarly, attempts to force all questions of obviousness into a 'problem-solution'

approach can lead to trouble, though often the test can be a helpful guide.”

80. In determining whether an invention is obvious, it can be helpful to consider whether it was “obvious to try”. In Medimmune v Novartis [2012] EWCA Civ 1234, Kitchin LJ provided some guidance:

“[90] One of the matters which it may be appropriate to take into account is whether it was obvious to try a particular route to an improved product or process. There may be no certainty of success but the skilled person might nevertheless assess the prospects of success as being sufficient to warrant a trial. In some circumstances this may be sufficient to render an invention obvious. On the other hand, there are areas of technology such as pharmaceuticals and biotechnology which are heavily dependent on research, and where workers are faced with many possible avenues to explore but have little idea if any one of them will prove fruitful. Nevertheless they do pursue them in the hope that they will find new and useful products. They plainly would not carry out this work if the prospects of success were so low as not to make them worthwhile.

But denial of patent protection in all such cases would act as a significant deterrent to research.

[91] For these reasons, the judgments of the courts in England and Wales and of the Boards of Appeal of the EPO often reveal an enquiry by the tribunal into whether it was obvious to pursue a particular approach with a reasonable or fair expectation of success as opposed to a hope to succeed. Whether a route has a reasonable or fair prospect of success will depend upon all the circumstances including an ability rationally to predict a successful outcome, how long the project may take, the extent to which the field is unexplored, the complexity or otherwise of any necessary experiments, whether such experiments can be performed by routine means and whether the skilled person will have to make a series of correct decisions along the way. Lord Hoffmann summarised the position in this way in Conor at 42:

“In the Court of Appeal, Jacob LJ dealt comprehensively with the question of when an invention could be considered obvious on the ground that it was obvious to try. He correctly summarised the authorities, starting with the judgment of Diplock LJ in Johns-Manville Corporation's Patent [1967] RPC 479, by saying that the notion of something being obvious to try was useful only in a case where there was a fair expectation of success. How much of an expectation would be needed depended on the particular facts of the case.”

[92] Moreover, whether a route is obvious to try is only one of many considerations which it may be appropriate for the court to take into account. In Generics (UK) Ltd v H Lundbeck [2008] EWCA Civ 311, [2008] RPC 437, at 24, 101 BMLR 52 and in Conor [2008] UKHL 49, [2008] 4 All ER 621, [2008] RPC 716 at 42, Lord Hoffmann approved this statement of principle which I made at first instance in Lundbeck:

“The question of obviousness must be considered on the facts of each case. The court must consider the weight to be attached to any particular factor in the light of all the relevant circumstances. These may include such matters as the motive to find a solution to the problem the patent addresses, the number and extent of the possible avenues of research, the effort involved in pursuing them and the expectation of success.”

[93] Ultimately the court has to evaluate all the relevant circumstances in order to answer a single and relatively simple question of fact: was it obvious to the skilled but unimaginative addressee to make a product or carry out a process falling within the claim. As Aldous LJ said in Norton Healthcare v Beecham Group plc (unreported, 19 June 1997):

“Each case depends upon the invention and the surrounding facts. No formula can be substituted for the words of the statute. In every case the court has to weigh up the evidence and decide whether the invention was obvious. This is the statutory task.”

81. The four-step approach traditionally applied in the UK to the question of obviousness has been applied in numerous cases in this jurisdiction and was acknowledged by the parties here as an appropriate way to assess the claim of obviousness in this case, though reference was also made, in particular by Bayer, to the problem/solution approach favoured in most

EPC jurisdictions and by the TBA. In Re Glaxo, Charleton J explained the four steps and also warned against the perils of hindsight:

“29. In Pozzoli S.P.A. v. B.D.M.O. S.A. [2007] F.S.R. 37, 872 at 879, Jacob L.J.

restated the traditional four step English test in a manner in which I regard as acceptable, as follows:

“1 (a) Identify the notional “person skilled in the art”;

(b) Identify the relevant common general knowledge of that person;

2. Identify the inventive concept of the claim in question or if that cannot readily be done, construe it.

3. Identify what, if any, differences exist between the matter cited as forming part of the “state of the art” and the inventive concept of the claim or the claim as construed;

4. Viewed without any knowledge of alleged invention as claimed, do these differences constitute steps which would have been obvious to the person skilled in the art or do they require any degree of invention?”

30. The test of obviousness cannot, as a matter of law or as correct reasoning, be approached with the benefit of hindsight. I warn myself about being misled by the simplicity of any solution into believing that it must have been obvious. If one falls into that error in looking at an inventive step, the concept of the common general knowledge becomes completely distorted. Rather, my task is to consider whether it would have been obvious to a person skilled in the art, and therefore having common general knowledge…. In the light of the common general knowledge, and the relevant cited prior art, can this be regarded as an invention?”

82. As observed by Bayer in its opening written submissions, the “objective of the structure for the obviousness assessment set out in Pozzoli … is to minimise the risk that hindsight will be applied in the assessment.”

83. Though there was little debate between the parties regarding the principles applicable to a challenge to a patent on the ground of obviousness – though, of course, significant issues emerged over how those principles should be applied – the relevance of the recent UK Supreme Court decision in Actavis Group PTC EHF v ICOS Corporation [2020] 1 All ER

213 was much debated. Sandoz has described the decision as an “exceptionally useful framework for analysis” and as having “particular resonance” for present purposes. Bayer by contrast expressed that they were at a loss to understand how the decision applied to the case being made by Sandoz, and argues that Actavis is concerned with a different type of obviousness challenge, one in which the challenger sought to contend that the purported invention could have been arrived at through routine trials alone. Given the emphasis placed on the decision, it bears considering in detail.

84. As here, Actavis concerned a challenge to a dosage patent for a drug, tadalafil, a drug for the treatment of sexual dysfunction. The drug was the subject of a patent, the Daugan Patent, which revealed the drug, but did not identify a dosing regimen.

85. Lord Hodge delivered the judgment of the Supreme Court. He summarised the issue neatly in the first paragraph of his judgment:

“A pharmaceutical company, which had acquired the patent, conducted extensive research into ascertaining the optimal dosage of the compound. It discovered a dose which not only was safe and effective but also, unexpectedly, could be administered in a new and beneficial manner, because of both the half-life of the compound and its minimal side effects at that dose. A number of generic drug manufacturers challenge the validity of the dosage patent on the basis that it involves no inventive step.”

86. In the High Court, Birss J had rejected the challenge, but this had been reversed in the Court of Appeal. Kitchin LJ concluded as follows:

“[152] … Drawing the threads together, I am satisfied that Mr Speck has made good his criticisms of the judge's reasoning. The judge has lost sight of the fact that, on his own findings, the claimed invention lies at the end of the familiar path through the routine pre-clinical and clinical trials' process. The skilled but non-inventive team would embark on that process with a reasonable expectation of success and in the course of it they would carry out Phase IIb dose ranging studies with the aim of finding out, among other things, the dose response relationship. It is very likely that in so doing they would test a dose of 5mg tadalafil per day and, if they did so, they would find that it is safe and efficacious.”

87. Lord Hodge summarised the central submission in the appeal as being:

“[50] … The question which the Court of Appeal should have asked was whether at the priority date, before the skilled team embarked on its investigation, it was obvious in the light of Daugan, and without knowledge of the alleged invention, that a 5mg per day dose of tadalafil would be a safe and effective treatment, with minimal side effects, for sexual dysfunction. The Court of Appeal erred by not adopting that approach and instead by holding that the invention was obvious, because the claimed invention was the product of a familiar and routine path of pre-clinical and clinical research, in which each step was likely to follow the outcome of the prior step. This erroneous approach failed to address the question whether it was obvious to try that low dose because there was a reasonable expectation of success at the outset.”

88. Having also summarised the response, he observed:

“[52] I am not persuaded that the law adopts the extreme position of either submission.

Lilly's approach would require the court to disregard the work which a skilled person would carry out after the priority date in order to implement the teaching of the Daugan patent. That approach, as Mr Speck submitted, is contrary to the basic scheme of patent law. Actavis's approach in its reliance on 'the skilled person's repertoire', in other words on what the skilled person could already do, cannot be a general test for obviousness as it would render irrelevant many of the factors to which the courts have had regard in the assessment of obviousness, some of which I mention below.”

89. And further:

“[59] The notional skilled person, while having the compendious knowledge of the state of the art which s 2(2) requires, has no inventive capacity. But that does not mean that the skilled person has no skill to take forward in an uninventive way the teaching of the prior art. In this case the notional skilled team comprises the clinical pharmacologist and the clinician specialising in urology (para [17] above). That notional team is treated as exercising the professional skills of its members in responding to the teaching of the Daugan patent. It follows that uninventive steps which the skilled team would take after the priority date to implement the Daugan patent are not excluded from consideration in assessing the obviousness of the alleged invention at the priority date.”

90. In an observation which presaged the dispute in this case, Lord Hodge stated that:

“[63] … Much of the interest and controversy which the Court of Appeal's judgment has generated relates to how people have understood or misunderstood the significance which that court has attached to the routine nature of the pre-clinical and clinical research which I have described.”

91. He then identified a number of factors to which regard could be had in determining the question of obviousness from paragraphs [65] to [73]. These include:

“[65] First, it is relevant to consider whether at the priority date something was ‘obvious to try’, in other words whether it was obvious to undertake a specific piece of research which had a reasonable or fair prospect of success…

[66] Secondly, it follows that the routine nature of the research and any established practice of following such research through to a particular point may be a relevant consideration which is weighed against the consideration that the claimed process or product was not obvious to try at the outset of a research programme…

[67] Thirdly, the burden and cost of the research programme is relevant. But the weight to be attached to this factor will vary depending on the particular circumstances…

[68] Fourthly, the necessity for and the nature of the value judgments which the skilled team would have in the course of a testing programme are relevant considerations as both the trial judge and the Court of Appeal held…

[69] Fifthly, the existence of alternative or multiple paths of research will often be an indicator that the invention contained in the claim or claims was not obvious…

[70] Sixthly, the motive of the skilled person is a relevant consideration…

[71] Seventhly, the fact that the results of research which the inventor actually carried out are unexpected or surprising is a relevant consideration as it may point to an inventive step, at least in so far as it suggests that a test was not obvious to try or otherwise the absence of a known target of the research which would make it less likely that the skilled person would conduct a test…

[72] Eighthly, the courts have repeatedly emphasised that one must not use hindsight, which includes knowledge of the invention, in addressing the statutory question of obviousness…

[73] Ninthly, it is necessary to consider whether a feature of a claimed invention is an added benefit in a context in which the claimed innovation is obvious for another purpose…”

92. He also emphasised (at paragraph [78]) that particular considerations arise in relation to dosing patents.

93. Having considered these factors, the role of an appellate court, and the relevant foreign judgments, he concluded as follows:

“[105] … The task which the notional skilled team would undertake was that of implementing Daugan. The target of the skilled team would be to ascertain the appropriate dose, which would usually be the lowest effective dose. The skilled team would know of that target from the outset of its research. The pre-clinical and clinical tests involved familiar and routine procedures and normally progressed to the discovery of the dose-response relationship in Phase IIb. In this case the trial judge’s findings of what would have been the sequence of the tests, which did not depend upon hindsight, included the finding, which the evidence clearly justified, that the team, having found a therapeutic plateau, would be very likely to test lower doses and so come upon the dosage regime which is the subject matter of the patent. For the reasons which I have given above, I am satisfied that the Court of Appeal was entitled to interfere with the trial judge’s assessment of obviousness and to hold that the 181 patent was invalid for lacking an inventive step.”

94. In addition to seeking to distinguish this case on the basis that it concerned obviousness by routine trial rather than obviousness over prior art, Bayer also draws an important distinction having regard to the fact that Actavis involved a dosing regimen for a drug which treated sexual dysfunction, whereas BAY 59 is an anticoagulant. As observed at the outset of this judgment, the adverse consequence of providing too little effective drug to a patient during a trial of tadalafil is, it is readily apparent, of little concern when compared with the risk of underdosing in a patient requiring an effective anticoagulant. In an anticoagulant, unlike the drug at issue in Actavis, the skilled team must be concerned about both the lower and upper limits of the therapeutic range. Thus, a step which may be routine in the case of one type of drug may not be routine in the case of a novel anticoagulant.

95. Though I agree with Bayer that there are distinguishing features between Actavis and the instant case, these are differences which derive from the different factual matrix in Actavis not from any underlying difference in principle depending on whether an invention is said to be obvious because it would inevitably have been reached by conducting routine steps and one which is said to be obvious because the prior art reveals it to be so. Of course, the evidence required to succeed with each argument may be different, and there was no evidence in this case that once-daily dosing would have been reached solely by routine trials. The principles, however, remain the same. The judgment in Actavis, and in particular the identification of matters relevant to consider when seeking to assess obviousness, can be of assistance in addressing the question of obviousness here. But as the UK Supreme Court indicated, echoing the words of Charleton in Re Glaxo, no approach to the question of obviousness should distract from the statutory question.

96. Finally, as will be discussed, reliance was placed on so-called secondary evidence in these proceedings, being documentation created by Bayer during the drug development process. Although a dispute emerged regarding the status of the particular evidence relied on, there was no dispute between the parties regarding the principles applicable to secondary evidence.

97. The role of secondary evidence is described in Molnlycke AB v Proctor & Gamble Ltd [1994] RPC 49 at 113:

“In applying the statutory criterion and making these findings the court will almost invariably require the assistance of expert evidence. The primary evidence will be that of properly qualified expert witnesses who will say whether or not in their opinions the

relevant step would have been obvious to a skilled man having regard to the state of the art. All other evidence is secondary to that primary evidence…

In the nature of things, the expert witnesses and the court are considering the question of obviousness in the light of hindsight. It is this which may make the court's task difficult. What with hindsight, seems plain and obvious, often was not so seen at the time. It is for this reason that contemporary events can be of evidential assistance when testing the experts' primary evidence…

Secondary evidence of this type has its place and the importance, or weight, to be attached to it will vary from case to case. However, such evidence must be kept firmly in its place. It must not be permitted, by reason of its volume and complexity, to obscure the fact that it is no more than an aid in assessing the primary evidence.”

98. Both parties agree that Molnlycke represents the correct approach to such evidence and agree as to the status which can be afforded to the secondary evidence: it cannot provide the role of primary evidence where that evidence isn’t there, but it can be helpful in testing the plausibility (which seems a more apt description than ‘credibility’) of the expert evidence regarding what the skilled team would have done when presented with the prior art.

The prior art

99. Much of the prior art relied on in these proceedings emanated from the 45th Annual Meeting of the American Society of Haematology (ASH) held in San Diego in December 2003. Professor Hirsh explained in evidence the process for the conference, or conferences like it (Transcript Day 2, p. 118):

A. Yes. So, there are many, many presentations at the ASH conference and people who aspire to present will write an abstract which will be -- and the abstract, there's a word limitation and you've to got to write it within that word limitation.

And the abstract is then reviewed by reviewers, members of ASH, or members of the Committee for Haemostasis and I was one of them once. And you get lots of abstracts to review and you rate them. And basically, you rate them as something, abstracts that are really interesting and, as a result, they'll be included in state of the art lectures or plenary sessions. Abstracts that qualify for a ten minute presentation - usually ten minutes – and abstracts that qualify for poster presentation and abstracts that don't qualify for any of those but they're still in the abstract book. Basically, that's the process.

100. A book of abstracts was prepared for the ASH Conference, entitled, simply, ‘Blood’. The book indicates the nature, if any, of any presentation which will be made and the location of the presentation. If a poster presentation, it indicates the time at which the author of the poster will be present to answer questions in relation to it.

101. Three particular abstracts, Abstracts 3003, 3004 and 3010 (the “Blood Abstracts”) were relied on in these proceedings, all of which related to poster presentations. All of the abstracts are of similar length, approximately half a column of text, although abstract 3003 also includes a short table of test results. In addition to the abstracts, reliance was placed on the posters to which the abstracts related. Reliance was also placed on a journal article published in the Journal of Thrombosis and Haemostasis entitled ‘In vitro and in vivo studies of the novel antithrombotic agent BAY 59-7939 – an oral, direct Factor Xa inhibitor’ by Perzborn et al.

The Blood abstracts

Abstract 3010 (“the Kubitza SAD abstract”)

102. Abstract 3010 is entitled ‘Single Dose Escalation Study Investigating Pharmacodynamics, Safety, and Pharmacokinetics of BAY 59-7939 an Oral, Direct Factor Xa Inhibitor in Healthy Male Subjects’. Five authors are identified, all associated with Bayer. The first-mentioned author is Dagmar Kubitza, hence the SAD (and MAD) abstracts and posters have been identified by reference to her. I will refer to the studies the subject of those posters as the Kubitza studies.

103. This abstract says that BAY 59-7939 (“BAY 59”) is an innovative direct factor Xa inhibitor in development for the prevention and treatment of thromboembolic diseases. It describes a single ascending dose (SAD) trial with 103 healthy men in which full pharmacodynamics profiles were determined for 24 hours after drug administration. The results of this study are described. It is stated that the pharmacodynamic and pharmacokinetic time profiles show a close relationship, with a correlation coefficient between 0.6 and 0.9 depending on the parameters used. It refers to a terminal half-life of 3 to 4 hours. It concludes that BAY 59 is safe and well tolerated at a wide range of oral doses (1.25 – 80 mg) with predictable dose-dependent pharmacodynamics and pharmacokinetics. These data suggest that BAY 59 has predictable anticoagulation with an excellent safety profile.

Abstract 3004 (the MAD abstract)

104. Abstract 3004 describes a multiple-dose escalation study investigating the pharmacodynamics, safety and pharmacokinetics of BAY 59. It has the same authors as the SAD abstract. The abstract describes a multiple ascending dose (MAD) trial with 64 subjects receiving multiple oral doses of BAY 59 for five days. The abstract notes that relevant changes in the pharmacodynamic parameters were still present after 12 hours. Standard safety parameters were unaffected and no signs or symptoms of bleeding were observed across the dose range. It notes a terminal half-life of 4 to 6 hours. It concludes that oral administration of BAY 59 was safe and well tolerated in doses up to 30 mg bid.

Abstract 3003 (The Harder abstract)

105. Abstract 3003 is entitled ‘Effects of BAY 59-7939, an Oral Direct Factor Xa Inhibitor on Thrombin Generation in Healthy Volunteers’. It lists 10 authors. These include five of the six authors of the SAD and MAD abstracts, together with another associated with Bayer, three authors from the Institute of Clinical Pharmacology, University Hospital Frankfurt in Germany and Hans-Klaus Breddin of the International Institute of Thrombosis and Vascular Disease, Frankfurt in Germany. It was accepted in evidence that Professor Breddin was a world-renowned figure in the field of thrombosis. As with the Kubitza abstracts, the first mentioned author is Sebastian Harder, and the abstract, poster, and the study are therefore referred to by reference to him.

106. The abstract describes a placebo-controlled randomised crossover study in which 12 healthy volunteers received a single 5 mg or 30 mg dose of BAY 59. Three different assays were used to measure thrombin generation. The assays are described. A table of results is presented. The abstract records that a single 30 mg dose exerted a sustained effect in some assays of thrombin generation for up to 24 hours. Factor Xa was inhibited dose-dependently after administration of BAY 59. Maximum inhibition was observed two hours following drug administration. It was stated that there was a close correlation between BAY 59 plasma concentrations and inhibition of Factor Xa activity. It was stated in conclusion that the data suggests that BAY 59 effectively inhibits thrombin generation via both intrinsic and extrinsic pathways.

107. Each abstract is merely a summary of a poster presented at the ASH conference. The three posters corresponding to the blood abstracts all follow a similar template. They comprise five columns of text and tables and/or graphs divided into the following headings:

introduction; objective; subjects and methods; results; and conclusions.

The posters

The Kubitza Single Ascending Dose (SAD) Poster

108. This introduction to the Kubitza SAD Poster reads as follows:

Despite recent developments in antithrombotic therapy, thromboembolic events cause significant morbidity and mortality after surgery, in acute coronary syndromes, and in patients with atrial fibrillation (AF).

Current therapies, namely the low molecular weight heparins (LMWHs) and warfarin, have reduced the risk of major thromboembolic events, but significant unmet medical needs remain. Although oral warfarin effectively reduces the risk of stroke, particularly in patients with AF, it has several severe limitations, especially in the elderly. LMWHs are administered by subcutaneous injection, making them inconvenient for most patients outside the hospital setting and unsuitable for long-term use.

New orally available anticoagulants with novel mechanisms of action, including direct thrombin and Factor Xa (FXa) inhibitors, have potential for the treatment of thromboembolic disease. Here, we present the properties of BAY 59-7939, a new, oral, direct, Factor Xa inhibitor currently in development.

109. The objective of the study is described as being to evaluate the safety, tolerability, pharmacodynamic effects, and pharmacokinetic profile of BAY 59 after administration of single oral doses in healthy male subjects.

110. In the subjects and methods section of the poster, it is explained that single doses of BAY 59 or a placebo were administered in nine steps to healthy male subjects in doses ranging from 1.25 mg to 80 mg. Results are reported under headings ‘safety and tolerability’, ‘pharmacodynamics’ and ‘pharmacokinetics’. Under ‘safety and tolerability’ it is reported that 103 subjects completed the study and that overall instances of adverse events were similar in subjects receiving BAY 59 and in those given a placebo.

111. In the pharmacodynamics results it is stated that all pharmacodynamic parameters had similar dose-dependent time response curves. The maximum inhibitory effect on FXa activity was observed 1 – 4 hours after tablet administration and returned to the normal range within 24 hours for doses up to 40 mg. FXa activity remained elevated beyond 24 hours for the 60 mg and the 80 mg doses (as observed in the UK proceedings this appears to be an error and the poster should likely refer to FXa inhibition rather than activity). A graph, Figure 1, was included evidencing these results:

Figure 1 from Kubitza SAD Poster

112. Additional figures show the results of PT, aPTT and HepTest assays.

113. The pharmacokinetic properties are explained in the following way. Plasma concentrations increased dose dependently after administration of oral solutions and tablet doses of BAY 59 up to 10mg. It was less than dose proportional at higher doses, likely due to low drug solubility and incomplete absorption. Plasma concentration time profiles showed rapid absorption after administration of the solution. Maximum plasma concentrations were achieved after 30 minutes and half-life was estimated to be 3 to 4 hours.

114. Pharmacokinetic and pharmacodynamic effects are said to be closely correlated.

115. The conclusions of the poster are as follows:

• BAY 59 – 7939 was safe and well tolerated at all doses (1.25-80 mg) and formulations (oral tablet or solution) in this study of healthy male volunteers. No clinically relevant signs or symptoms of bleeding were observed.

• BAY 59 – 7939 was shown to selectively inhibit FXa without affecting FIIa or antithrombin III. Dose-dependent and predictable effect time profiles were observed for the pharmacodynamic parameters FXa inhibition, PT, aPTT, and HepTest. A close correlation was observed between the pharmacodynamic effects of plasma concentrations of BAY 59 – 7939.

• The good safety profile, selective inhibition of FXa, promising pharmacokinetic and pharmacodynamic profiles indicate that further investigation of BAY 59 – 7939 is warranted.

The Kubitza Multiple Ascending Dose (MAD) Poster

116. The Kubitza MAD Poster contains the same introduction as the Kubitza SAD Poster. Its objective is similar save that BAY 59 is to be tested in multiple doses. In this study, six dose steps of BAY 59 placebo tablets were performed at doses of 5 mg once daily, 5 mg twice daily, 5 mg three times daily, and 10 mg, 20 mg, and 30 mg twice daily in 68 healthy males.

117. BAY 59 is reported to be safe and well-tolerated. Maximum inhibition of FXa was reached 1 to 4 hours after active drug administration. Maximum PT prolongations compared with baseline were reached after 1 to 4 hours. After administration of 20 and 30 mg twicedaily doses trough levels had not returned to baseline after 12 hours. Similarly, trough values of aPTT at steady state were still elevated in comparison with baseline indicating the effect on aPTT was sustained over the dosing period of 12 hours until the second evening dose was administered.

118. Half-life is reported as between 3.7 and 5.8 hours and it is stated that there is a close correlation between pharmacokinetic and pharmacodynamic effects.

119. The conclusions of the Kubitza MAD Poster are as follows:

• BAY 59-7939 was safe and well tolerated after multiple-dose administration at all the doses tested without signs or symptoms of bleeding. BAY 59-7939 inhibited FXa activity and dose-dependently affected the pharmacodynamic parameters PT, aPTT, and HepTest.

• Predictable dose-dependent pharmacodynamics and pharmacokinetics were demonstrated, and indicate that BAY 59-7939 is suitable for twice-daily administration up to 30 mg.

• Close correlation existed between the prolongation of the PT and plasma concentrations.

• The good safety profile, selective inhibition of FXa, promising pharmacokinetic and pharmacodynamic profiles indicate further investigation of BAY 59-7939 is warranted.

The Harder Poster

120. The introduction to the Harder Poster reads as follows:

BAY 59-7939 is a selective, highly potent, direct Factor Xa (FXa)) inhibitor that is being developed for the prevention and treatment of thromboembolic disease. It has been

shown to be well tolerated at single and multiple doses up to 30 mg, and is rapidly absorbed after oral administration, with a terminal half-life of 9-12 hours.

Activation of FXa plays a pivotal role in the blood coagulation cascade by catalysing the conversion of prothrombin to thrombin. Therefore, direct inhibition of FXa would be expected to interrupt this coagulatory ; however, the extent of the effect of FXa inhibitors on thrombin generation, especially in platelet rich plasma, has not previously been characterised.

Assays for the measurement of platelet-derived thrombin generation in platelet-rich plasma (PRP) utilise either the cleavage of a chromogenic substrate by thrombin (endogenous thrombin potential [ETP]) or the aggregation and coagulation of PRP with normal CA 2+ content in the presence of low concentrations of hirudin (plateletinduced thrombin-generation time [PITT]).

121. It is stated that the aim of the study was to evaluate the effect of orally administered BAY 59 on thrombin generation in healthy male volunteers. The poster describes a single blind placebo-controlled crossover study with 12 healthy male subjects with the subjects receiving a 5 mg oral dose of BAY 59 or placebo on day one and a single 30 mg dose or placebo on day 14 or vice versa.

122. Three assay types, PITT, ETP and PICT are described. Under the heading statistical analyses, it is stated that the results were analysed using analysis of variance (ANOVA) and that based on these analyses point estimates (least squares means) and multiple comparisons were performed.

123. Results were reported for the ETP, the PITT and PICT assays. In respect of the ETP assay, it was stated that ETP (peak or AUC) was reduced significantly (compared with placebo profiles) by both the 5 mg and 30 mg doses of BAY 59 with maximum effect at 2 to 4 hours. Inhibition of ETP peak and ETP-AUC by 30 mg BAY 59 was sustained over 12 hours. This was illustrated in four separate graphs.

Figure 3 from Harder Poster

124. Both PITT and PICT assays are stated to be prolonged significantly by the 30 mg dose. The increase in PITT was sustained by 30 mg BAY 59 over 12 hours and PICT was prolonged over 12 hours after treatment with both a 5 mg and 30 dose.

125. Reference is made to the Kubitza posters. The opening paragraph refers to it having been shown that BAY 59 was safe and well tolerated up to doses of 30 mg. The poster also states that in agreement with other phase I data, being the data reported in the Kubitza posters, FXa was inhibited dose dependently after administration of BAY 59. Maximum inhibition was observed two hours after treatment. FXa inhibition correlated closely with ETP as demonstrated in a separate graph.

126. The conclusions were as follows:

• Orally administered BAY 59-7939 dose dependently inhibited both intrinsic (collagen) and extrinsic (tissue factor) pathways of thrombin generation.

• The effect of BAY 59-7939 on thrombin generation was demonstrated in platelet-free assays and in PRP-based assays. In contrast, indirect (i.e.

antithrombin III-dependent) FXa inhibitors obviously only inhibit FXa that is not protected by the platelet-prothrombinase complex.

• BAY 59-7939 not only inhibited the lag time of thrombin generation (PITT-Tc), but also had a profound effect on both the maximum extent of thrombin generation (ETP-peak) and the total amount of generated thrombin (ETP- AUC). This observation suggests an additional feature of BAY 59-7939, because weaker FXa inhibitors may only prolong lag-time without affecting the total amount of thrombin generation.

• Some parameters (e.g. ETP-peak) indicate a long-lasting pharmacodynamic effect of BAY 59-7939, which suggests suitability for a once-daily dosing regimen.

• The effects of oral BAY 59-7939 on intrinsic and extrinsic thrombin generation and PICT, which are mediated by direct inhibition of FXa, indicate that BAY 59- 7939 is a promising anticoagulant that merits further clinical investigation.

127. It is the fourth of these conclusions, the suggestion of suitability for a once-daily dosing regimen, which was the main focus of the debate between the experts and the parties.

The Perzborn Article

128. By contrast with the other prior art, the Perzborn Article (“Perzborn”) received comparatively little attention during the hearing. Its significance was reduced by Bayer’s prudent concession – solely for the purpose of these proceedings - that the skilled team would have been able to discover the chemical formula for BAY 59, which formula is disclosed in the Perzborn Article.

129. Perzborn is another scientist associated with Bayer. In addition to disclosing the chemical structure of BAY 59, the article describes pre-clinical tests carried out in vitro and in vivo, in rats and rabbits.

130. The article concludes:

In summary, BAY 59 is an oral, direct FXa inhibitor that inhibited thrombus formation in established rat and rabbit thrombosis models at doses that did not significantly increase bleeding times. The clinical relevance of these data needs to be investigated.

Based on its potency, selectivity and efficacy, BAY 59 may offer a safe and effective oral therapy for the prevention and treatment of arterial and venous thrombosis; BAY 59 is currently undergoing clinical evaluation.

131. Prior art of a type similar to the conference abstracts at issue here was discussed in Illumina Inc v TDL Genetics Ltd [2020] RPC 9, a decision of Arnold J from the Patents Court. Bayer refers to the following passage:

“133. Being a conference abstract, Ikeda is a very short document which is lacking in detail and in certain respects is unclear. The skilled person would also be aware that it had not been peer-reviewed. For these reasons, the skilled person would approach it with a degree of caution.

134. It is common ground that the skilled person is deemed to read a prior art document with care, and in that sense with interest, but not to assume that the document has any relevance to the problem the skilled person is addressing. This is an important point in the context of the present case. In these proceedings Ikeda has been subjected to sustained, detailed, and high-powered forensic examination. That is not how it would have been read by the skilled person in 2003. I shall return to this point below.”

132. Sandoz calls attention to the following observation from the same case:

“146. Two points follow from this. First, it is common ground that the authors do not state that they have discovered that foetal cell-free DNA is shorter than maternal cellfree DNA. As Prof Lovett put it, if the authors had thought that they had discovered this, they would have shouted it from the rooftops. Thus the defendants are advancing an interpretation of Ikeda's results which apparently did not occur to the authors themselves.

147. Secondly, as counsel for the claimants submitted, in those circumstances, the skilled reader would not, when they turned to the method and results, have had their interest piqued or be expecting anything surprising. Moreover, they would assume that the method and results were consistent with the stated objective and conclusion. It follows that the skilled reader would not subject the method and results to sustained analysis looking for implications which the authors had not stated.”

133. It is fair to say that the prior art has been parsed and analysed by both sides in this case in a manner which bears no relation to how they might have been considered by the skilled team at the priority date. Insofar as both sides’ witnesses sought to draw support from the detail of the prior art for the positions which they had adopted, it is important to recall that what is at issue is how the skilled team would have read the prior art at the relevant time, not how it is capable of being interpreted now with the benefit of over ten days of expert evidence.

The arguments

134. In Bayer’s opening submissions, it states that once the Pozzoli steps have been taken, the court is left with the key question, whether the difference between the prior art and the inventive concept constitutes steps which it would have been obvious for the skilled team to take. The arguments were straightforward. On Sandoz’s case, the skilled team, possessing common general knowledge, would have included once-daily dosing in Phase II trials on the basis that such a dosing regimen was obvious to try. Sandoz places particular emphasis on the Harder Poster as having “let the cat out of the bag”. Bayer, by contrast, contends that there is nothing in the prior art which would have led the skilled team to design a once-daily dosing regimen and that the Harder poster, especially when read in light of the Kubitza posters, did not provide a sufficient basis for the skilled team to move forward with oncedaily dosing.

135. The burden of proof was at all times on Sandoz in these proceedings and, in those circumstances, Bayer was noticeably reticent about positively stating whether, on its case, the prior art would have led the skilled team to design any dosing regimen, i.e. to move to Phase II trials at all. Ultimately, and without much apparent enthusiasm, it seems to have adopted the position that the skilled team would not have moved to Phase II trials at all based on the prior art. As noted by Sandoz, this was not a position adopted by Bayer in any other jurisdiction.

136. Sandoz suggests that the court adopt the four-step approach to obviousness from Pozzoli described above. The first two steps are uncontroversial. It is agreed that the skilled team is made up of a skilled clinician and a skilled pharmacologist. The inventive concept is to dose rivaroxaban in a once-daily dose.

137. The third and fourth steps can be considered together. What is the difference between the prior art and the inventive concept and, in light of those differences, would the inventive concept have been obvious to the skilled team, i.e. would it have been obvious for the skilled team to move from the prior art to the inventive concept.

138. Sandoz argues that there is, in effect, no difference between the inventive concept and the prior art – which identifies, in Harder, the potential for once-daily dosing of BAY 59. In those circumstances, it suggests that it would have been entirely obvious for the skilled team to simply do what Harder suggests. Sandoz acknowledges, however, that it is not sufficient that a potential dosing regimen be obvious in the sense that it would occur to a skilled team as a possibility, rather it must, as put in Actavis, be obvious to actually try with a reasonable expectation of success: success, in this context, meaning that it would be both safe and effective. Applying the multi-factorial approach in Actavis, Sandoz argues that, in light of the prior art, once-daily dosing was obvious to try with a reasonable expectation of success and, therefore, not patentable. I note that one of the decisions of the Netherlands’ courts referred to below characterises the distinguishing feature between the prior art and the claimed invention as being the medical use of BAY 59. This seems to be simply a different way of framing the same issue.

139. Sandoz argues that its experts should be preferred over those of Bayer, and that the secondary evidence lends credibility to its experts’ evidence and undermines Bayer’s.

140. As regards comity, it urges me to afford particular weight to the decisions of the UK courts, stressing the similarity in evidence and argument, the manner in which the evidence was tested, and the status of the decision as final.

141. Bayer contends that the invention was not obvious by reference to the prior art. Unlike, for instance, in the UK proceedings, it does not purport to identify material differences between the prior art and the inventive concept (such as the reference to rapid release form, and five days dosing). In the TBA decision, the TBA dismissed the significance of the reference to “half-life of less than 10 hours” in the Patent, and Bayer placed no reliance on it here.

142. Rather, Bayer argues that the prior art, when the Harder and Kubitza material is read together, does not point to once-daily dosing at all, but also that, in effect, the skilled team would not have placed sufficient reliance on the prior art to consider trying once-daily dosing with a reasonable prospect of success. The weight to be afforded to the prior art was, in effect, the main area of dispute between the parties. Put otherwise, the main issue between the parties was as to the teaching of the prior art.

143. Bayer’s experts’ position, and ultimately Bayer’s, was that sufficient weight could not be placed on the prior art to enable any dosing regimen to be recommended for a Phase II trial, i.e., no Phase II trial could be designed. Sandoz argued that this breached the so-called

‘patent bargain’, that the quid pro quo for gaining a monopoly is that a patent must both disclose the bargain and enable its implementation once the patent expires. On Bayer’s case, argues Sandoz, the patent for rivaroxaban (the “Straub Patent”) could not be implemented.

144. The Straub Patent was granted on 23 February 2005, i.e. after the priority date. Section

13 of the Act expressly excludes from consideration of inventive step the contents of patent applications filed before the priority date. The parties agree, therefore, that I cannot consider the contents of the Straub Patent when considering the question of inventive step. Sandoz explains that it deploys the Straub Patent not to illustrate that the Patent did not involve an inventive step, within the meaning of section 13, but rather to challenge the credibility of Bayer’s evidence that, as at the priority date, no dosing regimen could be designed for rivaroxaban.

145. Leaving aside Bayer’s position that no dosing regimen could be designed, their main objection to Sandoz’s obviousness case is that it required the skilled team to ignore:

“[M]aterial limitations in the prior art (the presence of overlapping error bars in Harder, the absence of error bars in Kubitza, the absence of a P value, the lack of necessary information required to perform the assays, the variability of assays, the variation in reported half-life, the experimental assays in Harder, the issues about where the baseline was and so forth).”

146. Moreover, Bayer contends that Sandoz’s case impermissibly relies on presumptions as to what data in Bayer’s possession, but not reported in the prior art, demonstrated, and on the contents of the confidential material disclosed in discovery and on material published after the priority date.

147. Both parties were agreed that although the Kubitza posters and the Harder poster could and should be read together – since Kubitza was expressly referenced in the Harder poster

– and that these could also be read together with the relevant Blood abstracts, it was not permissible to read them together with Perzborn. This was described as an “impermissible mosaic” and it was explained that although the skilled team is deemed to have knowledge of all prior art in the public domain, reading two or more pieces of prior art together could itself be regarded as inventive. As a consequence, this further relegates the significance of Perzborn. In closing, Sandoz briefly contended that Perzborn was part of the common general knowledge – which would have permitted it to be considered with the other prior art – but this was not supported by the evidence and was a suggestion which was subsequently withdrawn.

The evidence

148. Evidence was given on each side by eminent experts. Each of Sandoz’s experts provided a total of four expert reports, an original report and a reply to Bayer’s replying expert report, then a supplemental report on the secondary evidence and a reply to Bayer’s replying expert report on the secondary evidence. There were two reports from each of Bayer’s witnesses. The expert reports ran to over 500 pages, and they were accompanied by three volumes of exhibits. It was not disputed that the contents of the reports broadly reflected the evidence given by the same experts in the UK proceedings, save that there was no secondary evidence in those proceedings.

149. I do not propose to summarise all the evidence given by each expert, rather I propose to address the material points of difference between them and the manner in which each expert dealt with those issues where relevant to the question which I have to decide. It is important, however, that any such assessment be carried out in light of an overall assessment of each expert’s evidence and the manner in which it was given. Each expert made the necessary declarations that they understood their duty in giving expert evidence, but that does not insulate their evidence from critical analysis.

150. The assessment of expert evidence is not a straightforward task, especially where, as here, despite significant agreement on many issues, the experts reach diametrically opposed positions on the central issue in these proceedings: whether the skilled team would have tried once-daily dosing as part of Phase II trials based on the prior art.

151. In Sandoz’s closing submissions, reference was made to the decision of the High Court in James Elliot Construction Ltd v Irish Asphalt [2011] IEHC 269 in which Charleton J discussed the assessment of expert evidence (at paragraph 12):

“Every expert witness has to be evaluated on the basis of sound reasoning. An expert witness is, however, no different to any other witness simply because he or she is entitled to express technical opinions; all of us are subject to human frailty: Exaggerated respect based solely on a witness having apparent mastery of arcane knowledge is not an appropriate approach by any court to the assessment of expert testimony. Every judge has to attempt to apply common sense and logic to the views of an expert as well as attempting a shrewd assessment as to reliability.”

152. Charleton J referred with approval to the following passage from Loveday v Renton [1989] 1 Med LR 117. In that case Stuart-Smith LJ says:

“The mere expression of opinion or belief by a witness, however eminent... does not suffice. The Court has to evaluate on the soundness of his opinion. Most importantly this involves an examination of the reasons given for his opinions and the extent to which they are supported by the evidence. The judge also has to decide what weight to attach to a witnesses' opinion by examining the internal consistency and logic of his evidence; the care with which he has considered the subject and presented his evidence;

his precision and accuracy of thought as demonstrated by his answers how he responds to searching and informed cross-examination and in particular the extent to which a witness faces up to and accepts the logic of a proposition put in cross examination or is prepared to concede points that are seen to be correct; the extent to which a witness has conceived an opinion and is reluctant to re-examine it in the light of later evidence, or demonstrates a flexibility of mind which may involve changing or modifying opinions previously held; whether or not a witness is biased or lacks independence.”

153. Charleton J concluded his observations on this issue by stating (at paragraph 13):

“Of these criteria, the most important reasons whereby I have chosen one expert over another have been the manner in which an opinion has been reasoned through and the extent to which opposing views have been genuinely and objectively considered on the basis of their merit. A judge must bear in mind that, notwithstanding that an expert may firmly declare a duty to the court, it is a natural aspect of human nature that even a professional person retained on behalf of a plaintiff or defendant may feel themselves to be part of that side's team. Of particular importance in this case, therefore, has been the extent to which an expert has been able to step back and to consider and to think through an opposing point of view.”

The experts

154. The four experts in these proceedings all gave evidence in the UK proceedings. Each provided a comprehensive curriculum vitae describing their background, at least two of which ran to over 100 pages. For convenience, I will adapt the admirably brief summaries of their impressive backgrounds given by His Honour Judge Hacon in the High Court in the UK.

155. Professor Jack Hirsh gave expert evidence regarding the views of the skilled clinician on behalf of Sandoz. Professor Hirsh is Professor Emeritus in the Department of Medicine at McMaster University in Hamilton, Ontario. He has had a long and distinguished career which has included pioneering the use of warfarin in a low intensity treatment and promoting an international standardisation of units for measuring such treatment. He has particular expertise in LMWHs. His work led to the regulatory approval of enoxaparin, an LMWH.

156. Bayer’s expert clinician was Professor Mark Andrew Crowther. Like Professor Hirsh, he is an academic based at McMaster University, being Professor and Chair of the Department of Medicine and Leo Pharma Chair in Thromboembolism Research. His work includes research and clinical practice relating to the management of thromboembolism in patients. He is a fellow of the Canadian Academy of Health Sciences.

157. Professor Martin Wilkins was Sandoz’s expert pharmacologist. Professor Wilkins is Professor of Clinical Pharmacology at Imperial College London, Director of the British Heart Foundation Imperial College Centre of Research Excellence and Director of the National Institute for Health and Care Research at Hammersmith Hospital. Professor Wilkins has practised in clinical pharmacology for over 40 years in Birmingham and London, during the course of which he has participated in many clinical trials.

158. Professor Bernd Meibohm was Bayer's expert on pharmacology. He is Distinguished Professor of Pharmaceutical Sciences, Chair of the Department of Pharmaceutical Sciences, Harriett S. Van Fleet Endowed Professor in Pharmaceutics and the Associate Dean for Research and Graduate Programs at the College of Pharmacy of the University of Tennessee

Health Science Center in Memphis, Tennessee. Professor Meibohm conducts research into the optimisation of dosing regimens, adapted to the needs of the individual patient.

159. Each of the experts, therefore, had the necessary experience and knowledge of drug development processes to provide their expert opinion on the views of the skilled team.

160. There was some feint suggestion by Sandoz that Professor Hirsh’s evidence should be preferred over Professor Crowther’s on the basis that Professor Hirsh was the more ‘eminent’ of the two. I do not need to decide on the relative eminence of any of the experts. The expertise of each exceeded by such a degree the expertise of the hypothetical skilled person that any purported difference in their respective prestige is wholly immaterial to the issues which require to be decided.

161. Just as Sandoz urged the court to prefer their expert clinician, on the basis of his preeminence, Bayer sought to tentatively suggest that Professor Crowther was in a better position to give evidence of the views of the skilled clinician – it referred in its submissions to Professor Crowther as the only practising clinician who gave evidence – but I am satisfied that Professor Hirsh was as well placed to provide a perspective on the views of the skilled clinician as Professor Crowther. Professor Hirsh’s description of having trained many research fellows with equivalent skill levels to those of the notional skilled clinicians provided a practical example of how Professor Hirsh addressed the task of putting himself in the shoes of a person not only less skilled than himself, but without much of his knowledge and without any of the knowledge that an ordinarily skilled person would have gained since the priority date.

162. Bayer also seemed to object that Professor Wilkins’s ability to give evidence of the views of the skilled pharmacologist was undermined by the fact that he had had training as a clinician. I do not consider that this undermined his evidence in any way. The only area in which it was suggested that his evidence didn’t reflect the perspective of a skilled pharmacologist, but of a pharmacologist with a particular clinical background, was in relation to his knowledge of the physical processes involved in coagulation. In that regard, one of the leading pharmacology textbooks, Pharmacology, Rang & Dale, 5th ed. 2003, includes a description of the coagulation cascade and a diagram of a type similar to that shown above, which provides a sufficient basis for what Professor Wilkins claimed would be the understanding of the skilled pharmacologist.

163. It can readily be seen that all four experts were significantly over-qualified and could not be said to represent the “skilled person”. In order to explain how a skilled person would have approached certain information at the priority date at a remove of twenty years, it is perhaps inevitable that the only persons who could credibly give such evidence are persons who are now dramatically more skilled than the notional person about whose mindset they are giving evidence. It is notable, however, that all four witnesses were significantly more qualified than the hypothetical skilled person even at the priority date.

164. The task required of each expert witness was a highly artificial one, made even more so in this case as all four witnesses had given evidence in the UK proceedings. All were given the task in this jurisdiction of assessing the prior art without regard to the contents of the Patent. They had been given the same task in the UK. Having regard to their participation in the UK proceedings, this was a task which was all but impossible in these proceedings. That said, I consider that all four experts were, prima facie, equally well able to put themselves in the hypothetical position of the skilled person coming to the prior art at the priority date without knowledge of the Patent and made good faith attempts to do so.

165. In dealing with the issues which arise in these proceedings, it has been of enormous assistance that the experts agreed the common general knowledge document. There remained, of course, some issues in dispute regarding what was common general knowledge but, as appears from the discussion of the disputed evidence below, these differences narrowed somewhat in the course of cross-examination.

166. I pause to note that the proceedings provided ample evidence for the benefits of crossexamination. Each of the witnesses provided a number of detailed witness statements, which they no doubt considered accurately reflected their honest views. All adopted the contents of their reports as their evidence at the outset of their direct examinations. However, through careful and detailed cross-examination, positions expressed in absolute terms in the written reports were shown to be conditional, firmly expressed opinions were revealed to possess more nuance and, in a number of cases, the experts’ oral evidence ended up differing materially from that which was contained in the reports.

167. Before moving to a discussion of the disputed evidence, there are a number of preliminary observations on the expert evidence which it is necessary to address at the outset having regard to the necessity to assess the experts’ evidence ‘in the round’.

168. At paragraphs 218 and 219 of his first expert report, Professor Hirsh sets out part of his instructions, including the assumptions he was instructed to make when considering the prior art. These are as follows:

“218. The skilled team would know that earlier steps undertaken resulted in promising outcomes but the skilled team would not have available the body of data generated that might be needed to satisfy any ethical, regulatory or other requirement to progress matters; however, if the skilled team considered they needed it, they could generate that data themselves by routine methods.

219. Bird & Bird instructed me to assume that the hypothetical skilled team would have the resources ordinarily available to a pharmaceutical company to undertake the development of a new oral factor Xa inhibitor; and specifically:

219.1 would be able to make the drug;

219.2 would not be constrained by any intellectual property considerations; and

219.3 would have the resources to be able to generate data of the kind referred to paragraph 218 above.”

* emphasis added

169. At paragraph 228 of his first written report, Professor Hirsh expressly states that “the Skilled Clinician would not be able to obtain approval for a clinical trial on the basis of the Harder Poster and the abstracts alone” and that it is necessary to rely on the above assumptions.

170. As discussed below in considering the disputed evidence of the experts, it appeared to me that there was at least some prospect that a significant point of difference between the experts could be attributable to the assumption that relevant data could be generated by routine methods and the appropriateness of Sandoz’s experts making that assumption for the purpose of addressing the question of obviousness. In particular, as will be seen, Bayer and its experts placed significant reliance on what was termed, by Bayer, the “data deficit” that the skilled team would be labouring under.

171. In the circumstances, I addressed this issue with senior counsel for Bayer during the hearing and again on the final day of the hearing, and, in particular, whether any reliance could be placed on “routine steps” which would be taken by the skilled team. As noted in the discussion of Actavis above, there is a distinction to be drawn between an argument that simply by the taking of routine steps the purported invention would have inevitably been discovered, a case not made here, and the issue with which I am concerned, whether a challenger could rely on evidence that reproducing the prior art only involved routine steps. Bayer’s final stated position was as follows (Transcript Day 14, p. 95):

Q. And my question really - and I'm concerned that I understand if there is dispute and what the parameters of that dispute are - is, is it legitimate for Sandoz to say you can assume that by routine steps you at least get to the same point that Bayer were when they published Harder?

A. No issue has been taken with that in the evidence, or in fact the pleading or arguments so far, Judge.

Q. So you accept that that is -- other than in relation to the assays perhaps, the reproducibility of the assays.

A. Yeah. I mean, it is not, in our respectful submission it's not a material point in this case that somebody could do Phase I routine trials and get data similar to the Kubitza posters.

172. As explained below, the reference to reproducibility of the assays reflects the fact that Bayer’s witnesses gave evidence regarding the difficulty which would be faced by a skilled team trying to reproduce the trial reported in Harder.

173. During closing submissions, I also raised a related issue, what Bayer’s position was in relation to what, if any, step the skilled team would have taken based on the prior art, i.e. was there any dosing regimen that it would have recommended for Phase II trials.

174. Professor Crowther’s evidence in the UK was that he could not recommend any dosing regimen based on the prior art. As noted by the High Court and Court of Appeal in the UK, that, in effect, meant that Professor Crowther’s evidence was that the skilled team would not have proceeded to a Phase II trial; it could not if it could not propose a dosing regimen which it expected to be safe and effective. Despite Professor Crowther’s evidence, the UK courts noted that Bayer did not advocate for such a proposition in those proceedings, where Bayer accepted that the skilled team would have commenced Phase II trials.

175. When Professor Crowther reiterated that position in his evidence in these proceedings

– that there was no dosing regimen that he could recommend based on the prior art – and he was questioned about the fact that Bayer had not stood over that proposition in the UK, Bayer’s counsel intervened to confirm that Bayer was “standing foursquare behind the evidence” of Professor Crowther. However, when pressed during closing submissions on whether this meant that Bayer’s position in these proceedings was, in effect, different from the position taken in proceedings in every other jurisdiction regarding the validity of the Patent, there was some equivocation, Bayer only saying that it was not in a position to make a concession that the court had not asked it to make. Rather, Bayer suggested that the only issue was whether once-daily dosing would have been tried.

176. If no dosing regimen would be tried on the basis of the prior art then, a fortiori, daily dosing would not have been tried and the challenge to the validity of the Patent would necessarily fail. It does seem to me, therefore, that, absent an acceptance by Bayer to the contrary, the question of whether once-daily dosing would have been tried involves a consideration of the a priori question of whether any dosing regimen could have been recommended based on the prior art, i.e. whether the skilled team would have moved to Phase II trials at all. The fact that Bayer has adopted a position in this jurisdiction which appears to be different than the position accepted in every jurisdiction in which the equivalent patent has been challenged raises its own issues, as discussed below.

177. Professor Crowther’s evidence on this issue became relevant to a related issue, Sandoz’s submission that less weight should be given to his evidence than of the other expert witnesses. There were two bases for this contention.

178. First, it was contended that Professor Crowther’s prior relationship with Bayer, over a period of twenty years, was such as to undermine his independence. In this regard, he had acted as an investigator, or in other similar roles, in a number of clinical studies conducted by Bayer, including regarding the use of rivaroxaban for paediatric use. He had promoted rivaroxaban through educational presentations, a task he also performed for other companies in relation to other drugs, and he had acted as an expert witness for Bayer in five sets of proceedings over the last ten years, including the UK and Australian proceedings regarding the dosing patent for BAY 59. Sandoz’s argument was that this long association should be regarded as “unconsciously” affecting Professor Crowther’s independence such as to diminish the weight to be given to his evidence.

179. In closing submissions, counsel for Sandoz compared the level of association which Professor Crowther had had with Bayer over the years with the disclosure requirements for experts as mandated by Order 39, rule 57(2) of the Rules of the Superior Courts, suggesting that the professional connection over many years was one which undoubtedly met the disclosure requirement. The argument was not, as I understand it, that there had been any breach of Professor Crowther’s obligation to disclose his connection to Bayer, but rather that the threshold for disclosure was low, suggesting that even a minimal connection might affect independence, and that a connection which clearly met the requirement for disclosure

– such as Professor Crowther’s with Bayer –went to the independence of that witness and thus the weight to be given to their expert evidence.

180. The second challenge to Professor Crowther was that he had gone beyond the role of the expert and, in his expert report, had given his opinion on an area outside his expertise and on an issue which he had been told was not in dispute. At the outset of these proceedings, Bayer had put in issue the question of whether Sandoz would have been able to determine the chemical composition of the drug the subject of the prior art, BAY 59. This had been resolved in Sandoz’s favour in the UK proceedings, the High Court concluding that it would have been straightforward for an information specialist to discover the chemical formula from publicly available sources, and, in any event and despite Bayer’s witness’ evidence to the contrary, Bayer would have simply provided it to a skilled team had it been asked. Unsurprisingly, therefore, it was ultimately conceded by Bayer, for the purpose of these proceedings, that the skilled team could have obtained the chemical formula.

181. Notwithstanding this concession, and despite his instructions to assume that it was possible to obtain the chemical structure, Professor Crowther addressed the issue in his first expert report and stated that he “did not consider it would be possible to identify the chemical name or structure from publicly available sources as a matter of routine.”

182. Sandoz argues that this illustrates that Professor Crowther did not properly confine himself to the role of expert in accordance with his duties to the court, but instead became an advocate for Bayer.

183. Bayer, in response, stoutly defended Professor Crowther’s independence, highlighting that Professor Crowther had expressed an opinion – and stuck to it – notwithstanding that it did not align with Bayer’s position in the UK proceedings and had not subsequently tailored his opinion in light of the judgment in the UK proceedings and his awareness that Bayer had not “stood over the extremity of” his evidence. This was described as the “gold standard of the independence of the expert”.

184. On balance, I think that the suggestion that Professor Crowther lacked independence was significantly overstated. It was apparent from each of the witnesses that it is commonplace for experts of the eminence of the four in these proceedings to work with or alongside pharmaceutical companies in the course of their academic, research, or clinical work. In that regard, there was nothing to suggest that Professor Crowther’s association with Bayer was out of the ordinary or that, as a consequence, he would be unable to independently express his views on the issues in these proceedings. It seems, in fact, that he did not tailor his views to align with Bayer’s litigation strategy, at least in the UK proceedings, which, as counsel for Bayer suggests, demonstrates independence. It is true, as counsel for Sandoz points out, that that is what is required of an expert and they should not “get credit for it”, but it does undermine somewhat the complaint that Professor Crowther had become an advocate for Bayer.

185. Nor do I think that Professor Crowther was going outside his remit in offering his opinion on the discoverability of the chemical code for BAY 59. He was offering his view on what the skilled clinician would have considered possible, not on what was possible to a skilled team including an information specialist. Moreover, one of the issues which is relevant to the question of obviousness is whether the skilled team would have been sufficiently encouraged by the prior art to carry forward the development of, in this case, the drug involved. It could have been relevant to that question whether a skilled clinician reading the prior art would have thought that it was worthwhile trying to discover the chemical formula of the drug described in the prior art.

186. Accordingly, I do not attach less weight to the evidence of Professor Crowther by reason of any purported lack of independence on his part, or failure to understand his role as an expert. As with all the witnesses, I found his command of his area of expertise to be hugely impressive. Professor Crowther was an engaged witness who was careful to ensure that he clearly understood the questions being asked and then provided clear answers.

187. I do, however, consider that there are other issues of concern with Bayer’s evidence. Where a witness’s evidence changes from that previously given, especially if the reason for that change is not adequately explained, this calls into question the weight to be given to the evidence. As discussed below, when considering the evidence of the experts regarding the teaching of the prior art, there were in my view, shifts in Bayer’s evidence, in particular that of Professor Meibohm, which materially affect the weight which can be given to that evidence.

188. One of them bears mentioning at this juncture because of the curious impact it has on the evidence of Professors Crowther and Meibohm when considered together, as it should be, i.e. as the evidence of the skilled team. In his witness report, Professor Meibohm identified dosing arms which he would recommend based on the prior art, subject to discussion with the skilled clinician. In particular, he suggests that he would have recommended 20 mg bid and 30 mg bid dosing (at paragraph 261 of his report). He then confirmed that, having reviewed Professor Crowther’s written report, his recommendations had not changed (at paragraph 289):

“I was asked to provide my views on the basis of the key input as provided by the Skilled Clinician relating to the common general knowledge and the prior art including that of particular relevance to the design of any dosage regimens of BAY 59-7939 to use or test in patients. I was provided, in draft but approved form, with the Expert Report of

Professor Mark Crowther. Having considered Professor Crowther’s evidence, there is nothing in Professor Crowther’s Expert Report that causes me to change any of my views set out in my report.”

189. However, at the close of his direct examination, Professor Meibohm made what can only be regarded as a dramatic change to his evidence when he said that he wasn’t in a position to recommend any dosing arms for a Phase II trial and that his apparent confirmation to the contrary at paragraph 289 in his written report was based on a misunderstanding on his part, that he was giving his views only on the hypothetical basis that he had considered the prior art without reference to the skilled clinician. This is completely at odds with what is stated in clear terms in his report as quoted above. Such a contradiction is very difficult to understand.

190. His explanation for this apparent change of view was as follows (Transcript Day 8, pp.

67 – 68):

A. Yeah, so when I started out with this, as I said, I had not talked with a skilled clinician about that. Later on I learned that the number of bleeding events, or the absence of bleeding events in healthy individuals is not a good indicator for safety in patients that had surgery, post orthopaedic surgery patients. So, clearly the safety aspect that I probably put to the 20 and 30 mg twice-daily is probably not appropriate in light of this assessment.

191. However, it is clear from Professor Meibohm’s own written report that this was information about which he was already well aware when providing his initial opinion, i.e. without needing to engage with the skilled clinician. It is referenced several times in his report. It could not and does not, therefore, explain his apparent change of mind. Nor did his explanation that he had subsequently had a conversation with the clinician, in circumstances where he did not identify anything ‘new’ arising from such a conversation. Professor Meibohm’s evidence in this regard was entirely unconvincing and remained somewhat confused, as evidenced by the following exchange with counsel for Sandoz after his direct evidence that he would no longer recommend any dosing regimen (Transcript Day 8, page 84):

Q. And, Prof. Meibohm, I suppose just for absolute clarity, what is your position today as regards what the skilled pharmacologist would recommend or think was obvious in light of the prior art; do you still continue to identify 20 mg and

30 mg twice-a-day as being something that was obvious to try?

A. I think what was obvious to try for me, as a clinical pharmacologist, was at least a dosing regimen that is twice, perhaps even three times daily with dose increments that basically would be supported by what is available at the time after I had a conversation with a skilled clinician.

192. Professor Meibohm later said that despite having proposed dosing arms in his written report, in the selection of dosing arms the skilled pharmacologist would defer to the skilled clinician “but obviously it would be a conversation” with the skilled pharmacologist (Transcript Day 10, p. 67). Admittedly, in proposing dosing arms, it was on the basis that it would be subject to consultation with the skilled clinician. But, as noted above, his report indicated no change despite such consideration of the skilled clinician’s expert report.

193. However, Professor Crowther’s evidence was to the almost opposite effect – that it was a matter for the skilled pharmacologist to make a dosing recommendation in the first instance, supported by the skilled clinician (Transcript Day 11, pp. 158 – 159):

Q. So, as you said in your evidence in England, you would defer to the views of the skilled pharmacologist in relation to devising the dosing regimen?

A. Yes, insofar as I would look to them to provide me with recommendations. And then the skilled team would assemble to consider those recommendations, based on all the information we have available to us. And then if we could, we would move forward to the dosing recommendation.

194. This leaves a somewhat troubling vortex at the heart of Bayer’s defence where, on their expert’s view, no one on the skilled team was responsible for making an initial recommendation regarding dosing arms.

195. A further issue, discussed in more detail below, is that Bayer’s position in these proceedings, and that of its experts, as regards whether the prior art would have enabled a skilled team to design any dosing regimen, is at odds with Bayer’s position in all other proceedings in which the Patent has been challenged. This necessarily undermines the plausibility of the witnesses and of Bayer.

196. One final issue which can usefully be addressed here going to the overall plausibility of the evidence adduced by the two sides, is the curious contention by Professor Meibohm that the purpose of the thrombin generation study was completely different to the purpose actually described in the Harder poster or relevant Blood abstract. The purpose, as clearly stated, was to evaluate the effect of BAY 59 on thrombin generation in healthy male volunteers. As discussed above, inhibition of thrombin generation as a possible mechanism for a new anticoagulant was the subject of significant interest at the priority date, so there is no ambiguity about the purpose of the poster. The conclusions drawn are all relevant to the expressed purpose.

197. Professor Meibohm, however, suggests (in his written report, at paragraph 188) that the study would be understood by the skilled pharmacologist as an “attempt to identify whether certain assays or biomarkers were useful for measuring the effects of BAY 59-7939 and therefore could be used in future studies to determine the activity of BAY 59-7939”. There is nothing in the poster which supports such an interpretation which, in its own terms, makes little sense. Professor Meibohm sought to explain what he meant during the course of his evidence (Transcript Day 9, pp. 160 – 169) and again the following day in evidence. He was cross-examined regarding the content of a post-priority date article by seven of the authors of the Harder poster entitled “Effects of the Oral, Direct Factor Xa Inhibitor Rivaroxaban on Platelet-Induced Thrombin Generation and Prothrombinase Activity” published in the Journal of Clinical Pharmacology (2007). The concluding paragraph was as follows:

“In conclusion, despite automation, the ETP method with freshly drawn PRP has a high technical standard for a routine laboratory test because it requires several processing steps and laboratory skills. However, PiCT, a simple plasma-based assay, might be useful to useful to monitor the effects of direct Factor Xa inhibitors, even at low levels of inhibition of FXa activity, are not detected by other coagulation tests. Overall, the results of the PiCT and ETP tests indicate that rivaroxaban was able to diminish the rate of thrombin generation when platelets are activated. Peak ETP and PiCT are still affected 24 hours after administration of rivaroxaban 30 mg, supporting the concept of once-daily administration.

198. Professor Meibohm interprets this paragraph as supporting his theory as to the “real”

purpose of the thrombin generation study. It does not; it merely observes that one of the assays used in the study has more potential for routine use than the other. Of more relevance in this paragraph is the interpretation given to the results, which Professor Meibohm dismisses on the basis that he didn’t have the data available to the authors. His alternative explanation for the study, therefore, remained unconvincing and reflects a determination to downplay the Harder study’s significance.

199. I turn now to address the relevant disputed CGK evidence.

Disputed CGK evidence

200. There were some disputes between the experts regarding what was common general knowledge, especially in their written reports. Following cross-examination, many of these issues narrowed. Insofar as the oral evidence of experts on both sides may have differed from that contained in their written submissions, they are, of course, open to the criticism that their evidence should be rejected for inconsistency. Both sides made some play of the fact that the other side’s experts had adopted the contents of their written reports and then departed from those contents. I think that there is a risk of overstating this criticism, especially where an expert is equally open to criticism for failing to be open-minded when giving their evidence and capable of changing their minds in the face of new evidence and new points of view.

201. Where evidence may have changed on certain issues, in particular, as to what was the common general knowledge of the skilled team at the priority date, the significance of that change will be determined by reference to the materiality of that issue in the context of the overall issue to be determined by this court, whether the skilled team would have considered it obvious to try once-daily dosing with a reasonable expectation of success.

202. It is also true that an unexplained departure from evidence previously given, whether in written reports or other proceedings, could give rise to some concern. In Ranbaxy, Clarke J addressed a criticism of experts for having changed their evidence from that given in proceedings in the UK (at paragraph 52):

“… Quite an amount of the cross-examination of both witnesses was concerned with what might be reasonably be described as suggestions of nuanced differences between the evidence now being tendered and that given on previous occasions. I have to say that such an eventuality, far from being a matter of criticism, is to be expected. The precise focus of the issues upon which a case may come to turn will undoubtedly crystallise in the course of litigation. If a second set of litigation, of an almost identical variety, is then conducted with the same expert witnesses, it is hardly surprising that the legal advisors of the parties and those witnesses will focus more closely on what have turned out to be the key issues. It should not be assumed that the evidence given on the first occasion is some form of sacred text which cannot be altered. There are doubtless things that were not said because questions were not asked or issues did not arise in precisely the same way. The fact that issues may arise in a more focused way in a second or subsequent piece of litigation between the same parties is almost certain to lead to some evolution in the evidence given. That is not to say that a court would readily accept a significant alternation in the evidence given by a witness. It would require a very telling explanation to justify a witness significantly altering evidence previously given as to the state of knowledge of the skilled addressee. However, a greater focus, within that evidence, on specific questions which may not have been explored as fully in previous litigation is likely, legitimately, to lead to some nuanced evolution of the evidence given.”

203. In that case, Clarke J considered that the expert evidence had not varied beyond what might be expected from that previously given and therefore rejected criticism of the experts on that ground. Save where stated otherwise, for instance, in relation to Professor Meibohm’s changed evidence regarding the recommendation of dosing arms, which is addressed above, I consider that any variation in the experts’ oral evidence from their written reports was “within tolerance” in this case and does not, of itself, affect the weight to be given to their evidence.

Wide or narrow therapeutic range

204. As discussed above, the therapeutic range is the range within which a drug is both effective and safe, the Goldilocks-zone where the amount of drug is not too little as to be ineffective and not too great as to be toxic. In the case of anticoagulants, this is particularly important, as there must be sufficient drug in the patient’s system to produce the required antithrombotic effect, or else there is a risk of thrombi forming. But as the same action which produces the antithrombotic effect increases the risk of bleeding, too much drug in the system can also cause adverse bleeding events. It was accepted by all parties that a novel drug where safety concerns arose in relation to both the upper and lower limits of the therapeutic range gave rise to additional challenges in drug development and required greater caution than, for instance, in the case of a drug, such as that at issue in Actavis, where there was no safety risk from underdosing.

205. It was agreed that the therapeutic range of BAY 59 was not known at the priority date. However, Professor Hirsh stated in his supplemental report that there was a perception that Factor Xa inhibitors would have a wide therapeutic window. He referred to this possibility having been discussed at conferences. However, he accepted in his oral evidence that there didn’t exist good evidence that this was so, and it was not common general knowledge that Factor Xa inhibitors generally or BAY 59 in particular had a wide therapeutic range at the priority date. Both parties agreed that it would require Phase III studies to be completed before this was known.

206. There remained a difference between the parties regarding whether this meant that, in proceeding to Phase II studies, it was appropriate to assume a narrow therapeutic window. Bayer’s witnesses suggested that this was appropriate both because it was known that warfarin had a narrow therapeutic range and as being, in any event, appropriately cautious. It seems to me that the fact that the therapeutic window is unknown is principally relevant to the overriding considerations of patient safety, which both sides agree is the primary concern in designing clinical trials. Any issue which might affect patient safety about which information is unknown will require a skilled team to proceed with caution. The skilled team would, therefore, have proceeded conservatively on the basis that the therapeutic range was not known and could not have discounted the possibility that BAY 59, like warfarin, had a

narrow therapeutic window. The comparison with warfarin is not appropriate, however, as discussed presently.

Reliance on LMWHs as comparator

207. As noted above, the starting point when designing a dosing regimen for a drug is its half-life, but there are drugs where there is a disassociation between pharmacokinetic and pharmacodynamic effects, where dosing is not in line with half-life. Notably, LMWHs are such a drug type.

208. In Professor Hirsh’s written reports, he suggests that the skilled clinician’s experience with LMWHs would have informed their expectations regarding BAY 59 (see, for instance, paragraphs 226 and 247 of his first report). However, in cross-examination, he accepted that the skilled clinician could not and would not have “read across” from their experience with LMWHs to an expectation that a Factor Xa inhibitor would behave in the same way. The reasons for the disassociation between half-life and appropriate dose in LMWHs are still not properly understood, LMWHs and Factor Xa inhibitors have a different mode of action, and the literature suggests that a read across is not appropriate.

209. I do not consider, as suggested by counsel for Bayer that the common general knowledge regarding the half-life of LMWHs therefore becomes irrelevant. As Bayer itself suggested in closing submissions, the skilled team comes to the prior art with all its knowledge and all its prejudices. In those circumstances, it seems to me that the fact that one of the three main anticoagulants – and the only one with a sufficiently predictable response not to require monitoring – was not dosed according to half-life would have some bearing on the question of whether the skilled team would be open to dosing a novel anticoagulant other than strictly by reference to half-life. In light of the common general knowledge regarding LMWHs, the skilled team would not have simply rejected that possibility as impossible or improbable, though they would not have had any expectation that this would be effective simply because of the experience with LMWHs. Similarly, the fact that LMWHs, at least in some jurisdictions, were dosed once-daily would have informed their expectations regarding the potential for once-daily dosing of BAY 59, at least by ensuring that there would be no preconception that such a dosing regimen was unfeasible.

210. I note that in similar manner to Sandoz’s reliance on the experience with LMWHs as regards half-life dosing, Bayer relies on the experience with warfarin regarding the therapeutic window. This seemed to me to be the same type of impermissible ‘read across’ as Bayer had criticised in relation to LMWHs, and I raised this with counsel for Bayer. It was suggested that the warfarin comparison had been advanced simply to counter-balance Sandoz’s reliance on LMWHs. In any event, I conclude that the skilled team would not have made any prediction about the effect of BAY 59 simply by reading across what was known about LMWHs or warfarin. However, I also consider that the skilled team’s experience with both would have informed their overall view about designing clinical trials for BAY 59. Put otherwise, there would have been no predisposition to reject: the possibility that BAY 59 had a narrow therapeutic window; that it was possible that it might be dosed other than strictly in accordance with half-life; or that a novel anticoagulant might be suitable for oncedaily dosing.

Value of assays

211. One of the most significant points of difference in the evidence between each parties’ experts was in regard to their treatment of the various assays described in the Kubitza and Harder posters.

212. There was a measure of agreement. In particular, it was agreed that none of the assays, whether those described in Kubitza or those described in Harder, had been correlated for clinical effect in Factor Xa inhibitors, i.e. it had not been established that an effect shown in an assay was predictive of clinical effect in a patient. Indeed, it was agreed that this could not be done until a Phase III study, if at all, and it appears that it has still not been done for any assay in relation to Factor Xa inhibitors.

213. It was also agreed that the assays referred to in Harder were “experimental” assays. However, it is important to understand that this description simply means that they were used in research, i.e. in laboratories, rather than in clinical practice. Although there was some suggestion that knowledge of these assays would have been completely outside the common general knowledge of the skilled team, it was accepted by Professor Crowther in his evidence that the ETP assay was a more commonly used research tool with which the skilled clinician would have some familiarity. He also accepted that they were the best assays available for measuring thrombin generation, or at least that there were no better assays (Transcript Day 12, pp. 17 - 18).

214. In their expert reports, Professors Crowther and Meibohm placed significant emphasis on the fact that the assays relied on in the Harder poster had not been correlated for clinical effect, as discussed above. This thread runs through Bayer’s evidence. Bayer’s position was characterised by Sandoz as the assays being of no value. Bayer repeatedly sought to suggest that this was overstating its position and that its evidence was only that the assay results could not bear the weight placed on them, whether in the Harder poster or Sandoz’s evidence. I think that, on balance, Sandoz’s characterisation is not unfair. Although Bayer’s witnesses, ultimately, accepted that the results were “informative” and “intriguing”, their position in substance was that they were of no utility in the context of these proceedings, where the experts were being asked to give their opinions on the views of the skilled team in light of the prior art, but could only be of use as part of a larger data set.

215. By way of example only, Professor Crowther, at paragraph 302 of his report, explains why the skilled clinician “would not have agreed with the conclusion made in the fourth bullet point” of the Harder poster, i.e. the suggestion that BAY 59 may be suitable for oncedaily dosing. A number of reasons are given, but they include the following:

“Most importantly, however, the clinical relevance of any differences between the assay measurements and baseline at any time point was unknown to the Skilled Clinician.

Therefore, it was not possible to say that the maintenance of any such difference (particularly when the assays are performed in healthy volunteers) could be expected to lead to a safe and therapeutic effect in patients, and the Skilled Clinician would not have considered it appropriate to draw conclusions as to the appropriate dosing regimen to test a novel compound in patients based on data concerning such assay measurement. In fact, to make such a claim is extraordinarily speculative.”

216. In cross-examination, he explained as follows (Transcript Day 11, pp. 127-128):

Q. Because again I just want to understand your position on this, Professor.

Prof. Hirsh has said that your position is effectively that the results generated by these assays have no value. I've asked you whether they produce relevant information. And I just want to understand, are you saying they have some value and are capable of producing relevant information?

A. Again, I go back to what I said earlier; it depends on the context. I think if the hypothesis is that you're going to take a drug forward based solely on this information, I think it has no value. I think if it's part of a larger data set that you have accumulated, it may be relevant.

217. In Professor Meibohm’s expert report (at paragraph 85) he states:

“However, Professor Wilkins appears to say… that if an “extended PD effect”, in the sense of an effect on a particular biomarker/physiological measure outlasting the concentration of the drug in the plasma, is observed in phase I studies, this may influence the dosing frequencies to be investigated in subsequent studies in patients. If that is what Professor Wilkins is saying, I disagree with him unless the extended effect in the biomarker in question has already been correlated with an extension of therapeutic efficacy or, of course, unless the drug/targeted disease is such that no safety issue arises.”

* emphasis in original

218. There are a number of difficulties with Bayer’s witnesses’ rejection of the relevance of the assays, or biomarkers, used in the thrombin generation study the subject of the Harder poster on the basis that they had not been correlated with clinical effect.

219. First, there is the apparent inconsistency between their apparent ‘trust’ in the results reported in the Kubitza poster and their lack of faith in the Harder results. Both are based on assays which have not been correlated with clinical effects for Factor Xa inhibitors, yet the Bayer witnesses place repeated emphasis, for instance, on the fact that the pharmacokinetic and pharmacodynamic data in the Kubitza studies were said to be closely correlated. Although the experimental assays used in Harder may have been less familiar than those used in Kubitza, a lack of correlation with clinical effect could not be a reason to consider it appropriate to place weight on one set of data but not the other.

220. Second, there is the logical impossibility of Bayer’s witnesses’ implicit suggestion that only an assay with correlated clinical effect could have any relevance for the purpose of designing a dosing regime. If a correlated assay was required to move from Phase I to Phase II trials, and a correlated assay could not be identified without Phase III trials, clinical development would be inevitably stymied. It appears to be suggested that absent a correlated assay, reliance would be placed on half-life alone in order to determine a dosing regime for a Phase II trial, but it was not explained how that could provide any information about pharmacodynamic effects necessary to move to Phase II trials.

221. Thirdly, and as Professor Meibohm was required to accept in cross-examination, the use of assays which have not been correlated is entirely consistent with the relevant guidance and literature at the priority date. In his cross-examination, when it was put to Professor Hirsh that Professor Crowther would “fundamentally disagree” with his proposition that if a pharmacodynamic effect was detected with the assays used in the thrombin generation study, it would provide a reasonable expectation of success in a Phase II study, he stated as follows (Transcript Day 3, pp. 130 – 131):

A. It’s the only way of doing it. There was no other. Otherwise you couldn’t go to a Phase II study.

… That’s all you had available. All you had available were tests that were performed and shown to be correlated with other anticoagulants, namely warfarin/heparin. But there was no way – they couldn’t have been correlated with the new drugs, because a Phase III had not been done.

222. Professor Hirsh’s view that assays or biomarkers which have not correlated can be of assistance in drug development and, in particular, in guiding dose selection is supported, even encouraged, by the Food and Drug Administration (FDA) and European Medicines Agency (EMA) Guidelines which were current at the priority date, and in the relevant literature.

223. Although a distinction is drawn in these documents between assays or biomarkers which are correlated with clinical effects or, in the language of the FDA, are surrogates, it is made clear that biomarkers which have not been so correlated can be used in designing clinical trials. For instance, the FDA document, Guidance for Industry, Exposure-Response Relationships – Study Design, Data Analysis, and Regulatory Applications (April 2003), described by Professor Meibohm as the “Bible” which everyone in drug development wants to follow despite its “nonbinding recommendations”, states as follows:

“Many biomarkers will never undergo the rigorous statistical evaluation that would establish their value as a surrogate endpoint to determine efficacy or safety, but they still have use in drug development and regulatory decision-making. Changes in biomarkers typically exhibit a time course that is different from changes in clinical endpoints and often more directly related to the time course of plasma drug concentrations, possibly with a measurable delay. For this reason, exposure – response relationships based on biomarkers can help establish the dose range clinical trials intended to establish efficacy.

… Biomarkers can also be useful during the drug discovery and development stage, where they can help link preclinical and early clinical exposure-response relationships and better establish dose ranges for clinical testing.”

224. After Professor Meibohm was taken through these documents in cross-examination, he made the following clarification to his evidence (Transcript Day 8, pp. 153):

A. …. So, my precise evidence on this is that I say you can, in extended PD effect in question, or a biomarker that has an extended effect can clearly be helpful in designing a dosing regimen, especially if it's a set of biomarkers. The question is what the biomarker actually measures and how it is related to an impact. And it may not necessarily be one biomarker that basically has the overarching evidence basically to suggest it, but a multitude of biomarkers taken together that have been characterised by the skilled clinician as being useful in that context could definitely have a value, even if they are not validated - what did I say here - validated as, or related to -- correlated with clinical efficacy.

225. A request for confirmation that the biomarkers would not have to be correlated in order to be relied on in clinical trials prompted the following exchange (Transcript Day 8, pp. 155

– 156):

A. Well, that would be ideal. But as I pointed out, there are instances, as in this one, where there is no correlation. So you have to start somewhere.

Q. Exactly. And otherwise you wouldn't be able to move to Phase II; isn't that right?

A. Well, I think some people would move to Phase II probably with a gut feeling where they wake up in the morning and think 'This is the drug that we should take forward.' And unfortunately there are, at least at the time there were scenarios like this. But that is not what you want to do. If you want to have a rationale that you can defend, that you basically can say 'This is why we want to do it,' obviously you have to come up with something, you cannot just say 'We don't have anything.' So it's imperfect, it's very imperfect. And then you start to pull in all kind of information that you can, from pre-clinical, from clinical.

You will not do this with the skilled pharmacologist and the skilled clinician, you will have a whole team of people that sit together and pore over all the data - safety, efficacy, pre-clinical, clinical - to ultimately come up with that

decision.

Q. Including ethics committees; isn't that right, Prof. Meibohm?

A. Well, I think you have to come up first with something and you have to -- you have to have a good rationale for the ethics committee that you can present it to them. Obviously if you don't have a rationale, the ethics committee say, 'Well, how did you come up with this?'

226. As Professor Wilkins explained, the assays may not have provided evidence of a clinical effect, no assay could, but it did show a biological effect, which was the most that could be shown from Phase I trials (Transcript Day 6, p. 70):

A. I think this is -- this is a biological assay and there's rich biological information in these graphs. And it's evident to me, as I've said, there's a pharmacological effect which is dose and time dependent and directionally it's all in the same direction and consistent over 24 hours.

227. He had explained his approach to biomarkers in the following way (Transcript Day 5, p. 95):

A. Oh, my position is that I would take a more expansive view [than Professor Meibohm] and if I could understand the context of the biomarker and its biological significance then I would use that to inform my judgment.

228. This is entirely consistent with the relevant guidance.

229. It is clear, therefore, that the apparent rejection by Bayer’s witnesses in their expert reports of the relevance of the thrombin generation study on the basis that the assays or biomarkers used in that study had not been correlated with clinical efficacy was not justified and, indeed, ran contrary to relevant guidance. Bayer’s witnesses did not adequately, or indeed at all, explain why they had placed such weight on the absence of correlation in a manner apparently at odds with the relevant guidance.

230. As is apparent from their evidence, the lack of correlation was not the only basis for Bayer’s witnesses’ rejection of the contention that the prior art would have led to a reasonable expectation of success of BAY 59 with once-daily dosing, but it was clearly a critical element in their consideration. The evidence clearly suggests that the skilled team would not have discounted the results reported in the Harder poster on this basis alone. In so far as Bayer’s witnesses’ evidence was premised on a contention that they would have done, I reject that argument. This undermines a central plank of Bayer’s case that the Harder poster did not point the way to once-daily dosing.

231. As an aside, I note that in many of the foreign judgments referred to in these proceedings, reliance has been placed by the relevant decision-maker on this noncorrelation. It is not clear what evidence was led on this point in those proceedings, but the evidence in these proceedings does not support a conclusion that, in effect, the Harder poster would have been discounted for this reason. Notwithstanding the obligations of comity, I am required to determine this case on the basis of the evidence before me. The evidence here leads to a different conclusion on this important issue and is, on this ground at least, a

basis for distinguishing those cases where this was a factor in the relevant court’s consideration.

Hip surgery low risk

232. It was agreed, as noted above, that the skilled team would know that any Phase II study into a new anticoagulant would involve the drug being used in patients undergoing voluntary hip surgery for the prevention of VTE. In this regard, it is important to understand that the drug is being given as a prophylaxis, to prevent a thrombosis arising, rather than to treat a thrombosis which has occurred. The Phase II study actually carried out by Bayer was frequently referred to in evidence simply as the ‘Hip Study’.

233. In Professor Hirsh’s witness statement, he describes the risks of symptomatic VTE as low and the risk of fatal PE as very low. However, in his oral evidence, he confirmed that he did not think that joint surgery was low risk. It was suggested that there was some contradiction in his evidence, especially given that he accepted that, insofar as reliance was placed on venograms to increase the safety of the surgery, this was of limited utility because a venogram, typically taken a number of days after surgery, could not assist if an embolism had occurred, it only gave information on whether a thrombus had formed. I asked Professor Hirsh how he squared the circle between what he had stated in his report and his evidence. He explained as follows (Transcript Day 4, pp. 17 – 18):

A. So, the difference here is most of the venograms, probably 95%, that are positive are on small silent thrombi. So, the risk, I said the risk of symptomatic venous thrombosis is low. We're talking about -- but the end point, the venography, takes in all the positive venograms. The vast majority of those are asymptomatic, small and insignificant.

Now, when I say that it's high risk surgery, yes it's high risk surgery, but the risk of a fatal pulmonary embolism, untreated, around that time was around about less than 1%. But that's still high risk, because these people are perfectly healthy except they need hip surgery. So it's considered high risk. So, the issue here is one wouldn't be able to do a Phase II study -- oh, there's one other thing, and that is when I'm talking in that book and I'm talking about patients getting no prophylaxis, but the first dose -- but these patients are getting prophylaxis.

And the hope is that the lower dose works. More than hope; the expectation is that the lower dose works.

But the dose range that's used in a Phase II study is about eightfold, between about 5 and 40 mg. But the expectation is that 5 mg will be effective, but will not be as effective as a higher dose. If you do the mathematics, the risk is very low. And that's why it's accepted by ethics committees.

234. Thus, it seems that there is a low risk of an adverse outcome – without prophylaxis – but that low risk must be seen in the context that it is otherwise healthy patients who are being exposed to that risk and the potentially very serious consequences if the most serious risk – PE – actually occurs. Surgery, of course, also carries with it the risk of excessive bleeding, capable of being aggravated by a new anticoagulant. Thus, the surgery must properly be considered high-risk, capable of being mitigated by an effective anticoagulant, which necessarily informs the skilled team’s approach, when introducing a novel anticoagulant in place of a proven drug in a Phase II trial.

235. One point worth observing on this issue is the evidence of Professor Hirsh that, although PE may be the most severe potential risk from surgery, surgeons are typically more concerned about excessive bleeding, on the basis, it seems, that it is the adverse outcome more likely to arise. In this regard, he suggested that at the priority date, there was in fact, a reluctance on the part of surgeons to use anticoagulants at all. During cross-examination, Professor Hirsh was directed to his own book, Low Molecular Weight Heparin (3rd ed.). Although the cross-examination and the text was directed to the importance of prophylaxis, it was suggested in the text, with which Professor Hirsh (unsurprisingly) agreed, that despite the risks of serious adverse events “some physicians and surgeons remain reluctant to use effective prophylaxis in all high risk patients”. Reasons are suggested in the text, including a perception that prophylaxis is not effective and also:

“The second major reason is that surgeons’ fears of anticoagulant related bleeding override their fears of clinically important postoperative venous thromboembolism.”

236. Professor Hirsh reiterated the point in re-examination (Transcript Day 4, p. 168):

Q. And how did clinicians, including surgeons, compare the relative risks of bleeding versus thrombosis?

A. Well, I think that there was a difference between surgeons and physicians.

Surgeons were the ones responsible for doing the operation. They were much more concerned about not producing bleeding than anything else.

237. This was accepted by Professor Crowther (Transcript Day 11, pp 141 – 142).

238. The skilled team, therefore, would have been just as keenly aware of the risks of overdosing a novel anticoagulant, excessive bleeding, as of underdosing, notwithstanding that VTE presented the greater risk of a catastrophic outcome. Concerns about the risk of excessive bleeding could only have been heightened by the experience with razaxaban as we will see now.

Half-life dosing

239. As set out above, there was no dispute about the fact that dosing according to half-life was the starting point in designing a dosing regimen. It was accepted that that was not an absolute and there were circumstances in which dosing according to half-life did not apply, in particular, where there was a disassociation between pharmacokinetic and pharmacodynamic effects. Nor was there any real issue taken with Professor Wilkins’ contention that where there is pharmacodynamic (PD) data which is at odds with the pharmacokinetic (PK)) data, greater reliance will be placed on the PD data: PD data trumps PK data. This is entirely logical; the main concern in drug development is with the drug’s impact on the body, not the body’s impact on the drug.

240. On this issue, Professor Wilkins suggested that razaxaban provided an example where dosing according to half-life had not been a safeguard against adverse outcomes. It will be recalled that the higher dosing arms used in the Phase II study for that drug had been abandoned because of excessive bleeding issues.

241. I am not sure that the evidence established that the dosing regimen in the Phase II trial for razaxaban had been determined by reference to half-life, still less that this was part of the common general knowledge, though I note that Bayer’s witnesses did not suggest otherwise. However, insofar as Bayer presented the experience with razaxaban as a cautionary tale, reinforcing the importance of a conservative approach in designing dosing regimens, there is no basis for suggesting that it serves as an example of what happens if dosing is other than strictly in accordance with half-life. Moreover, insofar as it was a cautionary tale, it highlighted the risk of overdosing a novel anticoagulant, not underdosing.

Once daily versus twice daily

242. One final issue which merits discussion at this point is the dispute between the parties regarding the relative merits of once-daily over twice-daily dosing and thus the motivation of the skilled team to investigate once-daily dosing. Bayer’s witnesses seemed to downplay the attractiveness of once-daily dosing despite the agreement in the common general knowledge document that it was preferable to twice-daily dosing. The undisputed evidence was that adherence to a dosing regimen was a relevant factor in determining the effectiveness of a drug and that, accordingly, a simplified dosing regimen was desirable from a patient convenience and adherence point of view and, as Professor Hirsh highlighted, a marketing point of view. One empirical study referred to in the evidence showed the benefits of onceor twice-daily dosing over three times daily dosing.

243. In their closing submissions, Bayer asserts that there was no “clinical imperative to pursue or prioritise once-daily dosing”. I accept that this statement is consistent with the evidence, but clinical imperative is not the only consideration when developing new drugs. Insofar as Professor Hirsh highlighted the attractiveness of being able to market a once-daily dosing regimen for a novel anticoagulant, assuming that it was safe and effective, he was much more convincing than Professor Crowther in his reluctance to accept that this was so.

244. Of course, the goal in developing a novel drug was to find the dosing regimen which achieved the greatest level of efficacy while maintaining the highest degree of safety. But the skilled team would have been motivated to investigate once-daily dosing had there been a basis for considering that it would be likely to be successful, even if twice-daily dosing also appeared likely to have a reasonable prospect of success. There was no suggestion by

Bayer that the skilled team would not have considered investigating both options in a Phase II trial if the data supported it, even if, as appears to be the case, only a single dosing regimen would likely be included in Phase III trials.

Foreign judgments

245. During the hearing, the parties identified a number of challenges to the Patent, or an equivalent patent, in multiple jurisdictions, most of which are signatories to the EPC. The Patent has been determined to be valid in some jurisdictions, invalid in others. At the date of hearing of this action, the UK was the only EPC jurisdiction in which the challenge to validity was final, i.e. not subject to any further appeal. The Court of Appeal of England and Wales upheld the High Court’s decision declaring the Patent invalid, and the Supreme Court has refused leave to appeal.

246. On the day the hearing concluded, the Swiss Federal Court of Appeal rejected Bayer’s appeal against the finding of invalidity made by the Swiss Federal Patent Office. A copy of the judgment, unreasoned, was provided to this court on 29 May 2025 and a reasoned judgment was provided on 20 June 2025. The parties agree that the Swiss decision is also now final.

247. In addition to the challenges in individual jurisdictions, the validity of the Patent has been upheld by the TBA. It has been held invalid in two non-EPC countries, Australia and South Africa. The decision in Australia is also final.

248. The principles regarding a challenge to the obviousness of a patent are not significantly in dispute between the parties. Although some jurisdictions, including the TBA, apply a socalled ‘problem-solution’ analysis rather than the four-stage test suggested in Pozzoli, neither party advocated for such an analysis here or, more importantly, suggested that such an analysis might point to a different outcome. The relevance of the foreign judgments, therefore, has more to do with comity than precedent.

249. The principle of comity was explained in the following way in Ranbaxy:

“54…. The principle of the comity of courts requires that the courts in one jurisdiction should not likely depart from a decision on the same issue made by a court of competent jurisdiction in another country which had to deal with that issue as part of litigation properly under its consideration. Thus, for example, where the courts in one jurisdiction have interpreted a contract in an particular way and where the same contract comes to be interpreted, in a separate dispute between the same or similar parties, in the courts of another jurisdiction, then the comity of courts requires that the interpretation of the contract in the second proceedings should not lightly depart from the interpretation given to the same contract in the first proceedings.

55. This latter principle, it seems to me, ought also apply, though obviously to a more limited extent, where the issue, while not identical, is very similar. For those reasons it seems to me to be appropriate, subject to the caveats relating to differences in statutory law, jurisprudence, the patents themselves, and the evidence which I have already identified, to pay appropriate regard to the international decisions in the related cases.”

250. In Gilead Sciences Ltd v Mylan SAS [2021] IECA 22, Costello J (as she then was) stated as follows:

“90. It seems to me that, in these circumstances (but noting that neither party in this case canvassed the possibility of an issue estoppel arising from a final decision of the courts of another jurisdiction in proceedings between the same parties and applying the same EU law test to identical SPCs – see Rio Tinto Zinc Corp. v. Westinghouse Electric Corp., RTZ Services Ltd v. Westinghouse Electric Corp. [1978] 1 All ER 434;

[1978] AC 547), this court should start from the proposition that it ought to follow the

decision of a final court of a member state of the EU where it is applying an identical EU law test to the identical SPC. In this instance, the decision of the English Court of Appeal is final as leave to appeal was refused by the UK Supreme Court. It is for the party who asserts to the contrary to satisfy this court why it should not follow this precedent. It may do so by reference to material factual differences or if it can demonstrate that a decision was reached on a clearly erroneous application of EU law.”

251. In Warner-Lambert Company v Generics UK [2019] 3 All ER 95, Lord Sumption referred to the decisions in other jurisdictions on the same claims operating as “a reality check” but noted that the force of the proposition that a court should not lightly depart from the decision of a court in another jurisdiction on the same claim depended “entirely on how far the factual and technical evidence before the foreign court” was the same as before the domestic court.

252. Actavis also involved a challenge to a dosing patent on the grounds of obviousness. Both parties had relied on decisions of countries which were EPC signatories. As here, the Patent had been declared invalid for obviousness in some jurisdictions, but in others it had been upheld. Lord Hodge observed that:

“101. Because of the differences in the evidence led, the manner by which it is tested, and the differing findings to which that evidence gives rise, one may derive support from the approach to the question and the methods of reasoning of other national courts but should never rely uncritically on the outcome.”

253. In Norton v Boehringer [2022] IECA 58, the Court of Appeal explained the obligation of the court where there are judgments of foreign jurisdictions on the same issue it is required to address:

“69. The Judge here was of course entitled – indeed obliged – to make his own assessment of the arguments and evidence heard by him. As Ranbaxy makes clear, the principle of comity did not require the High Court here to reach the same outcome as the other courts which had considered the validity of the Patent. In the language of Lord Hodge in Actavis Group PTC v ICOS Corporation, the Judge could not “rely uncritically on the outcome” of other challenges to the 220 Patent. But Ranbaxy required the Court to engage with the decisions made by those courts. That obligation went beyond merely giving a persuasive status to the legal principles applied in those decisions and it is clear that, in adopting that approach, the Judge fell into the confusion highlighted by Clarke J in Ranbaxy. The Judge was required to address the decisions made elsewhere and to give those decisions persuasive status.”

254. Where, as here, there are numerous decisions which go ‘either way’, the injunction to give the decisions themselves persuasive effect is necessarily challenging. It certainly cannot be met by reference to some overall scorecard: any suggestion that because more jurisdictions have upheld the validity of the Patent than not, this court should be persuaded to uphold its validity has little merit. Rather, what is required is some analysis of the decisions relied on in order to determine the extent to which the assessment by the foreign court mirrored the assessment to be undertaken by this court, i.e. was by reference to the same evidence tested in the same way. Regard must also be had to the status of the decisions in the foreign jurisdictions, that is, whether or not the decisions are under appeal.

255. It should be noted that no evidence was led in this court regarding the procedures in any of the foreign jurisdictions or regarding the evidence which was actually adduced in each case. Any analysis of the foreign judgments is necessarily confined to what is revealed about those matters by the decisions themselves. Moreover, the parties did not address at all a number of the foreign judgments other than to provide a brief written summary of what each side contended were the relevant or distinguishing features of each judgment. Most attention was focussed on the TBA decision and that of the UK High Court.

The Board of Appeal of the EPO – 27 October 2021

256. The European Patent (the “EP”) was granted with two claims, mirroring those in the Patent. A large number of oppositions were filed and the EP was revoked on a number of grounds including that it did not involve an inventive step. Bayer appealed that revocation to the TBA.

257. In considering the claim for lack of inventive step/obviousness, the TBA determined that the Blood Abstracts referred to above were the closest prior art. The posters, and in particular the Harder poster, were not referenced. Adopting the problem/solution analysis, the TBA concluded that the technical problem to be solved could be defined as providing a safe, effective and convenient oral dosage regimen of rivaroxaban for the prophylactic and therapeutic treatment of thromboembolic disorders.

258. The TBA decision indicates (at 9.19) that, based on the prior art, the person skilled in the art would have had a general expectation that the drug would provide clinical efficacy for this indication and that there “was agreement among the parties that the skilled person would have made the transition from phase I to phase II clinical testing.”

259. It noted (at 9.22.1) that “the court case differs in this aspect from the typical situation in other “dosage regimen cases”, where development is based on established therapeutic uses of the drugs concerned.”

260. It concluded (at 9.22.4) that “the serious concerns about safety in the case of new anticoagulant did not warrant a “try-and-see” attitude for the dosage regimen and the known, relatively short, half-life of rivaroxaban would not have supported an expectation of success with regard to once-daily dosing a rapid release rivaroxaban.” It is worth observing that the conclusion by the TBA that the half-life of BAY 59 was “known” based on the statement to that effect in the Blood abstracts related to the Kubitza studies is completely at odds with the approach advocated by Bayer’s witnesses where there is any data shortfall.

261. At 9.22.7, it noted that the statement in the blood abstracts that relevant changes in the pharmacodynamic parameters were still present after 12 hours but concluded that this could not change the conclusions based on general safety considerations and half-life:

“This is because it is not possible to infer from the statements that the effects observed will be maintained for longer than 8 to 12 hours and would also suffice to maintain the required level of anticoagulation the patient at risk thromboembolism over 24 hours.”

262. In relation to the Harder abstract, which contains a statement that a sustained effect of BAY 59 on thrombin generation for up to 24 hours had been observed, it rejected the argument that this provided a person seeking a solution to the technical problem with an expectation of success for once-daily dosing:

“The skilled person had no reason to assume that the data and statements in [the Harder abstract] were incorrect. However, owing to a lack of an established correlation of the asset parameters with clinically relevant effects (thrombosis and bleeding), it would not have been possible to make any predictions regarding dosing frequency on this basis.”

263. Though the proceedings before the TBA took place over two days, 26 – 27 October 2021, i.e. immediately prior to the decision, the respondents to the appeal did not participate.

Nor is there any reference to examination of witnesses. Rather there is a reference to a declaration of Dr Misselwitz dated 15 January 2018 and an expert declaration of Prof Dr Sylvia Haas.

Norway – Sandoz v Bayer – 9 June 2023

264. The patent was upheld by the Oslo District Court.

265. In that jurisdiction, the parties were agreed that the Kubitza studies were the closest prior art. The Oslo court considered that it wasn’t obvious to rely on the Harder poster for the solution of the technical problem, referring to the size of the study, the absence of PK data, the uncertainty of the results and the absence of correlation of the assays with clinical effects.

266. The Oslo court agreed with the reasoning of the TBA on dosing.

267. There were expert witnesses on both sides and the case was heard over 5 days. The

decision is subject to appeal.

The Netherlands – Sandoz v Bayer – 1 November 2023

268. The District Court of the Hague upheld the patent. It followed a one-day hearing attended by experts on both sides, including Sebastian Harder on behalf of Sandoz, during which the court asked questions. The prior art in issue included the Harder poster and abstract.

269. The judgment refers in detail to the decisions of the TBA and the Norwegian court referred to above.

270. The Dutch court rejected the idea that finding a convenient dosing regimen was part of the problem to be solved, and concluded that the only problem was finding a safe and effective oral dosing regimen for BAY 59. The court concluded that although the skilled person would consider the suggestion of once daily-dosing because of the reference in Harder, it would not consider the study sufficiently reliable to give rise to a reasonable expectation of success, citing the small sample size, the experimental nature of the assays and the variation in the placebo arm. It also relied on the known results from the razaxaban trials, that it was “generally held that a substance must be administered every 1 – 2 halflives” and that the skilled team would assume a half-life of 4 – 6 hours based on Kubitza, but that once-daily wouldn’t be tried even on the basis of the Harder half-life. It rejected the claim of a ‘read across’ from enoxaparin.

271. Reference was made to Bayer’s actual experience in developing the drug, though this was clearly not central to the court’s consideration.

Sweden – Sandoz v Bayer – 2 January 2024

272. The Patent and Market Court held that the patent was valid having considered the prior art, including the Harder poster and abstract. The court relied on written expert evidence.

273. The court noted uncertainties regarding the data in the Harder poster by reference to the evidence of one of the Bayer witnesses and concluded that the Harder poster does not “provide the person skilled in the art with enough scientific support in the sense that they bridge the doubts that the expert had regarding odd dosage.”

Germany – Hexal AG and Ors v Bayer – 3 January 2024

274. The Federal Patent Court upheld the patent in a short preliminary opinion from which Bayer quotes the following in its submissions:

“With regard to the appropriate dose and dosing frequency, the skilled person is as a matter of course guided by all clinical Phase 1 studies available to him. In doing so, the skilled person takes into account that thromboembolic disorders are particularly serious and life-threatening disorders and that these disorders can occur in patients, in contrast to the healthy test subjects that were tested to date, both in the case of underdosing and overdosing. Since the solution of the task nevertheless required the first-time administration of rivaroxaban to patients who needed effective protection against both life-threatening bleeding and life-threatening thromboembolism, the skilled person will carefully examine the available data situation when selecting the dosing regimen. In doing so, the skilled person will note that the clinical trial [in the Harder Poster and Blood 3003] bases its results on tests that do not allow him any judgment about the duration of a possible efficacy in patients. This is because on the priority date there did not exist any correlations or target values for these tests for ensuring effective and safe anticoagulation, which the plaintiffs were also unable to prove, and whereby these tests were even carried out with the blood of healthy volunteers. On the contrary, the literature contains information that confirms this lack of correlation and the associated lack of meaningfulness in terms of therapeutic efficacy. For this reason, the skilled person will not further consider the conclusion explicitly formulated in [the Harder Poster] regarding the possibility of an OD regimen, but will be guided by the plasma concentration half-lives of rivaroxaban according to [the Kubitza Posters, Blood 3004 and 3010]. The skilled person will therefore, based on the 4 to 6 and 3 to 4 hours specified therein, continue the BID and TID regimen of the [Kubitza MAD] study with a view to ensuring a sufficient amount of active substance in the body. This is because it is standard practice to devise proposals for dosing regimens for the clinical Phase II studies from the half-life determined in clinical Phase I studies …”

The Netherlands – Teva v Bayer – 27 March 2024

275. A similar challenge to the patent by Teva was also rejected by the District Court of the Hague following a similar procedure to the Sandoz challenge and based on the same prior art. The court’s judgment refers to the TBA decision and those of the Norwegian and German courts described above. Oddly, it doesn’t refer to the earlier decision of the same court.

276. It concluded that the skilled person would have rejected the suggestion in Harder regarding once-daily dosing because the risk of underdosing was too high. It noted that the type of tests used in Harder were not intended or validated to test the antithrombotic effect or bleeding risk of direct factor Xa inhibitors.

Lithuania – Sandoz and Ors v Bayer – 26 March 2025

277. The Vilnius District Court also upheld the patent.

278. It had regard to the Harder poster in addition to the Kubitza posters but was not persuaded, notwithstanding the views of its own court-appointed expert, that Harder justified departing from the decision of the TBA, having regard to the appropriate level of caution suggested by the TBA. It concluded that two or three times daily dosing would have been tried and that once-daily dosing was unexpected.

279. It appears that the decision was based on documentation only, including expert reports, and is subject to appeal.

280. The patent has also been upheld in Belgium, Croatia, Czechia, Portugal, Slovakia (in three separate challenges) and Sweden. In Croatia, the Intellectual Property Office concluded, in a decision dated 27 January 2025 that the prior art relied on in that case, including the Harder poster, could “at best, give the expert in the field hope for success that once daily dosing of rivaroxaban when administered in fast-release tablets will be safe and effective in patients, but this in no case constitutes a reasonable expectation of success.”

France – Sandoz v Bayer – 28 March 2024

281. The patent was declared invalid by the Paris Judicial Tribunal. The decision is under appeal. There was a one-day hearing.

282. The French court concluded by reference to Harder that it “strongly encouraged a person skilled in the art to consider once-a-day dosage, at least as a research alternative to the double dose referred to in” Kubitza. It stated that the court could only find that “the real difficulties raised by Bayer relate not to the solution proposed by the patent itself, but by the Phase II tests envisaged to confirm it.” It therefore rejected Bayer’s argument that there was no reasonable expectation of success for once-daily dosing. It found that the solution to the problem of a single daily dose was therefore a routine trial and not surprising or unexpected.

283. The decision is under appeal.

Switzerland – Sandoz v Bayer – 17 January 2025

284. The patent was deemed invalid by the Federal Patent Court, based on the same prior art as in issue in these proceedings. It appears that following detailed exchanges, including the provision of a Technical Opinion by a patent judge, there was a one-day hearing. The

decision was under appeal at the time of hearing of these proceedings but that appeal has since been rejected. The parties are agreed that that decision is final.

285. At page 46 of the first instance decision, the court addresses Bayer’s argument “at great length” that the skilled person would attach no significance to the Harder results because the assays were experimental, non-standardised, not validated for BAY 59 (not correlated with clinical effect) and not statistically significant.

286. The argument was described as “obviously not convincing” if only because it cannot be seriously claimed that the studies would have been done with assays which would ultimately be meaningless. It considered that the assays used were “very suitable for measuring thrombin generation”. Clinical effect could not be determined without Phase II studies, but these were sufficient qualitative indications for the next phases.

e

287. The Swiss court relied on post-patent publications which contradicted the evidence given by Bayer’s witnesses at the hearing regarding the suitability of the assays.

UK proceedings – Sandoz v Bayer

288. The decisions which were the subject of most debate during the course of these proceedings were the UK decisions. The High Court declared the patent invalid in a judgment dated 12 April 2024 ([2024] EWHC 796 (Pat)). The Court of Appeal rejected an appeal in a decision dated 24 May 2024 ([2024] EWCA 562 and leave to appeal was refused by the Supreme Court on 14 October 2024.

289. In the UK proceedings, the patent was challenged on a number of grounds, including obviousness. As noted above, Bayer also contended that the skilled team would not have been able to discover the chemical code for BAY 59, and thus would not have been enabled it to take any step in relation to it. This was rejected by the High Court and not pursued in the Court of Appeal.

290. In his judgment, Judge Hacon addresses the so-called “invention story” in some detail, noting that it had been explained by Dr Misselwitz in the Belgium proceedings, where there was no cross-examination, and where it appears to have been accepted. However, when Dr Misselwitz was cross-examined in South African proceedings, the explanation was rejected. Of course, no evidence was led of the invention story in these proceedings, so its main relevance here is as a purported point of distinction with the foreign judgments. Sandoz also argues that the lack of any alternative explanation for the inventive concept undermines Bayer’s argument that it was not prompted by the information in the prior art.

291. In assessing the evidence in relation to common general knowledge, Judge Hacon rejected the argument that the skilled team would have read across from LMWHs, and concluded that any information there may have been that FXa inhibitors had a wide therapeutic window was not common general knowledge at the priority date. Neither issue ended up being disputed in these proceedings.

292. It appears that Sandoz made the same complaint in those proceedings regarding Bayer’s experts’ disregard of uncorrelated assays as it makes here, that Phase II trials could never go ahead if a correlated assay was required. Judge Hacon did not interpret Bayer’s evidence in that way, although he accepted the underlying proposition by Sandoz – that this must be wrong (at paragraph 160):

“I did not understand Professor Crowther to be saying that. The go-ahead for a phase II trial can be justified by appropriately informative data from one or more phase I trials. If this was not what Professor Crowther meant, the clear bulk of the evidence proves it to be the case.”

293. The judgment refers in detail to the evidence of Professor Meibohm referable to the graphs shown above in the discussion of steady state.

294. It appears that in the UK, Bayer relied on differences between the prior art and the inventive concept, such as the absence of mention of a rapid release form, or the requirement for at least five days dosing in the prior art. Judge Hacon did not consider these significant and concluded that the real issue was whether the skilled team would have considered it obvious to conduct a Phase II trial with once-daily dosing having read Harder and Kubitza.

295. At paragraphs 230 to 239 he summarises the “headline” arguments and the disputed evidence in relation to that question, much of which appears to mirror what has been argued here.

296. Noting that Bayer had argued that the skilled team would have proceeded with caution, he sought to determine what level of caution was appropriate. He considered it mirrored that of a team seeking approval from an ethics committee and that whether something was obvious to try could be considered by asking whether the skilled team would have thought it was worth asking an ethics committee for approval.

297. Having weighed the evidence, he concluded that the skilled team would have considered it obvious to try once-daily dosing.

298. He also addressed a suggestion by Bayer that it must have had more data than disclosed in the Harder and Kubitza posters on which to design a Phase II trial (at paragraph 260):

“At one point counsel for Bayer submitted that Bayer must have had more data than is reported in Harder and the Kubitza posters on which to base their design of a phase II study. That may be, but it does not follow that the remaining data added up to anything materially different to what is reported in Harder and the Kubitza posters. Also, it would be odd if there was something materially different and therefore significant to a convincing explanation of the claimed invention, yet Bayer chose not to mention it in this court.”

299. Bayer’s appeal was rejected. One of its grounds of appeal related to the reliance by the trial judge on the analogy with an ethics committee. The Court of Appeal rejected this argument. Although it agreed that the views of an ethics committee were irrelevant, it considered that, in substance, the High Court judge had applied the correct test. The balance of the grounds were dismissed as being disputes about the judge’s factual findings without identifying errors of principle. In addition, the Court of Appeal rejected the argument that the trial judge had placed weight on the failure by Bayer to adduce evidence of how it came

up with the invention. This was rejected on the basis that the High Court had not, in fact, placed any weight on this factor.

Non-EPC countries

300. The patent has been declared invalid in both Australia and South Africa. In Australia, the validity of the patent was upheld at first instance, but this decision was reversed on appeal. The decision of the Australian courts appears to be final, as leave to appeal has been refused by the High Court in a decision dated 6 March 2025. The decision of the South African court was referred to by the UK High Court in the course of its judgment. The first instance judge granted leave to appeal in a decision dated 5 July 2024.

Assessment of foreign judgments

301. Although both parties sought to rely on the decisions where the foreign court’s judgment reflected its position in these cases, neither identified a single decisive basis for arguing why the decision in these proceedings should follow the decisions upholding or invalidating the patent elsewhere.

302. Sandoz highlights that it is clear from many of the judgments that some evidence of the so-called invention story was provided in many other jurisdictions. No such evidence was led here and, therefore, the decisions where the courts were provided with an invention story, i.e. an explanation for the inventive concept other than obviousness, can therefore readily be distinguished. On proper analysis of those decisions, this is a very slight basis for distinction. The invention story, or some version of it, may have been part of the background narrative in many cases, but it doesn’t seem to have been expressly relied on in any jurisdiction in the assessment of obviousness, save in South Africa, where it appears to have been rejected as not credible. In the circumstances, I do not place any weight on the presence or absence of the invention story.

303. The prior art relied on here was relied on in many jurisdictions but in some, such as Slovakia and Portugal, the Harder poster was not assessed. Given the centrality of that prior art to the debate in these proceedings, that creates an obvious point of distinction. Of course, the Harder poster was not discussed in the TBA decision, and therefore the strong persuasive force that body’s decision would ordinarily have is somewhat mitigated.

304. But in most jurisdictions where the validity of the patent has been considered, Harder was discussed, even if not accepted as the closest prior art. If a common feature can be identified in all the cases, it is that the decision-maker has engaged with the evidence regarding the weight to be given to the Harder poster. Where persuaded by that evidence that reliance can be placed on Harder, read together with Kubitza, the patent has been declared invalid. Where not so persuaded, the validity of the patent has been upheld.

305. There have been different witnesses in almost all cases to those here, save the UK and Lithuania. Sandoz has indicated that in the Lithuanian proceedings, the reports submitted on Sandoz’s behalf in the UK proceedings were provided and Professors Crowther and Meibhom prepared replying reports. There is no express reference to any of them in the Lithuanian court’s decision or to the UK court’s decision.

306. In two EPC jurisdictions the decision is now final, both declaring the patent invalid.

307. The UK proceedings involved the same witnesses and there was detailed crossexamination. The relevance of the decisions there is obvious, especially having regard to the observations in Ranbaxy and Gilead referred to above. However, the decisions overall serve to highlight that the question of validity turns, to a large extent, on the conclusions which can be drawn from the evidence regarding how the skilled team would read the prior art, in particular, the Harder poster in light of the Kubitza posters. The extent to which the decisions in other jurisdictions guide the decision of this court, therefore, can only be resolved by considering the extent to which the conclusions on the evidence given here mirror the conclusions on the evidence that was heard by other courts.

Approach to assessment of evidence

308. I have already referenced above some of the factors relevant to the assessment of the evidence.

309. Each of the witnesses was highly skilled and communicated their evidence in a clear, comprehensible way. The fact that, together, the Bayer witnesses each seemed to suggest that the other’s discipline would have initial responsibility for proposing dosing arms was a little confusing. That said, Professor Crowther was very clear that, on the basis of the prior art, he couldn’t recommend any dosing arm, and Professor Meibohm, at least in his oral evidence, supported him in that.

310. In demeanour, Professors Hirsh and Wilkins both seemed entirely comfortable with their overall evidence that once-daily dosing was obvious over the prior art; indeed they were somewhat at a loss as to how the Bayer witnesses could conclude otherwise. Professor Crowther came across as equally sure of his ground, though he did, repeatedly, stress that it was the absence of relevant data which was critical to his conclusion that once-daily dosing was not obvious.

311. Professor Meibohm also maintained his position regarding once-daily dosing, notwithstanding that he made important concessions in his evidence, in particular, as regards the evidential weight to be afforded to uncorrelated biomarkers. There was also a significant change in his evidence – regarding what dosing arms might be proposed on the basis of the prior art – which was not adequately explained. Professor Meibohm’s evidence, therefore, was somewhat less convincing than that of the other experts and was less internally consistent.

312. Bayer has highlighted that Professors Hirsh and Wilkins also made important concessions.

313. In relation to Professor Wilkins, one criticism made relates to paragraph 331 of his first written report in which he refers to Professor Hirsh having explained to him that the risk of symptomatic VTE is low and of fatal PE is very low. Professor Hirsh clarified what this meant in his evidence, as set out above. Nonetheless, Bayer asserts that Professor Wilkins had, in effect, underestimated the risk level of underdosing. This is not supported by the evidence. Professor Wilkins was clearly aware that underdosing was a risk, but he made clear that in proposing appropriate levels based on his understanding of the biological relevance of what he described as “what appears to be a real observation” reported in Harder, that would be subject to discussion with the skilled clinician (see, for instance, Transcript Day 6, pp. 112 – 121) and the skilled clinician would have the final say.

314. In relation to Professor Hirsh, Bayer relies on a number of matters used to support his views on the prior art which it was said he had “resiled from”. In their submissions, Bayer identifies the ‘read across’ from LMWHs, the assumption of a wide therapeutic window, the proposition that joint surgery was not high risk, and finally, that it was common general knowledge that BAY 59 had entered Phase II trials. The first three have been addressed above. All relate to common general knowledge and are factors which can be weighed in the balance in considering the obviousness of once-daily dosing over the prior art. Professor Hirsh did not, however, place anything like the weight on these factors as did the Bayer witnesses on the fact that the assays were not correlated.

315. As regards the last, this is the only jurisdiction in which that complaint could be made:

in every other jurisdiction, it has been accepted that BAY 59 would be taken into Phase II trials. Whether it was common general knowledge that this had already happened would, therefore, have been of no real significance at all. It has no significance independent of the question of whether, on the basis of the prior art, the skilled team would have thought it obvious to try Phase II trials of BAY 59 at all.

Discussion

316. Whether one applies the four-step Pozzoli test, or the problem-solution approach favoured by the TBA, the question to be addressed is simply whether once-daily dosing would have been obvious to the skilled team in light of the prior art. In this respect, there seems to be little dispute between the parties that, in answering that question in this case, it is appropriate to consider whether the skilled team would have considered once-daily dosing obvious to try with a reasonable expectation of success.

317. A significant difficulty for Bayer in the instant proceedings was that the prior art relied on by Sandoz in this case does not simply provide a basis for arguing that the Patent was obvious, it expressly identified the invention which Bayer subsequently patented, once-daily dosing. The fourth conclusion of the Harder poster was that some parameters “suggest suitability for a once-daily dosing regimen”. The prior art, therefore, expressly called the inventive concept to the mind of the skilled team. This is acknowledged in many of the foreign judgments, including those upholding the validity of the Patent.

318. Nor is there anything in that inventive concept which is in any way complex or contrary to what a skilled team would be predisposed to think about a novel drug. Once-daily dosing is, in one sense, the simplest of dosing regimens – this simplicity is the very thing which makes it attractive. There was no evidence that once-daily dosing of drugs, even of anticoagulants, was so unusual that it would be rejected out of hand by the skilled team, unlike the bagless vacuum cleaner at issue in Dyson.

319. In Inhale, the court noted that a document directed at solving the particular problem at issue will be “seized upon” by the skilled team. Adopting the problem – solution formula for a moment, the problem has been defined by the TBA as finding a “safe, effective and convenient oral dosage regimen of rivaroxaban”. The Harder poster expressly suggests a solution for that problem. It would undoubtedly have been seized upon by the skilled team looking for such a solution. There is, accordingly, simply no question but that the possibility of once-daily dosing would have occurred to the skilled team having read the prior art: the prior art identified that possibility in clear terms. The suggestion in the Patent that oncedaily dosing was “surprising” is at odds with prior art which explicitly suggests its suitability.

320. Bayer’s defence, therefore, is not that the inventive concept wouldn’t have occurred to the skilled team, but rather (i) properly read by the skilled team, the prior art wouldn’t be viewed as supporting the conclusion stated in the prior art, and (ii) even if the conclusion was consistent with the information presented, the skilled team did not have adequate information to place trust in that conclusion: it wouldn’t have placed sufficient trust in the prior art to believe that once-daily dosing was worth trying with a reasonable expectation of success.

321. Professor Crowther neatly summarised Bayer’s position on the first of those propositions when he stated (Transcript Day 11, p. 17) that there was “nothing in [the Harder] poster which suggests to me that this product would be safe and effective to introduce to patients on a once-daily dosing regimen.”

322. The second proposition is captured in the following way in Bayer’s submissions (paragraph 173):

“The assessment of the evidence on what the Harder poster would communicate to the [skilled team] comes down to whether the [skilled team] as … scientist[s], looks to read the information actually given, or whether they infer the existence of certain data, on the basis of a suggestion of suitability for once-daily dosing.”

323. Bayer’s position as to the correct approach is stated (at paragraph 37):

“[For] the purpose of the exercise that the Court has to undertake in order to determine whether there is an inventive step, prior art documents cannot be supplemented with extra information that they do not disclose or subjected to a legal fiction that, in addition to having the prior art document, the skilled team are deemed to have access to whatever underlying data gave rise to that particular document.”

324. Issue is taken in the submissions with the making of “impermissible assumptions” by Professor Hirsh. The following (Transcript Day 3, p. 27) is highlighted:

“There’s no way that any of these studies could have been done without the appropriate animal data and without the appropriate toxicology data.”

325. It is difficult to understand what could be objectionable about such an assumption, which objection, moreover, it can readily be seen, is somewhat at odds with the acknowledgement in oral submissions that it wasn’t in issue that “that somebody could do Phase I routine trials and get data similar to the Kubitza posters.”

326. There was no dispute amongst the experts regarding the progress of drug development trials. Phase I trials, the first trials in humans, are always preceded by pre-clinical trials. Indeed, it appears that many of these pre-clinical trials are regulatory requirements, though the precise nature of these regulatory requirements were not discussed in the evidence at all. The Kubitza and Harder posters made clear that they related to Phase I trials, i.e. trials in humans. It simply must have been the case, therefore, that pre-clinical trials had been completed and had produced results enabling the drug to be brought forward to Phase I trials. Any skilled team reading Kubitza and/or Harder would not simply assume, they would know, that pre-clinical trials had produced satisfactory results in relation to safety and toxicity.

327. Moreover, there was no suggestion that, had the skilled team been motivated to bring forward BAY 59 for development, it could not, as necessary, readily have conducted those trials and obtained similar results, just as was expressly accepted in relation to the Kubitza SAD and MAD studies. The evidence regarding those studies was that they were entirely standard.

328. This is consistent with the decision in Actavis in which the UK Supreme Court rejected a test for obviousness which “would require the court to disregard the work which a skilled person would carry out after the priority date in order to implement the teaching of” the prior art. True, that was a case regarding whether the patent was obvious because routine steps would have inevitably brought the skilled team to the patent from the prior art, but if routine steps can be relied on to get the skilled team to an outcome that has not been described, i.e. the inventive concept, it must be the case, that, if necessary, it can be assumed that by routine steps the skilled team can at least get to the prior art, an outcome which has been described. Of course, if getting to the prior art is not something within the competence or ability of the skilled team, then no such assumption can be made. Nor does it permit any assumption to be made which is not dictated by the prior art. So, in this case, the skilled team are entitled to assume positive pre-clinical results – otherwise there would have been no Phase I trial – even though they are not described, but cannot assume anything about the results from, for instance, the thrombin generation study which the skilled team would not otherwise understand from the results reported.

329. In considering the question of obviousness, I am obliged to consider the foreign judgments which have been delivered in proceedings challenging the same or an equivalent patent and to respect the principle of judicial comity. The parties have also relied on socalled secondary evidence.

330. As noted above, the extent to which judicial comity is relevant to the decision here depends to a material extent on the evidence led and the conclusions on that evidence. In relation to the foreign judgments, although there are references to some of the evidence led, all that is apparent from the judgment are the conclusions which the courts reached in arriving at those judgments. Accordingly, I propose, therefore, to set out my conclusions on the issues in dispute here, and then test those conclusions in light of those foreign judgments.

331. As regards the secondary evidence, given the dispute about the evidential status of the documents, I think that the safest course is to reach a conclusion without reference to them in the first instance. I will, however, then set out my views on the status of the documents and, insofar as they are admissible or relevant, consider their impact in light of the principles identified in Molnlycke.

Teaching of prior art

The Kubitza posters

332. There was very little in dispute regarding the interpretation of the Kubitza posters. A number of issues were stressed by Bayer in its submissions and its evidence regarding the interaction of the Kubitza and Harder posters. First, Bayer highlights that the shorter halflives reported in the Kubitza posters were supported by data, whereas the half-life reported in Harder was, in effect, a “bare assertion”. Second, it highlights Professor Wilkins’ agreement that the Kubitza poster reports a close correlation between pharmacokinetic and pharmacodynamic results. In the absence of disassociation, this supports the idea of halflife based dosing. Professor Wilkins gave examples of where such disassociation occurs, but did not explain how BAY 59 fit with any of those examples, most of which clearly didn’t apply. And it contends that Professor Wilkins’ evidence in relation to the interpretation of Figure 1 from the Kubitza SAD poster, reproduced above at page 34, differed materially from the evidence in the UK proceedings, providing a reason to distinguish the decisions there. I will take each of these in turn.

333. It is true that the half-lives reported in both Kubitza posters are supported by data and were studies on a larger group of people than the Harder study, whereas the Harder poster contained no data to support the figure given for half-life. It is also the case that determining half-life, or the pharmacokinetic properties of BAY 59, was one of the objectives of the Kubitza studies, though not the Harder study. However, that would not have caused the skilled team to entirely disregard what was said about half-life in the Harder poster for a variety of reasons. First, the half-life reported in Harder is reported in the opening paragraph of that poster, suggesting the authors attached some significance to it. Moreover, Professor

Crowther accepted as follows, having regard to the identity of the authors (Transcript Day 12, p. 12):

Q. … Would the skilled clinician have an expectation that the work shown in this poster would be a thorough, highly competent and put forward honestly and in good faith?

A. That's reasonable, again with the slight limitation that I would always be a little bit sceptical.

334. He also stated that there would be an expectation that the authors would report the facts as they understood them and that they would have understood the significance of the statement about half-life (see Transcript Day 12, p. 15). Professor Meibohm agreed (Transcript Day 8, p. 51) that the pharmacodynamic data in the Harder poster wasn’t inconsistent with the reported half-lives given in either the Kubitza posters or the Harder poster.

335. The identity of the authors is, I think, of some significance. First, the Harder poster was prepared having regard to the results of the Kubitza studies – this must be the case as it refers to the results of those studies – and was prepared by a group containing five of the six authors of the Kubitza posters which were, in effect, being presented alongside it. The skilled team would not ignore that fact and simply attribute weight to the reported half-life in Kubitza, because of the accompanying graph, and none at all to Harder. To do so would be illogical. More likely, it would consider that there was uncertainty regarding the half-life.

336. Moreover, one of the additional authors of the Harder poster i.e. not one of the authors of the Kubitza posters, was Professor Hans Klaus Breddin, a world-leading expert in the field of thrombosis. Although Bayer’s witnesses stressed the somewhat informal nature of these conference posters, which are not peer-reviewed, with the corresponding abstracts subject only to limited oversight, it is not credible that the skilled team, reviewing the posters, would have readily dismissed any of the data reported in the Harder (or Kubitza) posters. Rather, they would have the expectation that it was accurate. As noted above, the

TBA accepted that the half-life of BAY 59 was “known” based on the Kubitza abstracts alone.

337. I note that in many of the foreign judgments upholding the validity of the Patent, the relevant court has concluded, by reference to the presence of supporting data in Kubitza and the relative size of the studies, that the skilled team would, in effect, proceed on the assumption that the half-life of BAY 59 was as supported in Kubitza. Even if the skilled team considered that the information in Kubitza was more robust, for the reasons stated above and in light of the evidence in these proceedings, it would not have simply dismissed the statement about half-life in Harder: it would not have considered it had a sufficient basis for so doing.

338. It could be said that Bayer’s witnesses’ ‘faith’ in the data reported in the Kubitza posters stands in marked contrast to their scepticism about the data reported in Harder. Perhaps the better point, at least on this issue, is that any apparent inconsistency in data, whether within or between posters, undermines the faith in the data overall.

339. There is, for instance, an inconsistency between all three posters regarding the half-life of BAY 59, and a corresponding inconsistency that all the posters report correlation between pharmacokinetic and pharmacodynamic data. In the UK proceedings, the judgment reports that the detail of these graphs was not addressed, and Bayer draws a point of distinction here. The point doesn’t go anywhere. All the parties agreed that half-life can be difficult to calculate and, in any event, some variability can be expected. Insofar as none of the experts highlighted the discrepancy between the half-lives reported in the two Kubitza posters, it appears that this degree of variability was entirely expected. As regards the larger discrepancy between Kubitza and Harder, Professor Wilkins said (Transcript Day 6, p. 93):

A. … And so we have half-life data reported from Bayer - I think Bayer authored both of the studies. In one they've suggested a half-life of 4-6 hours. In another they've suggested a half-life of 9-12 hours. So, I'm assuming the half-life is somewhere between 4 and 12 hours. And I am not uncomfortable with that variation - it's not ideal, but it does happen in early phase studies.

340. As he also pointed out, both studies are internally consistent insofar as they suggest correlation between PK and PD results.

341. Insofar as it is suggested that Professor Wilkins’ evidence in relation to Figure 1 differed materially from that given on behalf of Sandoz in the UK, I do not believe that is made out. In his written report here, he says that the “30 mg and 40 mg doses show at least 10% inhibition at 24 hours, with the 60 and 80 mg exhibiting approximately 20% inhibition of FXa activity.”

342. His evidence in cross-examination was in the following terms (Transcript Day 6, p. 25):

A. -- is close to the square that indicates the 40 mg dose. Both myself and Prof. Hirsh read that as, you know, the possibility that there is 10% inhibition of activity out to 24 hours. We don't know whether that's significant, we don't know whether that's, you know, biologically or clinically relevant, but it does seem to be apparent from the graph.

343. Counsel queried his use of the term “possibility” and he explained Transcript Day 6, p. 26):

A. I would look at the graph, these are the data they're presenting and I would make that inference from the graph that there was some activity out to or inhibition of activity out to 24 hours. That's all I'm saying.

344. There is little between his oral and written evidence. More pertinently, given the suggestion by Bayer that this is a significant point of distinction between these proceedings and those in the UK, nor is there any material difference between what was said by Professor Wilkins here and what is reported about his evidence in the UK High Court judgment. After pointing out an apparent inconsistency between the correlation of pharmacokinetic and pharmacodynamic effect and inhibition of FXa after 24 hours, the judgment records (paragraph 253):

“As against that, it was resolved in the evidence of Professors Hirsh and Wilkins, and conceded by Professor Meibohm, is that the SAD Kubitza study shows graphical evidence of anticoagulant activity 24 hours after administration, including 24 hours after administration of a 30 mg dose.”

345. The simple fact is that that is what the graph does appear to show. It is true that the authors only referred to higher doses having a sustained effect and that the lower doses, including 30 mg, returned to normal range within 24 hours, a fact to which the skilled team would certainly have regard when considering the prior art, but they would also consider the evidence from the graph. This was the consistent evidence from both Professors Hirsh and Wilkins, which I accept, that the skilled team would assess the graphs presented in conference posters, not just what was reported about them. Insofar as Bayer’s witnesses seem to suggest that the skilled team would only take from the graphs what was highlighted in the text, this was not convincing. Though Professor Hirsh is criticised for saying that he would have “eyeballed” the graph, I think he meant no more than that he would have been able to read certain data from the graph. Professor Meibohm used the same terminology to the same effect (Transcript, Day 8, p. 43).

346. I agree with Bayer’s witnesses that, on the basis of Kubitza alone, the skilled team would not have dismissed what the authors had said in the text simply by “eyeballing” the graph and concluded that the poster supported once-daily dosing. However, when considered in light of Harder, which clearly raises that prospect, I think that the skilled team would have had regard to the fact that the graph in the Kubitza SAD poster was not at odds with the graphs in the Harder poster.

347. Bayer’s witnesses suggested that Figure 1 in Kubitza and other graphs couldn’t be relied on because the representation of the placebo was not zero, but that seems to me to be consistent with Bayer’s own evidence regarding the sensitivity of testing, the variability in tests and Professor Hirsh’s evidence regarding assays having lower limits of detection (Transcript Day 4, pp. 103 – 104). Although the proposition by Bayer seems to be that any effect detected couldn’t be assumed to be real due to the limitations of the test, this seems to miss the point which is that the information of interest is the difference in the results between the placebo and the BAY 59 dose.

348. The objective of the studies reported in the Kubitza posters was to evaluate the safety, tolerability, pharmacodynamic effects, and pharmacokinetic profile of BAY 59. Having conducted those studies, the authors concluded, inter alia, that it was “safe and well tolerated at all doses” and that “the good safety profile, selective inhibition of FXa, and promising pharmacokinetic and pharmacodynamic profiles indicate that further investigation of BAY 59 is warranted.” None of the experts cavilled with this conclusion. The skilled team would have been enthused about BAY 59 in light of the Kubitza posters.

349. Before turning to the real battleground in the case, the Harder poster, I propose considering the question of whether the skilled team would have progressed to Phase II trials at all.

A Phase II trial?

350. At paragraph 188 of its closing submissions, Bayer lists fifteen matters which it suggests must be borne in mind when reviewing the evidence, all of which, on its case, undermine the claim for obviousness. It then contends (at paragraph 189):

“Bearing in mind the foregoing and having regard to the evidence given, it is Bayer’s case that Sandoz’s position on designing a dosing regimen based upon the prior art is simply incredible.”

* emphasis added

351. This is, indeed, a bold assertion. The Patent has been declared invalid in a number of jurisdictions, including the only three jurisdictions, the UK, Switzerland and (non-EPC) Australia, where the decision is final. A challenge based on obviousness is thus, evidently, not “incredible” (of course, it is equally clear that the Patent has been deemed valid in other jurisdictions, illustrating the strength of the claim for validity). Moreover, it is Bayer’s position regarding dosing in these proceedings which, it appears, is at odds with all other jurisdictions in which the Patent has been challenged. In every other case, the challenge has proceeded on the basis that BAY 59 would be brought to Phase II trials based on the prior art. In almost every jurisdiction where the patent has been upheld, the relevant court or patent office has expressly concluded that the skilled team would have tried dosing twice or three times daily based on the prior art. This necessarily calls into question the credibility of Bayer’s position on dosing and, indeed, their witnesses’ evidence.

352. At one level, the decision not to concede that point here can be understood: a strategy which did not succeed in the UK High Court, that of not standing over the evidence of their experts, was, in effect, relied on against Bayer in the Court of Appeal. Faced with the same dilemma here, one can understand the desire for a different approach, or at least the reluctance to make the same concession as was made there.

353. It is not, however, entirely a matter of strategy. The question of what a skilled team would do in light of the prior art is primarily a question of fact. As noted by Collins J in Norton v Boehringer, it would be “disquieting” for a party to adopt different positions on the same issue in different jurisdictions. It is, to say the least, unsatisfactory that Bayer has adopted a different position in this jurisdiction on a question of fact, than in every other jurisdiction. That said, it is a position consistent with its experts’ evidence, and Bayer is entitled to take a position in light of the evidence led. Accordingly, the real issue is whether the difference in position here than that taken in every other jurisdiction has a bearing on that evidence. I think it must. First, it calls into question the credibility of evidence which is at odds with a proposition which has not even been put in issue in any other proceeding. Second, it tends to show that the reason for Bayer’s witnesses’ contention that no dosing regimen could be proposed based on the prior art is at odds with the law on obviousness.

354. Although Bayer’s witnesses had a number of objections to the suggestion that the Harder poster made the inventive claim obvious, the two most consistent grounds of objection were (i) that the assays had not been correlated for clinical effect, and (ii) there was a data deficit. I have addressed the first of those in detail above. The lack of correlation was not a basis for rejecting the relevance of the assays. Insofar as the Bayer witnesses suggest that decisions would not be made by a skilled team based on uncorrelated assays, I reject that contention for the reasons explained. Furthermore, in circumstances where Bayer’s witnesses placed repeated emphasis on this fact in their written reports, notwithstanding the acceptance in evidence that this position was inconsistent with the “Bible” for drug development, the FDA Guidelines, and other relevant guidance, this undermines their evidence and suggests a determination to paint a picture which corresponds with their underlying thesis, or illustrates that their underlying thesis was based on a faulty premise.

355. The second issue has also been touched on: whereas as Professors Hirsh and Wilkins assumed, as instructed, that the skilled team would not have access to data generated earlier in the process, but could generate it themselves through routine methods, Professors Crowther and Meibohm seemed to have assumed the contrary; that despite giving evidence of what the skilled team’s abilities were, those abilities could not be used (Transcript Day 12, page 62):

Q. And what I understand you to be saying, Professor, is that if you take the combination of Harder, the Kubitza posters and/or the Blood Abstracts, that any combination of those would not enable the skilled clinician to devise a dosing regimen?

A. I agree with that. In order to do this, you need to have a much deeper understanding of the data that informed those. I'm not saying that it's not possible if they had access to the full data set, but based on the information that's presented, I just can't imagine how they would.

356. There is, I accept, a separate issue regarding whether the Harder study could be replicated. In addition, it is important to clarify that even if that study could be replicated, that does not entitle the skilled team to make any assumptions about that data other than as disclosed in the Harder poster. That said, it seems to me that, the thrombin generation study aside, the Bayer witnesses have taken far too narrow a view of their role in giving the views on the skilled team. There was no suggestion by any witness that the pre-clinical trials, which must have been done, or the Phase I trials reported in Kubitza were anything other than routine and part of a standard drug development pathway. It was also made clear that the SAD- and MAD-type studies are the prelude to Phase II trials. In those circumstances, Sandoz’s witnesses appropriately assumed that that data could be generated. Of course, no one could, or would be permitted to, conduct a Phase II trial without having completed preclinical and Phase I trials. There was no basis for assuming access to the results of Bayer’s data in any other jurisdiction. The assumption or conclusion in all other proceedings that the skilled team would move to Phase II trials must, therefore, have been on the premise that the skilled team could reproduce the standard Phase I trials described in Kubitza or the preclinical trials which must have preceded them.

357. It may be, therefore, that Professor Crowther is not an outlier in suggesting that there was no basis to recommend a dosing regimen, but rather, he was asking himself a different question than was asked in other jurisdictions: assuming I can’t provide support for what is reported in Kubitza, could I recommend a dosing regimen? If so, he was asking himself the wrong question.

358. If that is so, Bayer’s witnesses’ ‘justification’ for concluding that no dosing regimen could be designed falls away. All the evidence was that the studies which were reported in Kubitza and the pre-clinical studies which must have preceded them were the standard path to Phase II trials. The Kubitza posters were overwhelmingly positive about BAY 59, which was “shown to be safe and well tolerated at all doses”. As Professor Crowther acknowledged, he would have hoped, on the basis of what he had seen in the prior art, that BAY 59 would have been progressed to Phase II trials. No witness identified any barrier to moving to Phase II trials other than the absence of data and the absence of correlated assays. It will be recalled that Professor Meibohm’s evidence, before it changed for wholly unconvincing reasons, was that he would have recommended a dosing regimen of 20 mg and 30 mg bid for a Phase II trial, notwithstanding the absence of data and correlated assays. His reason for resiling from that opinion did not hold water. The clear import of Bayer’s witnesses’ own evidence is that, absent these supposed obstacles, a dosing regimen designed by reference to half-life data would have been recommended. Based on the Kubitza posters alone, the skilled team would have considered it obvious to advance to Phase II trials with a reasonable expectation of success. The Kubitza MAD poster states, in terms, that BAY 59 is suitable for twice-daily dosing. No one has suggested that this was inconsistent with the data presented in the Kubitza posters. There would thus have been an obvious path for the skilled team to follow.

359. There remains the overriding issue that, in order for the skilled team to take that step, the skilled team would have to be sufficiently motivated by the prior art to think that it was worth generating the necessary routinely obtained data. This is reflected in the factors identified in Actavis relevant to determining the question of obviousness. Put otherwise, would the necessity of having to generate the routine data have negatively affected the skilled team’s expectation of success. In relation to the pre-clinical and routine Phase I data reported in Kubitza, the answer is ‘no’. Would they have considered it worth the effort? The evidence of the unmet need for an oral anticoagulant which did not require monitoring strongly suggests that the skilled team would have, if satisfied that BAY 59 was likely to be safe and effective. The burden and cost of developing novel drugs is significant in every case; it is not a basis for concluding that the skilled team would not have thought it worthwhile to advance to Phase II trials. The evidence here – including Professor Crowther’s “hope” that BAY 59 would have been progressed to Phase II trials based on what he saw in the prior art – clearly suggests that they would. Indeed, Professor Crowther agreed (Transcript Day 12, p. 42) that based on the posters “that the drug based on this should proceed to further clinical trials.”

360. Of course, based on the prior art, the skilled team would have to have considered Harder and the Blood abstracts before forming any expectation about Phase II trials. I have, in the discussion so far, not touched on the relevance of the Blood abstracts. As to content, it seems to me that they do not add or subtract (by introducing uncertainty) anything in the material contained in the relevant posters save, as discussed below, the reference in Blood 3003 (the Harder abstract) to the fact that a single dose of 30 mg exerted a sustained effect in some assays of thrombin generation for up to 24 hours. However, the Harder abstracts and one of the Kubitza abstracts were on the same page in Blood. It is realistic, therefore, to consider that the skilled team coming to Kubitza would also have been aware of Harder. If sufficiently motivated to pursue the teaching of Kubitza, it is not credible that they would ignore the Harder abstract and poster or not investigate their teaching.

361. It might be said that it is premature for me to conclude that, based on Kubitza, the skilled team would have proceeded to a Phase II trial. However, given Bayer’s position is that little weight can be afforded to the Harder poster, a complaint that its teaching would lead the skilled team away from a Phase II trial rings hollow. In any event, as we will see, that is not the teaching of the Harder poster.

The Harder poster

362. It is helpful at the outset to reiterate some observations already made.

363. First, the degree of scrutiny to which this, and the other prior art, have been subjected in these proceedings is not reflective of how the documents would have been read at the priority date. The posters were not intended to contain the wisdom of the ancients etched in stone and would not have been read as such. As described by Professor Meibohm (Transcript Day 8, p. 48):

A. … So, a poster clearly has, with regard to scientific value for me as a scientist looking at that, it has a completely different value compared to a peer reviewed publication that is published in a scientific journal. It obviously is intended to be shown, usually at a meeting to pique interest, to start a discussion, to show new results that have just come out and ultimately to create the starting point for discussion.

364. That is highlighted not to criticise any of the experts for their forensic analyses of the posters, but rather to remind myself that it is the perspective of the skilled team on the prior art which matters.

365. Second, although the prior art would have been read by the skilled team with a critical eye, it would not have been read with the excessively sceptical eye which Bayer’s witnesses brought to bear on the poster.

366. Professor Crowther, in particular, was sceptical about whether the data that the authors of the poster supported what they reported. Despite his belief that the authors would not misrepresent data, he doubts what is reported. By way of example, in relation to what was represented by the four graphs reproduced above at page 38 (Transcript Day 12, p. 30):

Q. And at the 24-hour mark, in three out of the four graphs there is a clearly identifiable difference between the 30 mg line and the placebo line; isn't that right?

A. So, I will agree that there's a visually apparent one, but again I think the point of my disagreement is that I have no confidence that those actually exist.

* emphasis added

367. Whatever limitations the skilled team might expect of a poster presented at a conference, it is simply not plausible to suggest that the authors, in particular Professor Breddin, would present a poster which did not reflect their results. This would not be acting “honestly or in good faith.” It is fair to be sceptical on the basis that the results don’t accord with prior expectations or that the authors have been over-enthusiastic in interpreting the results, but not on the basis that they would, in effect, have misrepresented them. Although not framed in those terms, much of Bayer’s witnesses’ doubts about the poster amounted to precisely that contention.

368. The skilled team would read the Harder poster and understand that the team who prepared it believed that it accurately reflected the relevant results from the study. Moreover, the skilled team would expect that it did accurately reflect those results. Professor Crowther’s invocation of scientific scepticism was, in truth, a call for scientific nihilism, where no statement is taken at anything approaching face value, irrespective of the source. There is nothing in the case law to suggest that that is the mindset to be attributed to the skilled team.

369. Where evidence is presented in the poster which was not at odds with the preconceptions which the skilled team would be expected to have, the skilled team would not have rejected that evidence simply on the basis that all the data underpinning that evidence was not contained in the poster. They would have had no expectation that all the underlying data would be presented, so its absence would not have undermined confidence in what was said; they would have had an expectation that what was stated in the poster was presented in good faith, in this case by, inter alia, a world-leading expert on thrombosis. Insofar as Bayer’s witnesses’ evidence was based on the contrary premise, that the conclusions stated in the Harder poster would have been rejected in the absence of all necessary supporting data, I disagree.

370. Moreover, it will be recalled that I do not accept the Bayer witnesses’ evidence that no reliance would have been placed on the results of the Harder assays simply because they had not been correlated with relevant clinical effects.

371. In considering the Kubitza posters, I agreed with Sandoz’s submission that it was appropriate to assume that the skilled team could be able to “get to” the prior art where all that was required to do so involved the application of routine measures. Different issues arise in relation to the experimental assays the subject of the Harder poster. These were not routine. Would the skilled team’s expectation of success be undermined by their awareness that the data necessary from the Harder poster was obtained through these experimental tests? In short, no, it wouldn’t.

372. The first thing to observe is that insofar as the tests are described as “experimental”, this refers to the fact that they were used in research rather than in the clinical setting, i.e. they were used for experiments. Moreover, it was accepted that the existence of, at least, the ETP test would have been known to the skilled team, albeit that it might not have used the test itself. Indeed, it appears that the test was developed by Professor Breddin.

373. Bayer’s counsel’s cross-examination of Professor Hirsh on the issue of variability of assays commenced with the following exchange (Transcript Day 3, pp. 139 – 140):

Q. Just to ask you about the variability of assays. Do you accept that assays will vary according to the person performing it, the reagent, the environment and so on?

A. Again you have asked three questions in one. If the assay is well established in the laboratory the technologist performing it would probably perform it the same way. The second part of that question was reagents. Definitely if the reagents are different you may get a different result.

Q. And if the environmental conditions are different?

A. Well, the assays are done under a certain temperature, 37 degrees centigrade;

if someone did it under 50 degrees the result would be different, but I don't see why they would do that.

Q. Or if the moisture content -- I am sorry, Professor.

A. I don't understand why they would do that.

374. It’s not clear to me why they would do that either, or, put more prosaically, why I should assume that the skilled team seeking to follow the teaching of Harder would not, if required, carry out the tests described by Harder in the most professional manner possible. Professor Hirsh then confirmed that his concern was with the validity of the results, not their

repeatability. But when pressed, having accepted that different laboratories may produce different results, he concluded (Transcript Day 3, pp.144 – 145):

A. They might get a slight difference quantitatively, but they wouldn't get a difference qualitatively. It would make no difference to the conclusion because I think there may be slight differences quantitatively.

375. Professor Hirsh was then referred to a paper by Kovacs and Ors, published in 1998, entitled ‘Three different chromogenic methods do not give equivalent anti-Xa levels for patients on therapeutic low molecular weight heparin (dalteparin) or unfractionated heparin’, which highlighted the variability of testing methods for anti-Xa for patients on therapeutic LMWHs or unfractionated heparins which, in some cases, were of clinical significance. The argument appears to have been that if there was variability even when using the same instruments, there could be no confidence in reproducing the tests described in Harder. It is an argument which proves too much, suggesting that no skilled team could have any faith that any result could be reproduced. As Professor Hirsh concluded in response thereto, you might not get the same results but they would all be in the same direction.

376. In his written report, Professor Crowther stated the following about what the skilled clinician would have understood about the ETP assay (at paragraph 184(a)):

“[There] was no standardized way to perform the assay (i.e. there was no single methodology which could be looked up), nor was there a standardized way to process and interpret its results. Consequently, the assay results were highly dependent on the particular methodology used in a given laboratory and the materials used to perform it (e.g. pre-analytic variables such as handling conditions of the sample, or preparation of platelet-rich plasma), as well as the environment in which the assay was performed (e.g. temperature).”

377. In cross-examination, he reiterated the sensitivity of the Harder assays to variables (Transcript Day 12, pp. 19 – 20):

A. These assays are extraordinarily dependent on methodology.

Q. Can I just ask you in relation to that, the point that you make is that even minor enough changes in the reagent or in the machine that's used or you go to a different laboratory you'll get different results, that's the central point you're making?

A. Yes, you have to have a very deep understanding of all of the characteristics, both in the machine and the reagent and the patients at the time they were performed if your desire was to replicate them.

Q. Okay. Would you agree with me that that degree of variability decreases if the assays are performed in the one laboratory using the same machine and using the same reagent?

A. Well, they do within the study. But if you had to replicate it outside the study, you would have to exactly replicate those conditions. So, absolutely, within the context of a study, it would be -- you always want to use the same laboratory within one study.

Q. That would be just standard practice?

A. If you're doing a study where it's possible to do it, yes.

Q. And also, would you agree with me that if you do use the same laboratory, you use the same reagent, same machine etc., you will be able to demonstrate from the use of this type of assay comparative effects?

A. Between doses?

Q. Between doses, yes.

A. Yes.

Q. And between doses and the placebo?

A. Yes, hopefully.

378. Insofar as the absence of a standard protocol for the assays described in Harder was raised as an obstacle by Bayer, Professor Wilkins proposed a straightforward solution – he would ask Sebastian Harder for his protocol “or ask him to do the study for me”. In the UK proceedings, the High Court concluded that Bayer would have provided the chemical formula for BAY 59 if asked, notwithstanding its witness’s evidence to the contrary. Dr Harder was not a Bayer employee. It is hard to imagine what reason, having presented a poster on the tests to be presented at the ASH Conference, Dr Harder would have had for refusing a request for details of his testing protocol. None was suggested in Professor Crowther’s evidence.

379. Overall, the evidence suggests that the skilled team would have confidence that the assays had been carried out competently and consistently for the purpose of the Harder study and that, insofar as they showed comparative effects, this wasn’t simply an artefact of the way the test was conducted.

380. Although the skilled team would have been aware that it couldn’t expect to replicate the results without precise details of the protocol used in the Harder study, I accept Professor Hirsh’s evidence that the skilled team would expect that the same assays conducted in laboratories qualified to do so would have produced qualitatively similar results.

381. Put in terms of the test for obviousness, insofar as it is suggested that the skilled team would not have had a reasonable expectation of success unless they could produce data similar to that reported in Harder, there is no reason to think that they would not have been confident that they could do so, either by having the tests carried out in a reputable laboratory, or if necessary, seeking assistance from the laboratory which carried out the original tests.

382. Turning finally to the contents of the Harder poster, the first observation to make is that Professor Hirsh and Wilkins’ enthusiastic response to the teaching of the Harder poster was far more convincing than Professor Crowther and Meibohm’s criticism of it. In this regard, as indicated above, I consider that their level of scepticism went far beyond that of the hypothetical skilled team, informed as it may have been by an overly narrow view of what the prior art could be said to teach, and also the unjustified weight they placed on the fact that the assays described in Harder had not been correlated with a clinical effect.

383. Professor Wilkins’ response to the study was more generous than that of Bayer’s witnesses:

A. But I would say, Judge, that, you know, I am interested in why this collaboration between academia and pharma decided to conduct an exploratory investigation, exploratory trial of their drug with a novel set of assays for thrombin generation. As we've heard, they've got data on the aPTT and the prothrombin time, the HepTest, so they have that information available to them. But, nonetheless, they have decided that it is worth spending time and resource on a study where they are employing these three assays. And the reason for doing it, I think, is very smart; because they are interested in thrombin generation. They realise the other assays are inadequate in that respect and so they have gone for three assays which report thrombin generation, which, despite all the discussion yesterday, is the clear final common pathway to thrombosis.

384. This acknowledgement of the inventiveness of the study provides a far more realistic reflection of the likely response of the skilled team to the Harder poster than the pettifogging attitude of Bayer’s witnesses, finding only fault or lacunae with a study the reported results of which they acknowledge to be intriguing.

385. The main point of contention between the parties regarding this poster, and in a very real sense the dispute at the heart of these proceedings, relates to the interpretation of the six graphs shown at figures 3 and 4 of the poster, the first four of which are reproduced at page

38 above and whether these provide evidential support for the fourth conclusion in the poster, the one which refers to the inventive concept.

386. As with figure 1 of the Kubitza SAD poster, on a plain reading of the graphs, it is clear that they do support the repeated references in the text to inhibition of thrombin generation being prolonged “over 12 hours”. Five of the six graphs, four of which represent ETP assays, the other two the PITT and PICT assays, show inhibition of thrombin generation prolonged at 12 hours and at 24 hours for the 30 mg dose, i.e. three different assays point in the same direction. Although the poster does not call out the 24 hour result, it will be recalled that Blood 3003 did. If that is a “real” result, it does suggest that BAY 59 would be suitable for once-daily dosing, and the skilled team would consider it worth trying with a reasonable expectation of success. It is also clear from the poster that this is the conclusion the authors reached: it is clearly stated in the conclusions section of the poster.

387. Sandoz’s witnesses consider that the skilled team would have read the results, been enthused by them and had sufficient confidence in them to give them a reasonable expectation of success for once-daily dosing. Bayer’s witnesses’ response was a case of “interesting, but…”

388. Many of the reasons for that “but”, or for Sandoz’s evidence that the skilled team’s scepticism would not have over-ridden their enthusiasm have already been addressed. The positive nature of the evidence from the three posters (and abstracts) together, the confidence in the authors, the awareness that the Harder poster had been prepared by many of the same team, together with a world-leading expert, in light of the Kubitza studies, the reliance on uncorrelated assays, the concessions made on both sides, the inconsistent half-lives in Kubitza and Harder, the baseline for the placebo in the graphs. The main remaining point of difference between the parties relates to the statistical significance of the data presented in the Harder poster.

389. There are a number of factors which must be considered here. First, the sample size for the Harder study is small, smaller than the Kubitza SAD and MAD studies. Against that, all the experts accepted that Phase I trials are not intended to be sufficiently powered to give statistical significance (see, for instance, Professor Crowther, Transcript Day 11, p. 139). That factor is not, therefore, decisive. Of more relevance is the fact that the graphs show error bars for each dose and that, in all cases, these overlap after, at most, four hours. The error bars reflect the fact that the values shown reflect means plus or minus one standard deviation. Professor Crowther’s evidence was that the use of a single standard deviation was a less precise measure than would be usual (two or three standard deviations) and the data might have been presented in this way because the authors were worried about a statistical outlier.

390. The experts were agreed that overlapping error bars, especially where only one standard deviation was used suggested that the results may not be statistically significant. But in the context of a Phase I study, insufficiently powered to provide statistical significance, this is not a reason for dismissing the results. Moreover, the poster highlights that the results had been subject to statistical analysis using “analysis of variance”, known as ANOVA.

391. In respect of each assay, the authors then highlight that the assay was reduced (in the case of ETP) or prolonged (in the case of PITT and PICT) “significantly”, before highlighting that the effect was sustained over 12 hours. I accept Professor Crowther and Meibohm’s evidence that, insofar as this suggests that the ANOVA analysis shows statistical significance, there is no way of knowing whether the values at 12 and 24 hours have

significance and that the analysis was more likely to be impacted by the 2 and 4 hour values where the impact was most pronounced.

392. Professor Crowther very fairly accepted that the skilled team would not be expected to have a detailed statistical knowledge. When Professor Crowther was asked what ability the skilled clinician would have to make the kind of statistical assessment conducted by him and Professor Hirsh, he explained (Transcript Day 12, pp. 31 – 32):

A. That is an interesting question. So, yes, I think we both have spent our lives immersed in this. I think a skilled clinician would understand the concept of a P-value. So they would understand that if a P-value is less than 0.05, that means that there's a very good chance that something didn't arise as a result of chance.

I think if you venture much farther off the pathway than that, they would have reduced knowledge. So, for example, if you said to a skilled clinician can you tell me what an ANOVA is, I would be surprised if they could interpret it in a significant -- they would be able to understand what that meant or provide any kind of commentary on it.

393. And then (Transcript Day 12, p. 33):

A. So, I think a skilled clinician would have some reasonable understanding of standard deviations and understand that we're all taught that if the bars overlap, that should raise a question in your mind. So, I would have a great deal more confidence that the skilled clinician, as I understand that, would look at these graphs and say, oh, look, between two, four, six, maybe even eight hours there's probably something real here. After that we see the confidence or the 95, the standard deviations overlap, and that would raise some concerns in their evaluation.

Q. But it would not lead the skilled clinician to discount the data completely, isn't that so?

A. No, they may find it informative.

394. Given the small sample size and in light of the way the data was presented, using a single standard deviation, Bayer’s witnesses expressed concern that the mean values shown in the graphs might be attributable to an outlier. But that is inconsistent with the suggestion that the authors would be expected to report their results in good faith. From their results, the authors drew a conclusion that BAY 59 might be suitable for once-daily dosing. The clear implication is that the mean values were representative, and that they had sufficient confidence that they were not skewed by an outlier from a group of healthy volunteers who, it was agreed, would have been screened for any condition which might be expected to affect the results, to enable them to make that statement. Professor Crowther’s speculation as to the possibility of an outlier requires a degree of interrogation of the poster by the skilled team and a degree of scepticism which the case law does not suggest should be attributed to it. Insofar as Bayer’s submissions suggest that the skilled team would have had regard to the fact that there was no P-values given in the Harder poster, this is not supported by the evidence which was to the effect that if there had been P-values, the skilled team would have understood their significance.

395. The key point about this evidence is not the extent to which the skilled team would have accurately assessed the statistical significance of evidence, but rather how they would view the evidence in the round. And in that respect, as Sandoz’s witnesses highlighted, the evidence all pointed in one direction. All of the six graphs show the most pronounced effect at 2 to 4 hours then a steadily waning effect thereafter. Five of the six show a diminished but continued effect at 24 hours. At paragraph 302 of his report, Professor Crowther seems to suggest that the focus by Professor Hirsh on the ETP-peak result involves “cherrypicking”. This is difficult to understand. Although ETP-peak indicates the most pronounced effect at 24 hours, only one of the six graphs suggests no difference between the placebo and the 30 mg dose at that stage. Three different assays indicate the same effect. And in fact, this is consistent with the results shown in the graph in Figure 1 of the Kubitza SAD poster. The reader is then told by the authors that they consider, in light of the evidence, that BAY

59 may be suitable for once-daily dosing. Professor Crowther thinks that the skilled team would disagree with them. In that, I can only conclude, he is not correct. Having read Harder, in light of Kubitza, the skilled team would also think that BAY 59 might be suitable for once-daily dosing.

396. In so concluding, I have borne in mind why, if proceeding to a Phase II trial, the skilled team might be reluctant to try once-daily dosing, and therefore could not try it unless they had an expectation of success, i.e. it being safe and effective, and the particular necessity for caution in trials of novel anticoagulants. A Phase II trial in patients undergoing hip surgery means a trial in patients who are at risk of bleeding, but also of VTE including PE. Any dosing regimen, where the clinical effect of the doses given cannot be predicted by reference to a correlated assay, necessarily carries with it some risk. That risk will be managed insofar as possible in the hospital setting where the hip surgeries take place, but that cannot eliminate risk. As accepted by Sandoz’s witnesses, venograms carried out after surgery offer little mitigation of VTE risks.

397. Bayer places huge emphasis on the necessity for any new anticoagulant in a Phase II trial to be effective and, in effect, argues that the risk that it would not be effective means that it couldn’t be considered obvious to try. This somewhat ignores their own evidence that a significant proportion of drugs which are brought to Phase II trials don’t succeed. In all cases, in order to be permitted to advance to Phase II trials, the evidence made clear that there must be an expectation of success – a drug company would not be permitted to dose a patient with a drug which wasn’t expected to be safe and effective. Nonetheless, drugs fail clinical trials and adverse outcomes occur. Professors Hirsh and Wilkins were realistic about this but did not think that the skilled team would abandon obvious avenues to try simply because risk couldn’t be eliminated. I agree. Were it otherwise, there would be little or no new drug development.

398. A 30 mg dose given once-daily carries different risks to, for instance, a 30 mg dose given twice daily, or a 30 mg daily dose given in two 15mg doses. Absent information which can never be obtained – from correlated assays – none of these doses can be given without any concern about VTE or excessive bleeding. Some comfort can be taken from dosing according to half-life, but as the experience with razaxaban would have informed the skilled team, that is not a perfect safeguard, and in the case of novel anticoagulants only highlighted the risk of overdosing.

399. Sandoz’s witnesses were comfortable, based on the information that they had, that trying once-daily dosing would have been considered an acceptable risk by the skilled team. Bayer’s witnesses thought otherwise. For the reasons stated above – the Bayer witnesses’ excessive scepticism, the undue weight placed by them on the lack of correlation of assays, the implausibility of their suggestion that no dosing regimen could be proposed by the skilled team based on the prior art, the dramatic and inadequately explained alteration in Professor Meibohm’s evidence on this issue – the Sandoz witnesses paint the more realistic picture of the mindset of the skilled team. Accordingly, I prefer their evidence to that of the Bayer witnesses when considering whether, when giving evidence that the skilled team would think once-daily dosing was obvious to try, they are attributing too great an appetite for risk to the skilled team.

Preliminary conclusion

400. Without considering the foreign judgments and the obligations of comity, or the secondary evidence, based solely on the of the expert witnesses, I conclude that the skilled team would have been motivated to investigate once-daily dosing, and would not have been dissuaded from so doing by the costs and uncertainty of clinical trials. The possibility of once-daily dosing would certainly have occurred to them in light of the prior art. Having regard to its common general knowledge, the skilled team, having reviewed the Harder poster together with the Kubitza posters, would have concluded that once-daily dosing was obvious to try and would have considered that it had a reasonable prospect of success, i.e.

of being both effective and safe. Based on those considerations, and subject to consideration of judicial comity and the secondary evidence, I conclude that the Patent, therefore, did not involve an inventive step within the meaning of section 9 of the Act and, accordingly, was not patentable.

Judicial comity

401. The obligations of judicial comity do not demand an alternative conclusion, rather they tend to confirm this conclusion. There are many cases in which the validity of the Patent has been upheld. I do not suggest that those were wrongly decided based on the evidence before those decision-makers. There are some explanations for why different outcomes may have been reached in other cases, a number of which have been explained above. All other cases proceeded on the basis that BAY 59 would have been progressed to Phase II trials. In some, such as the TBA decision, the prior art was different: the key Harder poster was not before the TBA. In many jurisdictions, there was no cross-examination of witnesses, the value of which I have already noted in this case. In some jurisdictions, it appears that Bayer may have offered an alternative explanation of how they came to the invention, although I don’t place great significance on this. In most of the jurisdictions where the patent has been upheld, the deciding body concluded that no weight would be placed on the Harder study, inter alia, because the assays used had not been correlated with a clinical effect, a conclusion which is not supported by the evidence in these proceedings. In the TBA decision, this seems to have been the sole basis for concluding that Blood abstract 3003 (the Harder abstract), which included the statement that a sustained effect on thrombin generation for up to 24 hours, would not have provided the skilled team with an expectation of success. None of the decisions upholding the patent are final.

402. There are also decisions in which the patent has been declared invalid. The most relevant of those is the UK decision, which is final. Its particular relevance comes from the fact that the same witnesses have given evidence in these proceedings as in those, and the evidence was tested in that legal system in the same way that it has been tested here. Bayer has sought to distinguish the UK proceedings by references to differences in the evidence actually led. The UK proceedings may have focussed on different issues which Bayer relied on for the purpose of arguing that once-daily dosing was not obvious to try, but none of those differences are so material as to displace the relevance of the findings of the UK High Courts and Court of Appeal on the same question having heard the same witnesses, still less to persuade me that despite my conclusions on the evidence here, I should nonetheless reach a different conclusion in this case.

403. Having regard to the conclusions in the evidence in this case, the principle of judicial comity would suggest a similar conclusion here to that reached by the UK courts. The principle is thus respected by the proposed conclusion here.

Secondary evidence

404. The parties are agreed as to the status which can be afforded to the secondary evidence:

it cannot provide primary evidence where the evidence isn’t there, but it can be helpful in testing the plausibility of the expert evidence regarding what the skilled team would have

done when presented with the prior art. Both parties agree that Molnlycke, quoted above, represents the correct approach to such evidence.

405. The secondary evidence in this case consists of documents provided by Bayer on foot of a request for discovery. The documents were provided subject to a confidentiality agreement made on 25 September 2024. Although the claim of confidentiality has been waived over many of the documents, there remained a dispute about some documents and a third party was appointed by the parties to reach a determination on confidentiality. That third party determined that some of the documents remained confidential save to the extent that the contents thereof were already in the public domain.

406. Following Sandoz’s opening submission, Bayer sought an order that information disclosed to Sandoz in these proceedings, designated as confidential information pursuant to the confidentiality agreement, be dealt with in such a way that that information is not released in whole or in part into the public domain or is otherwise caused to lose its character of confidentiality. The application sought, if necessary, that the proceedings, or parts of the proceedings, be heard in camera.

407. Sandoz proposed a more limited form of order, pursuant to which the confidentiality of the documents would be preserved but only pending further order, when the claim for confidentiality could be assessed in light of the evidence and after examination of the documents. In light of the decisions in Gilchrist v Sunday Newspapers [2017] 2 IR 284 and BOC Aviation (Ireland) Ltd v Lloyds [2024] IEHC 162, Sandoz accepted that the court had jurisdiction to make orders of the type sought and that the protection of confidential information could, in principle, provide justification for the making of an order placing some restrictions on the extent to which the entirety of proceedings be heard in public. However, it contended that the basis for making such an order had not been made out, and couldn’t be until the documents had been fully considered by the court.

408. As indicated in an ex tempore ruling on the second day of hearing, I considered it at the very least premature to make an order of the type sought by Bayer but made an interim order preserving the confidentiality in the documents pending further order, and granting liberty to Bayer to apply to make the order permanent in respect of some or all of the documents.

409. The parties were then in a position to address the contents of the documents, which both sides did in submissions and throughout the evidence of each of the four witnesses during the hearing of the action.

410. In preparing this judgment, I observed that, despite the evidence having been admitted and dealt with by both sides without objection, I hadn’t been informed of any agreement regarding the evidential status to be afforded to it, rather the submissions had been directed to its ‘secondary’ nature. In particular, it seemed to be that I was being asked to assess the credibility of the experts’ evidence by reference to documents, and that, accordingly, the parties must be affording some evidential weight to the documents. In light of the decision in RAS Medical v Royal College of Surgeons in Ireland [2019] 1 IR 63, which stresses the necessity for clarity on the status of discovery documentation deployed in proceedings, I sought confirmation from the parties of the basis on which the discovery/confidential documentation was put before the court and whether there was any agreement between the parties in this regard.

411. This led to an exchange of correspondence between the parties, and a short hearing on

18 June 2025 to enable me to understand the parties’ respective positions. In light of that hearing, I directed short written submissions, which were received, as directed, from Bayer on 20 June and from Sandoz on 23 June 2025.

412. In Bayer’s submissions, they characterise Sandoz’s initial position on the role of the evidence:

“Sandoz’s experts understood Sandoz as asking them to give their interpretation of what had happened during the development of the invention by reference to the discovery documents that Sandoz selected, and to give their opinion on whether or not such ‘facts’ supported their first-round opinion evidence on the objective question of what the person of ordinary skill in the art would think of the prior art.

Sandoz’s experts duly gave their opinion on those matters. Bayer understood them to be giving such an opinion, as did the Court.

This was the only basis for an assertion of relevance for the discovery, namely that the discovery could potentially be used by Sandoz to establish facts in relation to the development process that might support Sandoz’s expert evidence on the objective statutory test.

Under Molnlycke, (adopted in this jurisdiction by the Court of Appeal in Boehringer these kinds of facts can have secondary relevance.”

413. Bayer had raised no objection to the documents being introduced and put to witnesses on that basis. Their belated objection, following the court’s query, was on the basis of a perceived change of position by Sandoz in light of its letter written in response to that query which suggested that “…the discovery/confidential documentation cannot be deployed to establish facts” and “Sandoz cannot use the materials to assert as a fact that Harder was the predominant factor to moving to a once-a-day-dosing.” If that is so, says Bayer, the documentation becomes irrelevant and inadmissible.

414. Sandoz clarifies its position in its submissions, having first pointed out that it is too late for Bayer to raise any complaint, the documents having been put to witnesses on both sides in the proceedings without objection:

“In evaluating the evidence of the experts and deciding which expert evidence to prefer, the Court can take into account the extent to which it is consistent with or supported by the secondary evidence. So, for example, with regard to the dispute between the experts as to whether the data from the Thrombin Generation Study (reported in the Harder Poster) would have led the notional skilled team to trial OD dosing, the Court can have regard to the evidence in the documents which shows that this data was the primary driver of the decision by the Bayer development team to add an OD dosing arm to the Phase II studies in deciding whether to prefer the evidence of Prof. Hirsh or Prof.

Crowther.

It is accepted that, in order for the Court to determine the extent to which the documents support the evidence of the experts, the Court has to form a view as to what the documents show the Bayer development team actually did and the basis on which they proceeded. To the extent that there is a disagreement between the experts in this regard, the Court can make findings as to the facts disclosed by the documents. Furthermore,

insofar as there are differing views between the experts as to how the documents should be interpreted, the Court can decide which view it prefers.”

415. Insofar as is necessary, Sandoz relies on the provisions of the Civil Law (Miscellaneous Provisions) Act 2020. Bayer contends that it has disputed the admissibility of the documents and that they should, therefore, not be admitted pursuant to section 16 of the Act. Since neither party adverted to the 2020 Act at all during the hearing, this is a little difficult to understand, though it appears that Bayer accepts that, subject to objection, the documents are admissible in accordance with the terms of the Act.

416. It seems to me that what Sandoz has said in its written submissions reflects the position as Bayer understood it to be, according to its submissions, and to which it had not objected. It is not entirely clear whether it now objects to the documents having been admitted on that basis, but it appears not, and it would clearly be too late to raise such an objection if it did, whether by reference to the 2020 Act or otherwise.

417. I propose, therefore, to deal with the documents on the basis that they were being relied on for the purpose that Bayer says it understood them to have been deployed. I note, however, that the documents, in the absence of oral evidence from those involved in the development process, could only ever give a very limited picture of what was actually done by Bayer during development and the significance of the secondary evidence is therefore quite limited.

418. The relevance of the document is in testing the plausibility of the witnesses’ evidence that the thrombin generation study could, or could not have, provided a basis for introducing a once-daily dosing arm into the Phase II trials of BAY 59.

419. I will refer to a number of the documents discussed in evidence in order to examine the role that the thrombin generation study appears to have played.

420. The documents referred to at the hearing were a small selection of the discovery provided by Bayer and span a date range from 10 January 2002 to 19 June 2004.

421. The original clinical study protocol for the Hip Study in which BAY 59 was to be used in patients for the first time was dated 9 October 2002. It was a dose escalation study and six consecutive doses were proposed: 5 mg bid; 10 mg bid; 20 mg bid; 20 mg tid; 30 mg bid; 40 mg bid.

422. On 23 May 2003, an amendment to the clinical protocol was proposed by which a 2.5 mg bid dose was added to the trial. The reason appears from the “Description of the Amendment”:

“Data from the first and second dose step clearly demonstrated that the efficacy of the

5 mg bid and 10 mg bid doses were much better than anticipated, with only 7 and 3 DVT’s receptively [sic] detected by the local radiologist.

Since periand postoperative bleeding events are of great concern for surgeons, often limiting the general use of anticoagulation in patients at a higher bleeding risk, search for low-dose anticoagulants with a low risk of bleeding and still effective thromboprophylaxis is mandated.”

423. On 11 July 2003, there was an expert meeting in Stratford-upon-Avon in the UK. A minute of that meeting was prepared, which is a document disclosed by Bayer. It is not identified who prepared the minute.

424. It was explained in evidence that the holding of such expert meetings, where leading experts are invited to discuss the developmental progress of novel drugs, are a standard part of drug development. In this case, the experts met with Bayer representatives involved in the development of BAY 59 and discussed the results of ongoing trials and offered their input. A number of aspects of the minute were highlighted by each side and put to the witnesses for comment.

425. The meeting objective is described:

“The objective of this experts meeting was to validate the clinical trial program crosschecking with recommendations previously held experts meetings on BAY 59-7939 within the last 12 months. In addition, the meeting focused on the development objectives for prevention of VTE in patients with cancer.”

426. The minute contains an update on clinical phase I data including the study the subject of the Kubitza SAD and MAD studies. It also includes a description of the thrombin generation study:

“Single doses five and 30 mg BAY 59 -7939 do not only prolong lifetime thrombin generation into independent assays (ETP, PITT) they also decrease the peak (Cmax)

and total effect (AUC) of thrombin generation two hours post dose. Although one parameter had returned to baseline values after 24 hours.”

427. Professor Hirsh suggests by reference to the results reported in the Harder poster, that they should read “had not returned to baseline values after 24 hours”, though this was not Professor Wilkins’ interpretation.

428. The expert opinion/comments are reported as follows:

“Experts were unanimously positively impressed by the data, which support the plans the existing clinical trial programme. It was also commented that so far there is no compelling evidence for the need of twice-daily dosing. Even once-daily dosing may prove to be clinically sufficient in the future.”

429. An update on the Hip Study reports that “currently available date indicate that … BAY

59 – 7939 seems to be effective already at low doses.”

430. Under a description of what had changed in the competitive environment, it is noted that five or six other companies were working on other Factor Xa inhibitors and that “[o]nce-daily dosing may not be important now, but may play an important role in future.”

431. The most important success factors for new antithrombotic agents are described as: safety, convenience, efficacy, number of indications, cost effectiveness. Convenience was followed by “(oral administration, no monitoring, frequency of administration, eg. oncedaily)”.

432. Almost exactly a month after this meeting, on 12 August 2003, Bayer prepared a further amendment to the clinical protocol, to include two new dosing arms, both once daily. The description of the amendment is as follows:

“Daily doses of 30 and 40 mg are covered by the trial protocol, i.e. they are included in the dose range, which is proposed to be studied. Initially, bid and tid dosing regimens were planned. Since been dosing resulted in very satisfactory clinical results both with regard to efficacy and safety, the Steering Committee strongly encouraged to explore also once-daily dosing regimen within the pre-specified range of daily doses. This proposal is also backed by the observation in the phase I study (IMP 11140) that a single dose 30 mg resulted in a sustained inhibition thrombin-generation lasting for more than 24 hours. This observation points to the fact that pharmacodynamic effects of BAY 59-7939 may last longer that it could be expected on the basis of its pharmacokinetics.”

433. It was agreed that the phase I study referenced here is the thrombin generation study. It was also explained that the Steering Committee was distinct from the expert meeting and were the persons ultimately responsible for overseeing the clinical development. It was further explained by Professor Crowther that an amendment of a trial in this way was no small matter. The Hip Study would have taken place in multiple jurisdictions and multiple locations within each jurisdiction. In order to amend the protocol, it would be necessary to secure the agreement of the ethics committee in every single location.

434. Both sets of experts provided observations on this document. Professor Crowther’s view was that it was the early results of the Hip Study which were the “real reason” for the amendment and that the thrombin generation study was no more than a contributing factor, offering evidence in support of the amendment. Sandoz’s witnesses interpreted the document as showing that the amendment had been prompted by the thrombin generation study.

435. On a plain reading of the document, both factors seemed to have influenced the

decision. However, a consideration of the document in context suggests that Sandoz experts were correct in emphasising the role of the thrombin generation study. First there is the fact that the protocol had already been amended in light of the early results of the Hip Study. Moreover, there is nothing in the minutes of the meeting in June which suggests a significant development on that front since the amendment in February. Second is the fact that the amendment was made shortly after the thrombin generation study results were first reported in an expert meeting. Third, there are documents which come after which clearly suggest that it was the thrombin generation study which prompted the change to once-daily dosing and nowhere reference the early results from the Hip Study as causative. Fourthly, and most importantly, Bayer’s witnesses’ own evidence was that establishing safety and efficacy in low doses doesn’t tell you anything about the frequency with which you should dose the drug (see Professor Crowther, Transcript Day 12, pp. 116 and 117).

436. In the minutes of a further expert meeting on 8 November 2003 in Orlando, Florida, expert comments were reported on the subject of dose selection for Phase III trials, Bayer being reported as wanting to use only 1 dose if possible. It was noted in the comments that the “issue of once-daily vs. twice-daily dose [was] not solved.” Bayer emphasise the fact that the documents show that there remained a dispute about whether to pursue oncerather than twice-daily dosing well into 2004.

437. A teleconference took place on 15 December 2003 for which minutes were prepared. Dr Frank Misselwitz prepared a slideshow presentation for this meeting. Dr Misselwitz was centrally involved in the development of BAY 59. He was one of the authors of each of the three posters comprising the prior art. He has, as we have seen, given evidence in many jurisdictions and he has written subsequent journal articles about the development of BAY 59.

438. The slides for the December 2003 meeting contain a slide setting out recommendations from the expert meeting and steering committee on the development of once-daily dosing. It contains a number of bullet points including the following:

“In general, Experts got very excited on the 30 mg od results and encouraged us to complete adjudication process and CDP asap.”

439. In his first supplemental report, Professor Hirsh interpreted this as a reference to the thrombin generation study. In his response to Professor Crowther’s replying report, he accepted that this was not so, rather it referred to the amended Hip Study. However, when giving oral evidence, he once again asserted that this referenced the thrombin study. When cross-examining Professor Crowther, counsel for Sandoz accepted that Professor Hirsh had erred in his oral evidence and that this was a reference to the Hip study.

440. In their closing submissions, Bayer place some emphasis on this “confusion” by Professor Hirsh, suggesting that it meant that he was in error in suggesting that it was the thrombin generation study rather than the satisfactory results of the Hip Study which triggered the introduction of once-daily dosing, and that this was evidence of his reliance on hindsight.

441. The confusion on the interpretation of this document has nothing to do with the reason that once-daily dosing was introduced, so I don’t follow the logic of Bayer’s criticism. It was suggested by Sandoz in their closing submissions that Professor Hirsh’s error could be attributed to tiredness. This is precisely what I considered to be the position when this issue arose in evidence; the witness seemed tired at the end of a long day and long week of detailed and skilful cross-examination, notwithstanding that additional short breaks had been taken throughout his evidence to accommodate him, and made a mistake. Given that the evidence related merely to his interpretation of a document which is of marginal relevance, I attach very little significance to Professor Hirsh’s apparent misunderstanding.

442. The minutes of the meeting itself refer, in the first bullet point to the “commercial value is highest in Afib and has potentially most to gain from once-daily formulation. Even if we come into the market second, but first with once-daily formulation we could still take the market.”

443. Another slide deck was prepared by Bayer dated January 2004, entitled “Options/Scenarios for Once-daily Development”. It contained a slide headed ‘OD History’.

444. The slide identified 6 milestones, starting with the initial Phase II plan to compare bid and tid dosing, followed by a multiple dose study suggesting a good chance for bid dosing and then the commencement of Phase II trials with bid administration.

445. The next bullet points are as follows:

Q2/2003

- Results of thrombin generation study give first hints for prolonged action of Bay 59- 7939 at 30 mg/volunteer

Q3/2003

- Additional dose arm with 30 mg Bay 59-7939 was added to the phase IIa study

446. The following slide contains the following graph.

Slide No. 4 of Slide Deck “Options/Scenarios for Once Daily Development”

447. As can be seen, this is in substance the same graph as that contained at Figure 3(a) of the Harder poster reproduced above at page 38, albeit presented slightly differently, for instance with the error bars separated, and with the 24 hour data point highlighted.

448. The subsequent slides describe the results from the Hip Study.

449. These are, of course, just slides, the purpose of which is unclear. Nor is it clear who prepared them or what their knowledge of the overall development process was. However, taken at face value, they could only be understood as indicating that it was the results of the thrombin generation study that prompted Bayer to introduce once-daily dosing.

450. This is stated in terms in a document entitled ‘Bayer Pharmaceuticals Strategic Review’ prepared by an outside agency, clintarget llc, dated 10 June 2004. Under the heading ‘Oncedaily dosing strategy’, the document records:

“Feedback from key experts as well as an assessment of the competitive landscape has led to the strong desire to develop BAY 59-7939 as a once-daily medication. Multiple dose Phase 1 data initially indicated that bid dosing was appropriate and thus the ODIX-HIP Phase IIa study was conducted with bid administration. In the meantime, results from the thrombin generation study (Study 11140) indicated anticoagulant activity for BAY 59-7939 over 24 hours given at 30 mg (not effective at 5 mg), which led to incorporation of a 30 mg od treatment arm into the ODIX-HIP study.”

451. Similarly, minutes of an expert meeting held in Venice, Italy on 19 June 2004 appear to refer to once-daily dosing being supported by the pharmacodynamic effects shown in the thrombin generation study. Dr Misselwitz’s comments recorded in the minutes of a Global Projects Team Meeting dated 23 January 2004 were to similar effect.

452. Sandoz also referenced a number of publications post-dating the priority date in which those involved in the development of BAY 59 described the process. For instance, in ‘The discovery and development of rivaroxaban, an oral, direct, factor Xa inhibitor’ by, inter alia, Perzborn, Kubitza and Misselwitz, published in Nature Reviews: Drug Discovery in 2011, the authors observe:

“Rivaroxaban inhibited factor Xa activity in a dose-dependent manner after single dosing in healthy subjects. Rivaroxaban also inhibited thrombin generation in healthy subjects, and some parameters of thrombin generation remained inhibited for 24 hours after administration of a 30 mg dose. These results offered the first indication that oncedaily dosing might be feasible.”

453. And in ‘Evidence-Based Development and Rationale for Once-Daily Rivaroxaban Dosing Regimens Across Multiple Indications’ by, inter alia, Kubitza and Misselwitz, published in Clinical and Applied Thrombosis/Hemostasis in 2016, the authors write:

“During the course of [the Hip] study, 2 observations suggesting that OD dosing rivaroxaban might be feasible were noted. First, a phase I study of the pharmacokinetic/pharmacodynamic interactions of rivaroxaban and enoxaparin showed that both drugs had a similar rapid onset of anti-factor Xa activity (time to Cmax of 2 – 4 hours), with a nearly identical time course. Because enoxaparin administered OD had been the SOC comparator in the [Hip] trial, this close similarity in the pharmacodynamic time course of rivaroxaban and enoxaparin suggested that OD rivaroxaban might be a viable regimen, especially because anti-factor Xa activity if the leading anti-coagulant effect for low molecular weight heparins (LMWHs) such as enoxaparin. Second, in another phase I study, rivaroxaban-induced inhibition of factor Xa activity and reduction in the ETP were still evident at least 24 hours after administration doses approximating therapeutic concentrations. As noted previously, reduction in ETP was evident 24 hours after a 30-mg dose of rivaroxaban but not after a 5-mg dose. On the basis of these observations suggesting that OD dosing might be feasible, a further rivaroxaban dose group, 30 mg OD, was added to this study.”

454. The documentation and articles are, as highlighted, of only secondary relevance in these proceedings. The experts were asked to comment on, in effect, to interpret the documents. Save to the extent that all the experts were able to confirm that the drug development process revealed by the documents was consistent with their expectations for how a new anticoagulant drug would have been developed at the priority date and clarification of some terminology, it does not seem to me that their views on what the documents mean carries any particular weight and I do not propose to deal in detail with their evidence in this regard.

455. There are, however, two takeaways from the documents. First, insofar as the thrust of Professors Hirsh and Wilkins’ evidence was that the thrombin generation study would have provided evidence which would be used by the skilled team when designing a Phase II study, the documents suggest that that is how the Bayer team actually dealt with the study, lending plausibility to their evidence. Insofar as Professors Crowther and Meibohm’s evidence was that the study would not have been relied on by a skilled team because the assays used were not correlated with a clinical effect, that view is not supported by the documents.

456. The second relates to the approach of the witnesses to the documents.

457. In Bayer’s submissions, it repeatedly criticises the evidence of Sandoz’s experts for their impermissible reliance on these documents, and other material from after the priority date, as evidence of them using hindsight to support their opinions. I do not believe that this is what they were doing. Both Professors Hirsh and Wilkins had given their evidence in the UK proceedings without any reference to, or awareness of, the contents of this documentation. In the UK proceedings, they had been cross-examined in detail by Bayer on the basis that they were, in effect, wrong in asserting that the skilled team would have opted for once-daily dosing on the basis of the evidence available.

458. Their initial reports and their replying reports in these proceedings were also delivered without any awareness of the discovery documents. There can be no question, therefore, that they reached their views on the prior art by reference to material not in the public domain at the priority date.

459. The plain interpretation of the secondary evidence is that the thrombin generation study, the subject of the Harder Poster, played a significant part in Bayer’s decision to include a once-daily dosing arm in its Phase II trials. It is scarcely surprising therefore, that Sandoz’s witnesses would draw comfort from the fact that “in the real world” Bayer seems to have afforded significant weight to information that the witnesses had regarded as important but in respect of which, in both these proceedings and in the UK, it was Bayer’s case that they were mistaken in so doing. Professor Wilkins, in cross-examination, observed that, by reference to the documents, he was, I think he meant, reassured that he “wasn’t fooling [himself], there was evidence that other people saw of value.”

460. Bayer’s criticism of Sandoz’s witnesses on this ground is not well made.

461. By contrast, it is a striking aspect of Bayer’s witnesses’ evidence that their evidence was almost entirely unaffected by the secondary evidence, which they made clear that they had never seen prior to the receipt of Sandoz’s expert reports addressing the discovery documentation. On even the most neutral interpretation of the documentation, the thrombin generation study, the significance of which Bayer’s witnesses had, at best, downplayed, and more realistically, discounted, marked an important step in the development of a once-daily dosing regimen.

462. By the time they gave evidence in these proceedings, they were aware of what the documentation appeared to indicate about Bayer’s development process. Not only were they unmoved, but they also both then offered interpretations of the documents which, respectfully, simply do not reflect the overall tenor of those documents, in a manner which suggested that both witnesses were, no doubt inadvertently, mounting a defence of positions

which they had already taken, and thus, indirectly, of Bayer, rather than, as required, reassessing those positions in light of new information.

463. In disputing the clear implication of the secondary material, Bayer and its witnesses conjured the spectre of the “vast” or “enormous” amounts of material which were available to Bayer but not to the notional skilled team and that it must have been this vast amount of material which enabled Bayer to move to once-daily dosing in circumstances where the skilled team, armed only with the prior art, interpreted in light of their common general knowledge, could not.

464. The existence of other material can readily be inferred: the actual results of the Kubitza and Harder studies, for instance, would have been available to Bayer but are not contained in the posters or abstracts. Some Phase II results seem to have been available at the time of the decision to introduce OD dosing arms. Other studies are referred to in the discovery documentation. But this must be the case in every challenge based on obviousness over the prior art. If the fact that the inventor had access to detailed data not available to the skilled team automatically operated as an obstacle to the argument that an invention would have been obvious to the skilled team, it is hard to understand how any such obviousness challenge could succeed.

465. Insofar as Bayer sought to imply that there was some particular information within their possession which enabled it to try once-daily dosing in circumstances where it would not have been justified for the skilled team to do so, Sandoz says this is impermissible. Just as Sandoz cannot rely on the secondary material to fill any gap in its primary evidence, nor can Bayer. Having called no factual witness for the purpose of identifying particular knowledge available to Bayer other than the prior art, they cannot rely on their experts’ apparent assumption that such material must exist.

466. In this regard, as discussed in the section on foreign judgments, in many jurisdictions, Bayer has called factual witnesses, or has provided statements from them. In particular, it appears from a number of the judgments that in some jurisdictions, Dr Misselwitz has given evidence of what Judge Hacon calls the “invention story” in the UK proceedings. This, in simple terms, was an explanation of what enabled Bayer to introduce once-daily dosing without reference to the prior art. In other words, it was an alternative explanation for how

Bayer came to take the inventive step than that offered by Sandoz, that it was obvious in light of the prior art.

467. Having chosen not to give factual evidence, whether of a specific invention story or otherwise, I agree with Sandoz’s submission that it is impermissible to ask the court to infer the existence of some unspecified alternative data or explanation for the introduction of once-daily dosing, in effect, to introduce an unexplained invention story by the backdoor, without having led evidence and exposed any witness giving that evidence to crossexamination.

468. For obvious reasons, a consideration of the secondary evidence only serves to reinforce the conclusion that once-daily dosing was obvious to try with a reasonable expectation of success. The secondary evidence is consistent with Sandoz’s case that the skilled team would have considered once-daily dosing obvious to try in light of the results of the thrombin generation study; it certainly doesn’t undermine it.

469. It also provides corroboration of Sandoz’s witnesses’ evidence that once-daily dosing would be commercially attractive. The secondary evidence, which suggests the enthusiasm on Bayer’s part for the development of a once-daily dose is consistent with Professor Hirsh’s view. In this regard, Bayer, in its submissions, highlighted Professor Crowther’s evidence that it was very unusual to move from a twice-daily dosing regimen to a once-daily dosing regimen during the course of a clinical development program and that this would require “a large degree of confidence” that it is going to work. If that is what happened, as the documents appear to indicate, it only serves to illustrate the apparent attractiveness of oncedaily dosing from a drug company’s point of view.

470. I accept that Bayer would have additional data available to them that is apparent from the prior art, but for the reasons explained above, that is not an answer to an obviousness challenge.

Summary of conclusions

471. Sandoz has sought a declaration that the Patent, for once-daily dosing of the drug rivaroxaban, is invalid on the basis that it was not patentable within the meaning of the Act and the EPC because it did not involve an inventive step. It claims that the invention was obvious to a person skilled in the art in light of the prior art. Bayer claims that the invention was not obvious.

472. Each side has called expert witnesses to support their positions. For the reasons stated above, I consider that on the key issue, whether the skilled team would have considered once-daily dosing obvious to try with a reasonable expectation of success, the evidence adduced on behalf of Sandoz is far more convincing. I have concluded, therefore, that the evidence in these proceedings establishes that the claimed invention was obvious and did not involve an inventive step.

473. This conclusion is consistent with all foreign judgments, including two from EPC jurisdictions, in which the validity of the Patent has been finally determined. No material basis for distinguishing those decisions was identified. Although there are jurisdictions in which the Patent has been upheld, no such decision is final, and there are points of distinction in those cases, either by reference to the prior art relied on, or the conclusions reached on the evidence before the foreign tribunals. A decision that the Patent is invalid is, therefore, consistent with the principle of judicial comity.

474. In addition to the primary evidence in this case, secondary evidence was adduced. Although that evidence is of limited relevance because it provides only a partial picture of the development process actually undertaken by Bayer, insofar as the evidence illustrates anything, it tends to support the primary evidence of the Sandoz witnesses and to undermine that of the Bayer witnesses.

475. Accordingly, I propose making a declaration that the Patent is invalid. I will hear the parties as to the form of order and in relation to any issues arising.