Sandoz Limited & Anor. v Bristol-Myers Squibb Holdings Ireland Unlimited Company & Anor.

Apixaban, trade name ELIQUIS, is a successful drug for treating thromboembolic disorders. In a judgment of 7 April 2022, [2022] EWHC 822 (Pat), I held that a patent covering apixaban as such, and which also claimed apixaban for use in treating a thromboembolic disorder, was invalid. As a result, an associated SPC was also invalid. I will refer to that as “the First Judgment”, and to the trial which led to it as “the First Trial”.

The patent in issue at the First Trial had a priority date of 21 September 2001.

This trial concerned four further patents relating to apixaban (“the Patents”), all from the same family as one another and all having priority dates of 25 February 2010. They cover formulations of apixaban.

In the Introduction to the First Judgment I explained the parties. The situation was the same at this trial, except that Pfizer as well as BMS is a patentee. Nothing turns on that and I will refer to BMS and Pfizer collectively as “BMS”.

As was the case at the First Trial, infringement is admitted if the Patents are valid.

The Patents are:

Sandoz reached a settlement with BMS so that only the first Patent (which I will refer to as “`021”) remained in issue so far as it was concerned, and from that Patent, only claims 1-6 (it was these claims that Sandoz admitted were infringed if valid). That, however, is now only of historical interest (and possibly of relevance to costs) because the independently valid claims in issue dwindled over time, and by the start of the trial, only claim 1 of `021 needed to be addressed. All the other Patents and claims stand or fall with it.

When this trial began, `021 and the second Patent (ending `811) had been revoked by the Opposition Division of the European Patent Office. During this trial, which was prolonged for reasons explained below, the third Patent (`660) was revoked by the Opposition Division, and reasons were given on 27 June 2022, when the drafting of this judgment was at an advanced stage. In each case, revocation is stayed pending appeal (though BMS has not yet actually appealed the `660 decision, with time for it still running). Teva submitted that the 27 June 2022 reasons supported its case. BMS submitted that the prior art, arguments and evidence were different. In my view BMS is right about this. Although the reasons are broadly consistent with my conclusions, and I do not in any way criticise them, it would not be right to put weight on them and I do not do so. I should also mention, if only for completeness, that another related patent has been conclusively revoked by the Technical Board of Appeal for added matter, an issue I am not addressing (the patent was included in these proceedings at an earlier time but dropped out when conclusively revoked).

Much as I recorded in the First Judgment, Sandoz and Teva are separately represented but have made common cause, submitting joint skeleton arguments (except for a short supplemental skeleton on dependent claims submitted by Teva alone when such claims were still live) and sharing expert witnesses. Their Counsel again split the oral advocacy. They have run the same arguments with one exception, a prior art citation which only Sandoz ran and which has faded from the case. I will once more refer to Sandoz and Teva together as “the Claimants”.

BMS has also applied to amend the Patents. All that matters for the purposes of this judgment is that claim 1 of `021 is in proposed amended form. The amendment was to address an insufficiency allegation, which the Claimants accept that it has done (the allegation was of an omission to stipulate test conditions for a claim feature, and by the amendments the conditions were brought into claim 1 from a dependent claim). There is no issue over the amendments other than that the Claimants say they do not cure any invalidity.

As with the First Trial, there has been related litigation in two actions in Canada. It is of some, albeit modest, relevance because evidence given there by Prof Davies (BMS’s formulation expert) was used to cross-examine him here.

The evidence and argument took place live in Court. Unfortunately, however, Prof Davies contracted COVID and his recovery took a while. It goes without saying that this was no fault of his, and naturally the priority of all concerned was that he should make the fullest possible recovery, unburdened by this case. As a result, the trial started in April and did not end until June, despite being only a few days in Court.

I should mention that:

When opening skeletons were submitted, the issues were:

However, it seemed unlikely that the Press Release would add anything to the arguments over Carreiro, and so it proved. Wei was covered in the cross-examination of Dr Stott (for the Claimants) but also seemed unlikely to be of separate importance. Both were formally dropped towards the end of the oral evidence.

The parties each put in evidence from two experts:

As I shall explain below, the areas of dispute between the clinicians, never great to begin with, shrank still further during trial and had vanished by the closing arguments.

There was no fact evidence.

The Claimants’ experts were:

BMS’s experts were:

Prof Laffan was cross-examined only briefly, and the Claimants decided not to cross-examine Dr Weitz at all.

BMS made clear that they had no personal criticisms of Dr Stott. He was well qualified and a good, fair and clear witness.

The Claimants attacked Prof Davies in a number of respects:

I will start with the basic principles and then identify some other aspects of the law said to be relevant. I will also explain where these points were said to be significant to the arguments.

The basic approach is as set out in Actavis v. ICOS [2019] UKSC at [52] – [73]. At [63] the Supreme Court endorsed the statement of Kitchin J, as he then was, in Generics v. Lundbeck [2007] EWHC 1040 (Pat) at [72]:

The Claimants relied on Brugger v. Medicaid [1996] RPC 635 (at 661), as approved by the Supreme Court in Actavis v. ICOS for the proposition that an obvious route is not made less obvious by the existence of other obvious routes. I have commented on it in other recent judgments – it must not be taken too far. In the present case the Claimants do not rely on it to do much work: they merely said that if it was shown that particle size reduction and excipient optimisation were both obvious things to do to improve dissolution rate, the former was no less obvious because of the availability of the latter. In any event, however, I will have such regard as is appropriate on the facts both to the Brugger principle and to the caution needed in respect of it.

BMS relied on what Jacob LJ said in Leo Pharma v Sandoz [2009] EWCA Civ 1188 at [9]-[10]:

Picking up on the reference in [10] to the steps leading to the invention having to be “really likely” or “almost mandated”, BMS referred to Actavis v. ICOS where obviousness arose because although the 5mg dose in the claims could not be predicted, routine research would (BMS’s emphasis) have led to its being tested with success.

I thought there was a danger that BMS’s submissions were really just trying to approximate the facts of the present case to those in Leo, and/or to contrast the facts with Actavis, rather than identifying legal principles. The former case was an extreme one where the obviousness attack was that something wholly new would emerge in the course of rather blue-sky testing, the testing being done without any particular expectation of success. The latter was a case where the research was routine, and done for a solid reason with the expectation of getting a useful result, albeit one which numerically speaking could not be predicted in advance. Neither really maps onto the present case, where the obviousness case is that dissolution rate testing would be done (as is accepted), knowing that it would either give reassurance that all was well, or show that improvement was needed.

The Claimants referred to the well-known explanation of Jacob LJ in Actavis v. Novartis [2010] EWCA Civ 82 by reference to the 5 ¼ inch plate: arbitrary and non-technical limitations which do not solve a problem are not inventive for that reason. Birss J (as he then was) considered this in Apple v. Optis [2020] EWHC 2746 (Pat) and explained that having an arbitrary feature in a claim is not a ground of invalidity but it does mean that the feature cannot be relied on for inventive step.

This principle was invoked by the Claimants because claim 1 of `021 contains detailed numeric limitations as to dissolution rate and particle size. As I understood it, the Claimants wanted to head off an argument that for example, while it might be obvious to go for a “small” particle size, they could not show that specifically a D₉₀ of 89µm would be achieved. This is perhaps not quite the same as the 5 ¼ inch plate situation, but in any event I did not understand BMS to run a point based on the exact numbers. I have made findings about CGK particle size ranges below; I do not consider that there was any magic to 89µm and had BMS really run such an argument, I would have held that the precise limits in the claims were arbitrary; similarly with dissolution rate.

The Claimants argued that a patent monopoly must be justified by a technical contribution and that a patent which does not solve any problem is not inventive. They referred to Generics v. Yeda [2013] EWCA Civ 925 at [49]. I think that case is really more to do with the proper scope of a claim, but I accept the principle as stated by the Claimants. The reason this point mattered, at least potentially, is that the Claimants’ obviousness case assumed that the skilled addressee would find a problem with the dissolution rate of apixaban in a prototype formulation. But other than in the figures of the `021 Patent there was no demonstration of such a problem. The Claimants wanted to be able to say that either there was a problem of slow dissolution to be solved (in which case it would be identified routinely with smaller particle size being an obvious solution) or, if there was in fact no problem then there was no inventiveness. This was rather elaborate and I did not understand BMS to argue that there was no problem to be solved. Nonetheless, I consider the application of this principle briefly below.

It is well-established that cases of obviousness over common general knowledge alone have to be scrutinised with care because they may be too abstract and allow a case to be run which is unrealistically and unfairly free of real world problems and details (see e.g. Ratiopharm v Napp [2008] EWHC 3070 (Pat)). In the present case, obviousness is alleged over a concrete piece of prior art, Carreiro, but that just provides the clinical jumping off point for what is in substance a formulation invention, and the obviousness of the formulation aspects is indeed based on common general knowledge alone. So in my view I do have to exercise appropriate caution, and I have tried to do so.

It was accepted that the skilled team, working from Carreiro, would include a clinician and a formulator. It was also common general knowledge that the clinician would define what was desired and the formulator would seek to put it into effect. For reasons that I come to shortly, the issue of obviousness turned entirely on the formulator’s work. There was a trivial dispute about how many years’ practical experience the formulator would have. It does not matter. He or she would have a relevant degree or degrees, and practical experience. From here on, if I refer to “the formulator” I just mean the notional “Skilled Formulator” (the term used in the parties’ CGK document). This does not mean the addressee is not a team, it is simply that there is in general no need to say anything further about the clinician to decide the issues.

The parties provided a statement of the Agreed CGK. I have edited it down quite extensively to remove most of the general background clinical material, and some parts of the formulation section. My objective was to preserve the material most central to understanding this judgment while reducing its length and taking out material of limited importance or usefulness. My removing material does not detract from its agreed status.

The upshot of the clinical CGK is that apixaban was a promising drug that had reached advanced clinical trials and would be obvious to progress in an appropriate formulation. When I come to Carreiro I will explain that it identifies a number of doses (in particular 2.5mg and 5mg tablets) that BMS accepted would be obvious for the clinician to recommend to the formulator to make into tablets for twice-daily administration; these would be immediate-release formulations. The clinician would not have to have any further involvement relevant to the issues at this trial.

For many years before the priority date, warfarin was the main oral anticoagulant in clinical use for treating thromboembolic disorders including deep vein thrombosis, pulmonary embolism, stroke (often caused by atrial fibrillation), and acute coronary syndrome.

Limitations with warfarin (including a need for a tailored dose for each patient that was hard to identify) prompted a search for oral anticoagulants that could be given in fixed doses without coagulation monitoring and which were at least as effective and safe as warfarin. This search led to the development of the DOACs. [Weitz 1/7.19]

The DOACs were designed to be orally administered and were small molecule inhibitors designed to target components of the coagulation cascade namely either thrombin or Factor Xa alone. [Laffan 1/60; Weitz 1/7.20]

By 2010 two DOACs had been licensed for use in the UK - dabigatran, a direct inhibitor of thrombin, and rivaroxaban, a direct inhibitor of Factor Xa. Both had been licensed for use in the UK for the prevention of thromboembolic problems after joint replacements and were in advanced trials for other indications.

In 2010, the Skilled Clinician would also have been aware of another DOAC being developed, although not approved: apixaban. They would have been aware of apixaban from the scientific literature and conferences. [Laffan 1/63; Weitz 1/7.23]

The Skilled Clinician would have been aware in 2010 that apixaban was a direct acting oral Factor Xa inhibitor and that it was being trialled in a number of clinical trials, several of which were at an advanced stage (Phase III). [Laffan 1/64; Weitz 1/7.23].

The Skilled Formulator typically joins a drug development project after the drug discovery team has identified a small number of lead candidate compounds, but before studies in humans have been carried out. [Davies 1/46]

In the circumstances of this case, the choice/identification of a lead compound is not of relevance: the Skilled Formulator would be tasked with formulating apixaban specifically.

Starting with a lead compound, in this case apixaban, the development phase of the R&D process would initially comprise a series of pre-formulation studies, to gain an understanding of the physicochemical properties of the API. [Stott 1/38; Stott 1/39; Davies 1/49] During pre-formulation there is typically only a small amount of the API available for analysis. [Stott 1/39]

The data from the pre-formulation studies feeds into the subsequent formulation development and may highlight any potential difficulties that the Skilled Formulator may encounter in formulating an effective dosage form. [Stott 1/39; Davies 1/51]

Before they start work, the formulator will be advised of the proposed dose range and route of administration. [Davies 1/52; Stott 1/38]

The goal is to formulate a dosage form which:

The skilled team, including the formulator, will know that should any problem emerge with the API or formulation during later stages of the development phase, a significant delay and reworking of the formulation might be required. The R&D costs to bring a product to market are substantial and hence, any delay or setback will cost time and money and significantly delay the launch of the product. [Davies 1/55]

For a drug exerting its therapeutic effect in the systemic circulation the API must reach the blood / plasma. For the majority of drugs, there are four main processes that take place before the API in a tablet reaches the systemic circulation:

The process of drug absorption from oral solutions is different. There are no disintegration or dissolution steps because the medicinal ingredient is pre-dissolved in the solution formulation. [Davies 1/73; Stott 1/165]

Bioavailability can sometimes be described as being "dissolution rate limited" or "permeability rate limited", which refers to the slowest step that effectively controls the rate at which the API enters the body. [Stott 1/67; Davies 1/59; Davies 1/69; Davies 2/29].

The amount of API that gets absorbed, among other factors, usually determines the nature and extent of its clinical effect. If not enough is absorbed, the effect may be minimal or insufficient. Too much, on the other hand, can result in unwanted side-effects. Accordingly, absorption is an important consideration in formulation design. The level of API in the blood at a given time is commonly represented as a plasma concentration-time curve, an example of which is shown below: [Stott 1/48; Davies 1/66]

The following terms are used to describe plasma concentrations and would be familiar to the Skilled Formulator:

The relative amount of an administered dose of a particular dose of a particular drug that reaches the systemic circulation intact and the rate at which this occurs is known as "bioavailability". [Davies 1/119]

If all of the API gets into the bloodstream (like it will for intravenous drugs), then the bioavailability is 100%. If only a portion of the API gets into the bloodstream, the bioavailability will be less than 100%. For oral dosage forms, bioavailability is expected to be below 100% because many diverse factors can affect how much API is absorbed into the blood stream (e.g. incomplete absorption and first-pass metabolism by the liver and in the gut wall). The bioavailability of orally administered drugs can vary greatly between different drugs and different dosage forms of the same drug. [Stott 1/52; Davies 1/120]

Pharmacokinetic studies are used to measure the bioavailability of an API from a dosage form that is under evaluation by assessing the parameters AUC and C_max (discussed above). T_max may also be assessed. [Davies 1/121] The study of absorption, metabolism, distribution and excretion of an API is known generally as the field of pharmacokinetics. [Davies 1/126] The values of C_max, T_max and AUC are used in bioavailability studies to assess the rate of systemic absorption (C_max and T_max) and extent of systemic exposure (AUC) of an API from a pharmaceutical formulation. The value of C_max is used to assess the maximum plasma concentration value for the API. [Davies 1/125]

The terms "absorption" and "exposure" are often used synonymously when discussing "bioavailability", although strictly speaking, they are each slightly different concepts. Absorption is strictly the process of absorption of the API into the bloodstream from the gastrointestinal tract for an oral dosage form. Exposure is more synonymous with bioavailability as it is a measure of the amount of API that is absorbed into the bloodstream after an API is administered and can be evaluated by considering the rate (C_max and T_max), and the extent of exposure (AUC) of the API in the bloodstream and the maximum concentration of the API in the bloodstream (C_max). [Davies 1/128]

As noted, in the pre-formulation stage, the physicochemical properties of the API and its stability are tested. It is essential to get an understanding of these properties prior to the formulation of the API into a product. [Stott 1/70; Davies 1/75]

Standard pre-formulation tests would include: equilibrium solubility; stability; permeability; investigating alternate forms such as crystal forms, hydrates, solvates and salts; purity; particle size, shape and surface area; particle density, powder flow and compressibility. Excipient-API compatibility studies may also be done. [Davies 1/76; Stott 1/71]

The equilibrium solubility of an API (i.e. the amount that will make a saturated solution) is typically expressed as a mass of the medicinal ingredient that can be dissolved in a unit of liquid, such as milligrams per millilitre (mg/ml), although it can also be expressed in percentages and sometimes in parts i.e. number of parts of a solute dissolved in a number of parts of a solvent. This is typically determined by suspending an excess amount of the API in the solvent at a fixed temperature and stirring the mixture for a long period of time (usually many hours) until the concentration in solution remains constant. [Stott 1/60.1; Davies 1/80]

The equilibrium solubility of an API is typically one of the first properties of a new potential API to be measured in pre-formulation studies following routine protocols. [Davies 1/81]

The Skilled Formulator would typically test an API’s equilibrium solubility at human physiological temperature (37 ºC) in a range of media across a range of pH to mimic the changing pH of the GI tract, and usually over a time period of at least 24 hours in order to allow equilibrium to be reached. The aqueous equilibrium solubility is of particular interest as the environment in the GI tract is aqueous, but solubility in non-aqueous solvents will also be checked. If an API is an acidic or basic molecule and therefore capable of being ionised, the pH will affect its equilibrium solubility. If the API is non-ionisable, then the Skilled Formulator would expect the pH of the medium to have no effect on the solubility of the API. [Stott 1/73; Davies 1/80-81]

The equilibrium solubility is not a measure of how quickly a substance will dissolve. The latter is called the "dissolution rate". The dissolution rate can be shown in the Noyes-Whitney equation below. According to this equation equilibrium solubility and surface area are factors affecting the dissolution rate of an API in vivo.

Dose is considered by the formulator in terms of the "dose-solubility ratio". The reason for this is that the Skilled Formulator would want to put the equilibrium solubility of the API in context with the amount of API that actually needs to dissolve. [Stott 1/76; Stott 2/8; Davies 1/85-86]

The formulator would be told which dose(s) or dose ranges to formulate by the clinician. [Stott 1/38]

The propensity of an API to cross the intestinal membrane is a function of its permeability. The extent of permeability depends on the chemical structure of the API (including its molecular weight of the API, its lipophilicity and the extent of ionisation), the mechanism of transport of the API across the membrane, the extent to which the API is metabolised within the cells of the intestinal membrane, and any counter-efflux. [Davies 2/14; Stott 1/65]

Along with equilibrium solubility, the permeability of a compound is an inherent property of a new API to be considered in pre-formulation studies. [Stott 1/71; Davies 1/84]

Permeability is not a characteristic which is particularly easy for the Skilled Formulator to influence. This is because the Skilled Formulator cannot typically add anything to the formulation that changes the inherent properties of the API. There are agents that may modify the barrier properties of the GI membrane (permeation enhancers) but these have very limited applicability and often cause deleterious side effects. [Stott 1/66]

IDR testing is a method of looking at the dissolution rate of the API itself. IDR testing employs methods to control factors such as surface area, temperature, agitation or stirring speed, pH and ionic strength of the dissolution medium. The IDR is obtained by exposing a disc of compressed API to the dissolution medium under specified conditions. Dose is not taken into account. [Stott 1/footnote 11; Davies 2/19; Davies 2/20]

APIs can exist in different physical forms and these will be investigated at the pre-formulation stage. APIs can be in crystalline or amorphous form (or sometimes an intermediate form, i.e. crystalline with some amorphous content). In crystalline structures, the API molecules will be in a regular and orderly structure, whereas in amorphous APIs the molecules will be arranged in a random manner. The Skilled Formulator would usually select a crystalline form, rather than develop an amorphous form, if it is possible to do so, as crystalline forms are more stable (physically) and therefore less prone to change form during and after product manufacture. [Stott 1/80]

During pre-formulation the Skilled Formulator will investigate the characteristics of the particles of the API such as particle size, shape, and the compressibility of the API. [Stott 1/88; Davies 1/76]

Particle size distribution may be presented in the form of a % of particles that have an equivalent diameter of less than, or more than, a particular value, typically given in µm and displayed as D_V[%] (or D_[%]). For example, a hypothetical population of particles where 90% of the particles have an equivalent volume diameter less than 200 µm would be described as having a particle size D_v90 (or D₉₀) of less than 200 µm. [Stott 1/91]

The shape of the particles will also be evaluated, usually by microscopy. [Stott 1/93]

An immediate release dosage form is designed to immediately release the drug that is inside the tablet once it has been swallowed (discussed above). Various well-known excipients like disintegrants are included to aid in the tablet breaking up after it has been ingested. [Stott 1/37; Davies 1/59] By contrast, modified release formulations release the API over a longer time period. [Stott 1/37]

Pharmaceutical formulations typically contain the API along with excipients. Excipients are usually categorised according to their function and include:

A surfactant reduces the surface tension at the interface between the solvent and solute. [Stott 1/103.2; Stott Supplementary 1/4.9; Davies 1/107; Davies 1/113; Davies 1/137] Details of commonly used surfactants, including recommended amounts to be used in pharmaceutical compositions, would have also been available to the Skilled Formulator in textbooks such as the Handbook of Pharmaceutical Excipients. For instance, some commonly used surfactants at the priority date and recommended ranges for specific applications (as a % concentration in the total pharmaceutical composition) are set out below in Table 1 below:

As noted, the disintegration of a tablet can be influenced by well-known excipients, such as disintegrants. Disintegration can also be influenced by the type of process used to manufacture the tablet, the selected process parameters, and the amount of force used to compress the tablet. [Stott 1/56]

A key decision to be made is the manufacturing method to be used and this will be decided early on in the formulation development process. The manufacturing method itself and selected process parameters can affect many of the tablet characteristics, including the disintegration, hardness, friability or dissolution rate of the tablet (and therefore bioavailability of the API). Excipients can be used to try to counter these effects. [Stott 1/96; Davies 1/58]

The manufacturing process typically involves mixing the API with the excipients and then compressing the resultant powder blend into a tablet. It may also include an agglomeration or granulation step used to impart favourable properties on the powder blend for subsequent processing steps. The ways in which tablets are made include direct compression, dry granulation and wet granulation. [Stott 1/45; Stott 1/96; Davies 1/58]

Direct compression involves compressing the powder blend into a tablet without further manipulation. In granulation primary powder particles are made to adhere to form granules containing multiple particles in order to modify their flow or compression properties or to facilitate the uniform distribution of the API. [Stott 1/96; Davies 1/58]

In dry granulation, the API and excipients are mixed and aggregated typically at high pressure to form granules without the use of a liquid. [Stott Supplementary 1/4.12]

Wet granulation involves the mixing of the API and excipients with granulating fluids, such as water, ethanol or isopropanol, and the subsequent removal of this liquid by drying. Water is commonly used as a granulating fluid. Binders are added to ensure that the API and other excipient particles aggregate and remain bound together once the granule is dry. [Stott Supplementary 1/4.13]

The dissolution testing conditions employed can affect the dissolution rate and hence the results obtained on measurement. The USP therefore sets out guidance on the dissolution test conditions that should be used by the Skilled Formulator, in order to ensure standardisation across the pharmaceutical industry. The USP identifies four different apparatus for testing the dissolution rate of solid oral dosage forms. In terms of apparatus selection, Apparatus I (USP I) and Apparatus II (USP II) are used most frequently for solid oral dosage forms, with Apparatus II being preferred for tablets. [Stott 1/99; Davies 1/115]

The following factors of dissolution testing are broadly standardised by the pharmacopoeias:

The Skilled Formulator will sample the dissolution test. The sample will then be analysed to determine the concentration of API present. [Stott 1/101]

There are a number of different methods available to the Skilled Formulator to reduce the particle size by compression forces, impact, cutting and/or attrition. The equipment selected depends on, amongst other things, the target particle size (or sizes) and the properties of the material. A common technique is milling. Alternatively, a microniser can be used. Particle size reduction any lower than 1 µm would typically be referred to as "nanosizing" and is not typically used in tablet applications. [Stott 1/103.1; Stott Supplementary 1/4.4]

At the priority date, the Skilled Formulator would have been aware of different types of milling / micronizing methods using different types of mills such as cutter mills, hammer mills, pin mills and fluid energy mills can be used for the milling/ micronizing step. Broadly speaking, each of these mills applies mechanical or kinetic energy to break down particles to a desired level. [Stott Supplementary 1/4.6]

The choice of an appropriate method would be dictated by various factors including the target particle sizes required, the physical nature of the particles and cost, and the method chosen would be performed by a person with expertise in particle size reduction techniques. [Stott Supplementary 1/4.8]

Bioequivalence is a measure of whether two different formulations containing the same amount of the same API are equivalent in terms of their absorption. Bioequivalence studies are typically based on AUC and C_max values although T_max may also be assessed. [Stott 1/105; Davies 1/129-130]

Bioequivalence is achieved when there is no statistically significant difference between the tested pharmacokinetic properties of two products. [Stott 1/106; Davies 1/130] The formulator would have a reasonable understanding of the basic methodology, and would be able to interpret the data generated from bioequivalence studies. [Stott 1/108; Davies 1/131]

The Biopharmaceutics Classification System ("BCS") is a well-known method for classifying APIs according to their solubility and permeability. [Stott 1/78] It was originally published by a group of four leading formulation scientists in a well-known 1995 article, "A theoretical basis for a biopharmaceutic drug classification: the correlation of in vitro drug product dissolution and in vivo bioavailability", G. Amidon et al., 12 Pharmaceutical Research 413 (1995) 12: 413-429 ("Amidon"). [Davies 1/88] The Skilled Formulator would be aware of this paper and the BCS system. [Stott 1/110; Davies 1/88]

In accordance with the BCS, APIs can be classified into four categories:

An API has ‘high solubility’ under the BCS where the highest dose strength is soluble in 250 mL or less of aqueous media over the physiological pH range (of pH 1-7.5) at 37ºC, calculated using the API’s equilibrium solubility value. This can also be described as a "dose-solubility ratio" of less than 250 mL. The 250 mL is supposed to represent the minimum volume anticipated in the stomach when an oral dosage form is taken with a glass of water by a fasting patient. [Stott 1/111]

The other criterion the BCS uses to base its classification is the API’s permeability. An API has ‘high permeability’ if it is at least 90% absorbed or where permeability has been established in vitro. [Stott 1/112; Davies 1/94]

The BCS was originally developed as a framework for procuring biowaivers. [Davies 1/90] In the context of obtaining a biowaiver, instead of conducting in vivo bioequivalence studies to prove bioequivalence to the earlier product, a biowaiver can be applied for to demonstrate bioequivalence by in vitro dissolution testing alone. [Stott 1/113]

The parties also prepared a document indicating the disputes about CGK. Regrettably, they could not agree what was not agreed. This may be inevitable to a minor degree in some cases, but one would hope that it would happen only in small measure, and at the periphery.

The situation when opening skeletons were put in was that the parties did identify by agreement two disputed issues on the clinicians’ CGK; others were not agreed. None of them matters any longer because the obviousness arguments all turn on formulation. So I am not going to resolve them.

On the formulation side, one disputed issue was put forward by the Claimants, broken into three sub-issues. BMS put forward ten. The importance of these all changed during the trial and I am going to organise the CGK disputes in the way I think most helpful to identifying the real issues at the end of the trial.

The central arguments on the formulator’s CGK concerned the BCS system. There were a number of aspects to this which are interrelated.

It is worth reiterating the starting point for these disputes:

A nuance is that even where some drug was not dissolved in the stomach, then permeation might still be slow enough that it was the limiting factor, with the rest of the drug dissolving in the gut.

The key disputed matters on this aspect of CGK were:

These points were all explored in the context of a number of contemporaneous documents. I will deal with them in chronological order, save that I group two EMEA documents together.

Amidon is a 1995 paper entitled “A Theoretical Basis for a Biopharmaceutic Drug Classification: The Correlation of in Vitro Drug Product Dissolution and in Vivo Bioavailability”, and which provided the basis for the BCS system. It was clearly CGK, as was common ground before me.

Amidon defined the four classes in Table II:

And it contained the following commentary on Case 1 and Case 3 (which correspond to Class I and Class III):

The FDA published a 1997 document entitled “Guidance for Industry”, subtitle “Dissolution Testing of Immediate Release Solid Oral Dosage Forms”. The following parts were relied on in particular at trial and the subject of oral evidence:

I need not quote the text, but note that section VI of the document deals with the relevance of dissolution testing and profiles for “SUPAC-IR”, which concerns the regulatory requirements surrounding changes to manufacturing site, or scale, and for “biowaivers” for lower strengths of a dosage form.

The FDA issued another guidance document in 2000. It was also entitled “Guidance for Industry”, but with the subtitle “Waiver of In Vivo Bioavailability Studies for Immediate-Release Solid Oral Dosage Forms Based on a Biopharmaceutics Classification System”. It had a more intense and detailed focus on biowaivers and the procedures for getting them than did the 1997 document. Its contents relevant to this trial are very similar to the 1997 document and so I will limit how much I quote. The key section the subject of evidence and argument was this:

So this was a slightly less stringent parameter than the 85% in 15 minutes contained in the 1997 document.

I note that both FDA documents cited Amidon.

In Europe there was similar guidance for like reasons. A specific document discussed at trial was an EMEA 2001 “Note for Guidance on the Investigation of Bioavailability and Bioequivalence”; it was exhibited to both formulator experts’ reports but not the subject of specific cross-examination. I was also taken to a 2010 document called “Guideline on the Investigation of Bioequivalence”; this was only introduced for the cross-examination and not put in with any expert’s report. So the coverage of these in evidence was a little patchy and it was not really clear that the second document would have been CGK by the priority date. Each document refers to biowaivers being potentially available if 85% of the drug is dissolved in 15 minutes. The 2010 document also allows for the possibility of 85% dissolution in 30 minutes, with measurements at more time points being required.

“Aulton’s Pharmaceutics - The Design and Manufacture of Medicines” is a standard work and, I find, a source of CGK of the classic kind. I was referred to the 3^rd edition, from 2007. Both experts had relied on it.

The following text was relied on:

This is more general than the other documents; it does not contain any concrete numerical indications of what rate of dissolution should be targeted. BMS relied on it as saying that the BCS system was a way to classify drugs and predict bioavailability and that in relation to Class III drugs the reference to “release” concerned disintegration because sufficiently rapid dissolution could be assumed. The Claimants took the position that “release” included dissolution and that the reader would understand that dissolution needed to be ensured and would not be left to chance. I agree there is some ambiguity; I prefer the Claimants’ position but do not think it matters very much. In any event, any ambiguity over this text has to be seen in the overall context of all the documents.

This was a chapter entitled “Application of the Biopharmaceutics Classification System Now and in the Future” in a book entitled “Drug Bioavailability”. It was put in as an exhibit by Prof Davies.

The arguments over Abrahamsson were quite detailed and rather fiddly. As with the slight ambiguity in Aulton, I think the position as to CGK has to be considered in the round in the light of all the documentary sources. I also think that the end point in relation to Abrahamsson is not materially different from the other documents. However, since it was the focus of quite a lot of argument and in case it should matter, I will deal with the points taken.

Abrahamsson contains the following:

Thus there is a general introduction of the BCS system and an explanation of IVIVC (“in vitroin vivo correlation”). It may be beguiling to think, on first encountering the term, that IVIVC would be an unqualifiedly good thing because it would enable a prediction about the in vivo situation to be made from an in vitro measurement. In fact the position is very different: for BCS Class I drugs, one wants fast dissolution such that absorption of the drug is only affected by gastric emptying. Provided the dissolution rate remains above that level, a change to it will not affect in vivo behaviour – there is no correlation - and that is a good thing.

The authors seek to illustrate this by reference to Figure 19.12, reproduced above. It is potentially a bit confusing because the x-axis, “In vitro dissolution”, is time to dissolve 50% of the drug, so dissolution rate is dropping rather than increasing from left to right.

There is also a problem in understanding Figure 19.12 because it relates to “a fictive class I drug”. It is unclear whether this means a variety of fictive Class I drugs with different dissolution rates, or a single Class I drug whose dissolution rate can be varied, perhaps by changing formulation. Before me, BMS argued that it was a single drug and said that Dr Stott had accepted as much. The Claimants said it was a range of drugs. I do not think Dr Stott did accept that it was a single drug, or at least not clearly so. However, so far as it matters (and I think it does not), my view is that the text is simply not clear as to whether it is one drug or a number. The authors’ intention was to give a very general indication that there would be a range of dissolution rates, represented by the plateau in the left hand side of the figure, over which gastric emptying would be the limiting factor, but then there would be a tipping point, beyond which dissolution rate would increasingly limit absorption.

The fact that the authors’ intention was as general as this is supported by the fact that there are no values on either axis.

Figure 19.12 and the text that I have quoted above concern class I drugs. Class III drugs are addressed on the following page:

This is a very condensed explanation seeking to transpose Figure 19.12 to the case of a Class III drug where permeability is low. In my view it would be plain to the reader that Figure 19.12 could not simply be applied without modification for a Class III drug, because the figure does not show permeability or its effect. In my view the authors were just saying, as I think is clear from the text, that for class III drugs there would be a conceptually similar tipping point where the dissolution rate became slow enough that the time for permeation would be reduced, and more so as dissolution became yet slower.

None of this is a criticism of the authors of Abrahamsson. Had they been trying to answer exactly the same questions as arise at this trial they no doubt would have gone into more detail and expressed themselves differently. They were just dealing with matters at a rather more general and conceptual level.

BMS characterised its position in its written closing submissions as follows (emphasis in the original):

Contrary to BMS’s position, I think it is useful to start with whether the dissolution rate could in fact hinder absorption. The answer is quite clearly that it could. It is a basic proposition because if dissolution is too slow then not all the drug is available to permeate the gut wall when, or soon after, it leaves the stomach. This is so basic that in my view it must also be CGK; any formulator would understand it and the expert evidence supported it. To be fair, Counsel for BMS backed away in oral closing submissions from the proposition that the formulator might not understand that slow dissolution could affect absorption. The more realistic line that BMS took, and which Prof Davies had maintained, was that with a Class III there would be a very high degree of confidence that it would dissolve fast enough that it would not hinder absorption. He also said that the degree of confidence would be so high that the formulator would not trouble to test IDR at an early stage – I deal with that below.

As I understood it, Counsel for BMS accepted that a drug’s being Class III could not guarantee sufficiently rapid dissolution, if only because the classification depends on equilibrium solubility, which tends to be correlated to dissolution rate, but is not the same thing, and so sometimes a soluble drug may be slow to dissolve. In any event, even if not accepted that would be my finding, to the CGK standard.

These basic propositions are, in my view, all supported by and consistent with the documents that I have quoted above. They tend to explain matters in terms of having certainty: that if the dissolution rate is in fact established to be fast (whether 85% in 15 minutes or in 30 minutes) that there will not be a problem. But the formulator would understand that this clearly implies that if dissolution gets slower, there is a risk of its limiting absorption.

Understood in this light, the disagreement between the experts was more a matter of emphasis than a binary one. Prof Davies’ consistent position was that the formulator’s confidence for a Class III drug would be high, and so they would not regard 85% in 15 or 30 minutes, or any similar values, as a necessity or as an invariable criterion to be adopted at the start of a project. Adequately fast dissolution, he said, might be achieved with less demanding figures.

I agree that if they thought about it, the skilled formulator would accept that such was possible. But based on Dr Stott’s evidence, and what I think is a very clear picture that emerges from the documents I have referred to above, I conclude that the CGK was that the safe and natural way to proceed with a Class III drug was to aim and test for a dissolution rate of about either 85% in 15 minutes or in 30 minutes. With an expensive drug development process there would be no incentive to take a chance, and every reason to play it safe. The motivation to play safe would be for regulatory reasons to do with SUPAC and the possibility of a later biowaiver, and more fundamentally simply to have confidence that there would be one less thing to worry about going wrong.

I note that this conclusion is fully consistent with what Prof Davies said in his Canadian evidence. For example, in a report of 21 August 2020, in a section dealing with CGK, he wrote:

Reference 30 is Amidon. Prof Davies sought to qualify this by saying that it was in response to the other side’s experts’ reports in that case but I could not see how that would make any difference or qualify the simple meaning of his statement about “target”. I also note the reference to “solution-like behaviour” which supports my view that it was a known concept, and desirable.

It is convenient at this point to sweep up two other points on the evidence, which are related:

I reject both of these points. In his first report at e.g. paragraph 114 Dr Stott had referred to Amidon’s teaching that “all the API dissolves in the stomach and the rate of absorption is controlled by when the API leaves the stomach”. This is talking about solution like behaviour in different words. I also repeat that Prof Davies had no trouble with the concept and had used the expression in his Canadian evidence to which I have referred above.

In the same paragraph Dr Stott referred to 85% in 15 minutes and noted in a footnote that the FDA later reduced this to 85% in 30 minutes. There were many other occasions when Dr Stott referred to 85% in 30 minutes but it is wrong to say that he did not also raise 85% in 15 minutes.

The technique of IDR testing was part of the agreed CGK, as described above. Prof Davies’ evidence was that for a Class III drug he would probably not bother with it, and certainly not at an early stage. It is not an important point given the nature of the obviousness argument as explained below, but I prefer Dr Stott’s evidence that IDR testing would be done, including for Class III drugs, and generally at an early stage. I think this conclusion is also consistent with the somewhat cautious attitude I have described above in relation to dissolution rates.

Dr Stott also described something he called “kinetic testing”. This was a sort of shortcut in which aliquots would be taken at various time points during equilibrium solubility testing so as to give a rough idea of dissolution rate. It was said by BMS that Dr Stott had not mentioned kinetic testing in his written evidence. I find that he had, albeit rather briefly. More importantly, Prof Davies disputed that it was CGK. On this point, I agree with BMS that the technique was not CGK. I do not doubt that Dr Stott had used it, but that does not mean it was widely used or known. Prof Davies said at one point that he supposed that some industry players might use it, but in context he did not mean that he accepted that it was widely used. There were no documents referring to it as CGK.

It was common ground that dissolution rate would have to be tested once the formulator reached the stage of having prototype formulations.

There was no dispute that the CGK was that the first thing that the formulator would do if it were found that prototype formulations were too slow to dissolve was to assess and if necessary improve the disintegration of the tablet. This makes sense because, as was common ground, if the drug did not escape the tablet in the first place it would not have the opportunity to dissolve and other tweaks would not help.

It was agreed to be CGK that particle size reduction was a way to improve slow dissolution rates. Other ways existed such as changing or improving excipients. BMS said the latter would be preferred to the former. I do not think any such per se preference was made out. Both approaches would be in the CGK notional “toolkit” of the formulator.

Aultonsays that most pharmaceutical powders have crystals in the range of 0.5-300µm, but that distributions are often smaller, such as 0.5-50µm, to help blend homogeneity.

The experts agreed that there comes a lower limit where matters such as handling difficulties, aggregation and the like start to pose problems. It is hard to be precise, but I find that the CGK was that that lower limit was at about 10µm.

Prof Davies said that the formulator would not reduce particle size so readily with Class III drugs as they would with Class II or Class IV; subject to that limitation he accepted that Aulton’s range was CGK. I do not think that the difference in approach with different Classes was soundly based in any document and it may have been a consequence of his attitude that the formulator would have an overwhelmingly high degree of confidence of the dissolution rate of Class III drugs, which I have rejected.

I therefore find that Aulton’s suggestion was CGK. Dr Stott’s evidence was that 10-100µm was typical. I think this reflected the concern mentioned above about very small particle sizes, and an upper limit which is somewhere between the two values mentioned by Aulton (50µm and 300µm).

In any event, and cutting to the chase, the D₉₀ value of 89µm in claim 1 of `021 is well within the CGK of the formulator and would be regarded by him or her as completely normal and unsurprising.

`021 seeks to explain the invention first of all in paragraphs [0005], [0006], [0011]. [0005] says:

The last sentence of [0005] is referring to tables 3 to 5 which were not said to be relevant to my task. [0006] and [0011] are as follows:

Prof Davies said that paragraph [0011] contained a mistake and that “dissolution rate” should instead have said “absorption”. The Claimants responded that it was unclear. I agree with Prof Davies that the reader would think there was probably a mistake and “absorption” is quite likely what was meant. I agree with the Claimants to the more limited extent that one cannot be entirely sure as to the error or what was intended. I think the overall gist about what is said to be inventive or “surprising” is reasonably clear, but I do not agree that it is surprising.

The statements in these paragraphs are consistent with the case advanced by BMS at trial, but that does not mean that they are correct. As I explain in relation to CGK and when I come to obviousness below, the formulator would realise by routine means that apixaban was a Class III drug and would appreciate that that meant it was “high” solubility as the BCS uses that term. But he or she would know that that did not guarantee that the dissolution rate did not affect bioavailability. They would know that it might have an effect if it was too slow. They would not be particularly surprised that apixaban had turned out to be a case of a Class III drug where dissolution rate affected absorption, and they would not be at all surprised that the patentee had tested that experimentally.

Of course, how the skilled formulator would react to a patent’s teaching is not part of the assessment of obviousness. Obviousness has to be assessed without knowledge of the patent. I am dealing with these paragraphs because of the way that BMS relied on them and sought to characterise the inventive step: as the realisation that for apixaban, dissolution rate does affect bioavailability.

Paragraph [0013] gives test conditions for assessing dissolution rate. Nothing turns on the details of this.

Paragraphs [0014]-[0017] give various definitions which I need not go into because of the agreed CGK.

Paragraphs [0018]-[0020] give details of the synthesis of apixaban and crystallographic parameters of two forms of it. Again, nothing turns on this.

Paragraphs [0021]-[0024] give information about granulation processes and refer to the use of a surfactant. This is all consistent with the agreed CGK. Paragraph [0025] says that the apixaban tablets may contain 2.5 or 5 mg of the active substance, usually taken orally twice a day. As will appear below, these are among the dosing parameters disclosed in Carreiro.

Paragraphs [0026]-[0029] give some experimental details and reiterate the use of a surfactant.

The experimental data begins at [0031] to [0033], which relate to some tests with dry and wet granulated tablets, but as I have already said, the parties’ arguments did not focus on this so I pass over it.

Paragraph [0034] introduces Table 6 and Table 6a as follows:

The argument before me focused on Table 6 which, by contrast with Table 6a, uses a surfactant (SLS). The role of a surfactant is, as I have said, explained in [0024] and [0028], although the skilled formulator would know it from their CGK in any event.

Table 6 is as follows:

Tablets A and B are said to have been made by wet granulation and Tablet C by dry granulation. Other than that, details of their composition are not given, and nor is particle size. It can be seen that Tablet B was the slowest-dissolving. By contrast, the priority document for `021 did give particle sizes. The Claimants made a criticism of this, and BMS responded that one does not need composition or indeed particle size to determine whether dissolution rate affects bioavailability (which is the subject of the experiments that follow, as described below). I agree with BMS on this, although it does not affect my reasoning, and it is rather odd to have removed the particle sizes.

Paragraph [0035] gives some further explanation and introduces Figures 1 and 2 which give data for C_max and AUC for solutions and for tablets A to C:

Figure 1 is as follows (it is not necessary to reproduce Figure 2 for the purposes of this judgment):

From this it can be seen that the slower dissolving tablet (77% in 30 minutes) is not quite bioequivalent to a solution, but it is on the cusp of being so. It is the basis for the dissolution rate limitation in claim 1. Counsel for the Claimants accepted that the fact that the tablet was just outside the limit for bioequivalence was not a separate reason for invalidity. 77% in 30 minutes is very close to the normal target dissolution rates from the CGK; it is very slightly less ambitious.

Paragraph [0036] then introduces Figures 3 and 4, which plot the percentage dissolved in 30 minutes against the D₉₀ in microns. Figure 3 is for a 2.5mg apixaban tablet whereas Figure 4 is for 5mg. This does not make much difference, but since both were mentioned in evidence I reproduce them both here:

Figures 3 and 4 each show the “minimum dissolution requirement” by means of the horizontal dotted line and which corresponds to the 77% figure in the claims of the Patents (which, as I have said above, comes from [0035]).

The Claimants said the specific particle size of 89µm taken into the claims is not important because the dissolution rate threshold of 77% in 30 minutes is reached with particles of around 110µm or 120µm for tablets of 2.5mg and 5mg respectively, as shown by figures 3 and 4. So, the Claimants say, the precise size of 89µm is arbitrary in that any value below 110µm or 120µm could have been chosen. BMS responded by relying on [0036], which explains that for 5mg tablets the particle size at the 77% in 30 minute cut-off is 119µm, but that the claimed size limit is 89µm to allow for variability in dissolution measurements. I agree that a somewhat different number than 89µm could have been chosen, but the approach of leaving a margin of safety over experimentally determined limit is a rational one, so to call the value entirely “arbitrary” is an overstatement. A slightly different number would have done the same job, but the number used represents a safety margin over what was determined experimentally and is meaningful to that extent.

The experiments described in the specification are, I find, routine ones done for routine, CGK reasons. It could not have been predicted in advance of doing them that apixaban’s dissolution rate could affect bioavailability, but nor would it have been surprising when that result was obtained. The particle size experiments are also routine and done for CGK reasons.

As I have already said, I need only consider claim 1 of `021, which, as proposed to be amended, is as follows:

This is a combination of granted claims 1 and 2: features (f) and (g) were originally in claim 2. This had led to an insufficiency attack, as I have already mentioned, which was that the claim was uncertain if the way of doing the dissolution test was not specified. The combining of claims 1 and 2 cured that.

However, it was not said by BMS that features (f) and (g) contributed to inventive step in any way (as opposed to clarifying the scope of the claim), and the same applies to feature (e), which was present as granted. Feature (c) is trivial and irrelevant to inventive step.

Feature (a) requires up to 5mg crystalline apixaban, and Carreiro, as I will come to, discloses tablets of either 2.5mg or 5mg.

So what really matters to my decision is whether it was obvious to make an apixaban tablet of 2.5mg or 5mg with the particle size profile of feature (b) and the dissolution rate of feature (d).

Obviousness is alleged only over Carreiro. As I have mentioned in the Introduction above, there is no dispute that Carreiro would motivate the clinician to recommend the formulator to make immediate release tablets of apixaban in 2.5mg and 5mg doses. Although this is agreed, I will still briefly summarise Carreiro’s teaching that justifies the conclusion, before coming to the real meat of the obviousness arguments, which are on the formulation side.

The authors of Carreiro are two workers at Lenox Hill Hospital in the USA.

The abstract of Carreiro is a fairly general introduction to apixaban against a historical backdrop:

A longer introduction section covers the same ground, and then there is an explanation that apixaban was designed as a follow-up compound to razaxaban. Section 3 gives some pharmacokinetic and metabolic attributes of apixaban in animal models:

Information about pharmacokinetic and pharmacodynamic studies in humans are given in section 4, which begins with the following:

Section 5 moves on to discuss clinical trials. The section is organised according to the various indications under investigation. I will not set out the narration because the overall position is helpfully set out in Table 1:

Focus during the oral evidence was mainly on the APROPOS trial, the ADOPT trial, the APPRAISE trial, the ARISTOTLE trial and the AVERROES trial. One can see that the doses used varied, but 2.5mg twice daily and 5mg twice daily were used or planned in a number of instances.

Section 7 is the conclusion and is as follows:

The upshot is that Carreiro describes tests of apixaban in a variety of clinical trials. It gives the dose sizes and dosing intervals used in the trials and basic pharmacokinetic and pharmacodynamic parameters. It explains why direct Factor Xa inhibitors would be attractive. The only completed trials that it describes are phase 2 trials and its optimistic tone is qualified as a result, with statements that the outcome of the phase 3 trials cannot be taken for granted.

I agree with BMS that Carreiro contains no specific teaching about actually formulating apixaban, although the information that I have just summarised would provide necessary and useful inputs to the formulation task.

I have covered the skilled addressee and CGK above.

The necessary steps to the claims of `021 are the choice of D₉₀ and dissolution rate. These are not changes to anything in Carreiro (it is a “filling in gaps” kind of obviousness case) but have to be supplied from CGK. Hence why I said above that this is in a sense a CGK-alone attack on a formulation patent, albeit based on the clinical information in Carreiro. I have said that I have to take that into account, and I will.

It would be the formulator’s task to seek to implement the clinician’s recommendations. The formulator would come to it with the understanding that the task was a feasible one, since tablets of the relevant doses had been used in the clinical trials listed in Carreiro, although, as was common ground, the formulator would not have any information about how they had been made.

To articulate the parties’ arguments I will have to set out a number of steps in the argument. I will need to consider in due course whether this means that there is a Technograph issue (i.e. injecting hindsight by breaking the invention down into small individual steps).

The Claimants’ case was that the formulator would first measure the equilibrium solubility of apixaban during routine pre-formulation tests. This much was common ground, as was the fact that the formulator would find that the aqueous solubility of apixaban was relatively low, at 40µg/mL.

Next, the Claimants said that the formulator would regard the low aqueous solubility as a warning sign that apixaban might have a slow dissolution rate so as to cause a problem in vivo. I agree with this but it must be noted that the Claimants’ case is only that this would be a warning sign, not a trigger in itself for further specific steps at that early stage.

The Claimants also argued that the formulator would do kinetic solubility testing and IDR testing. I agreed with the latter but not the former, given my conclusions on the CGK. If done, these could raise further warning signs.

The parties agreed that the formulator would make prototype formulations and test them. The Claimants’ position is that by this stage and probably a good deal earlier it would have been obvious to settle on a target dissolution rate of 85% in 15 or 30 minutes. BMS’s position is that the formulator would have had a lot of confidence that the dissolution rate would be all right, simply from apixaban at the doses under consideration being a Class III drug (or alternatively that a much less ambitious target would be chosen). For reasons given in connection with the CGK above, I agree with the Claimants on this issue.

Next, and assuming that there was a problem concerning dissolution rate, it was common ground that the CGK would lead the formulator first to check and if necessary improve the disintegration.

The Claimants submitted that if there remained a problem, one CGK way to address it would be by reducing particle size. BMS did not really dispute this, but said that excipient optimisation would be a more preferred route. I find that it would be uninventive to address dissolution rate issues by choosing a more appropriate particle size, and I have found in dealing with CGK that the size specified by the claims of `021 is well within the CGK range. Exactly what size would be chosen would depend on the specific circumstances of any given case including the details of the prototypes.

I recognise that this analysis involves multiple sequential steps, as Counsel for BMS emphasised in closing submissions. I also recognise that the attack is a rather conceptual one based on CGK alone. It does not involve putting forward a concrete, complete recipe, for example. These are reasons for caution and I have consciously aimed to take them into account.

Nonetheless, I am of the clear view that the obviousness case is made out. On my findings as to the CGK it would be an obvious choice and probably the most obvious choice faced with the complex and costly task of drug development to aim from the outset (and if not then certainly in due course) for the well known goals of either 85% in 15 minutes or 85% in 30 minutes (both are within the claims of `021). I think Prof Davies’ confidence arising from apixaban being a Class III drug was much greater than the notional formulator would have, but it is also rather beside the point; it could not be a reason why it would not be obvious to set a concrete target recognised by the art and by the regulators and then test it.

Although the attack involves a number of sequential steps, they are the systematic sequence known to the CGK. For example, trying to remedy disintegration first would be basic and logical.

As to particle size, as I have already said, I think the availability of other possible solutions to slow dissolution, particularly excipient optimisation, is a true Brugger situation: both routes would be obvious. I am fortified in this view by the fact that the claim covers very typical particle sizes which would not be regarded as unusual or problematic. BMS’s case might be improved if there was e.g. such a small size required that handling would become a problem, but that is not the case. The limit in the claims of the Patents is entirely uninventive.

It cannot be assumed that the formulator would in fact experience any problem. It could perhaps be the case that by careful and/or lucky choice of excipients the targeted dissolution rate would be achieved without particle size reduction; it might also be the case that small enough particles would be chosen in the first place. BMS did not press this hard and in my view it cannot help to meet the case of obviousness. The claimed dissolution rate and particle size ranges are, for reasons given above, obvious ways to address a potential problem. If and to the extent there were in fact no problem then the Patents do not provide a technical contribution. I should also make clear that BMS cannot argue that this is the kind of case where appreciation of the existence of a problem requires invention. On my findings, the formulator would know that there might be a problem, would look for it by routine means, and would be able to solve it with CGK.

In relation to this point, BMS argued that the Claimants should have led evidence about the typical particle size of apixaban as an API at the priority date, to give concrete reality to their case and to show that particle size reduction would, as opposed to could, be used. I reject this. The Claimants are in the business of making apixaban now, but trying to prove the situation at the priority date would be disproportionate and impractical, if not impossible. If anything, it is BMS that could have brought evidence of real particle sizes. But I do not think it is necessary to resolve the issues.

I conclude that:

I will hear Counsel as to the form of Order if it cannot be agreed. I direct that time for seeking permission to appeal shall not run until after the hearing on the form of Order (or the making of such Order if it is agreed). I draw attention to paragraph 19.1 of the Patents Court Guide, which says that a hearing on the form of Order should take place within 28 days of hand down. Since I am giving this judgment in July that will, understandably, not be met but the argument can, I hope, take place in September and not much outside 28 days. I ask the parties please to liaise straight away to find a suitable time within that period. If there are any difficulties with this they should be communicated via my clerk promptly.