England and Wales High Court (Patents Court) Decisions

IN THE HIGH COURT OF JUSTICE
CHANCERY DIVISION
PATENTS COURT

		Royal Courts of Justice Strand, London, WC2A 2LL
		12 July 2002

Mr Justice Pumfrey :

Introduction

1. This is an action to revoke United Kingdom patent GB 2297550 (‘the patent in suit’), which stands in the name of the defendants (‘SB’). The action came on for trial at the same time as another action to revoke the patent brought by Generics (UK) Limited. The latter proceedings were compromised on the day of the hearing, but SB relied on certain material introduced by Generics (UK) Limited in their proceedings to support their contentions as to the non-obviousness of the invention. I made certain orders intended to protect the confidentiality of this material, which originated from a company not a party to the proceedings at all, but permitted SB to rely on it.

The patent in suit

2. The patent in suit is concerned with a particular polymorphic form of a known pharmaceutical, paroxetine hydrochloride (‘PHA’). This pharmaceutical is a 5-hydroxytryptamine uptake inhibitor sold in the United Kingdom under the name Seroxat, and is prescribed for depression, among other indications. It was among a number of drugs which were acquired by SB from Ferrosan, a Danish company, in the mid-1970’s. Protection by patent and Supplementary Protection Certificate for the basic pharmaceutical expired in January 1999.
3. This case is about a particular form of PHA, called the anhydrate. This means that the crystals contain PHA only, and do not contain within their lattice structure either water (in which case they would be referred to as hydrates) or any other solvent (in which case they would be called solvates). In hydrates or solvates the molecules of water or other solvent occupy specific lattice points within the crystal lattice. They can be sought to be identified by using conventional analytical techniques.
4. The patent in suit is entitled ‘Paroxetine hydrochloride anhydrate substantially free of bound organic solvent’. It was amended upon application to the Comptroller before these proceedings were started on 17 April 2001. One of the inventors, Mr Neal Ward, gave evidence before me. The case turns on difficult points of construction, and I make no apology for examining the specification in some detail. It starts with acknowledgment of EP-B-223403 (‘’403’), which is a patent granted to Beecham Group plc, SB’s predecessor in title, claiming priority from GB 8526407 (‘’407’) which is the principal item of prior art relied on. The patent in suit states (correctly) that Example 8 of ’403 describes the preparation of PHA anhydrate by crystallisation from a water containing solvent. It calls the Example 8 material ‘Form Z’. It then sets out the statement which is central to the present discussion:

‘Subsequent repetition of the preparation described in Example 8 has failed to yield any type of paroxetine hydrochloride anhydrate, and there is no clear teaching elsewhere in the document of any alternative route or modification to the process which would generate the anhydrate.’

5. I shall return to this when I come to consider the case of obviousness advanced on the basis of ’407, the disclosure of which differs somewhat from that of ’403. A further disclosure is acknowledged at page 1 line 17:

‘Paroxetine hydrochloride is also purported to be disclosed in the International Journal of Pharmaceutics 42, (1988) 135 to 143, published by Elsevier. The anhydrate is said to be produced by crystallising paroxetine hydrochloride from anhydrous propan-2-ol. Subsequent repetition of this process has resulted in a propan-2-ol solvate of paroxetine hydrochloride. That is to say that there is bound propan-2-ol in the product. This bound propan-2-ol cannot be removed by conventional drying techniques such as vacuum oven drying.’

6. The patent then states that paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol has not been described in the literature, and identifies it as what the patent is about:

‘Paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol has not been described in the literature, nor has any method been disclosed which would yield such a product as an inevitable result. A method for preparing Paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol has now been found.’

7. After a reference to four new forms of the anhydrate which have also been found, labelled Forms A, B, C and D, the specification continues:

‘Accordingly, the present invention provides paroxetine hydrochloride anhydrate substantially free of bound organic solvent.

The present invention also provides paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol with the proviso that it is other than Form Z.’

8. The phrase ‘substantially free’ is then explained:

‘Substantially free of bound organic solvent is to be interpreted to be less than the amount of propan-2-ol which would remain solvated, i.e. bound, within the crystal lattice of the product under conventional vacuum oven drying conditions.’

9. On the face of it, this is intended to widen the scope of the words ‘substantially free’. Given their natural meaning, they would suggest that any material with more than de minimis content of propan-2-ol would fall outside the claim. This meaning is displaced by the quoted sentence, whose effect is to require the phrase to be construed by reference to ‘conventional conditions’ for drying. There was a very substantial dispute as to what conditions, if any, could be described as conventional in this context. The further problem is that it defines the phrase ‘substantially free of bound organic solvent’ in terms of one organic solvent alone, propan-2-ol. So, if the preparation of paroxetine hydrochloride anhydrate takes place in (say) acetone in the absence of propan-2-ol, how is the reader of the specification to decide whether the result of his preparation in fact infringes?
10. It should also be observed that the passage refers to propan-2-ol ‘solvated, i.e. bound, within the crystal lattice’. BASF relied on this passage as emphasising the manner in which the propan-2-ol was to be bound, viz. as solvent of solvation. I consider this submission further below.
11. There follows a statement about purity. Five preferred purity ranges (greater than 50%, 60%, 70%, 80% and 90%) are identified, and the most preferred being ‘substantially pure’, itself defined as more than 95% pure. The question of water and organic solvent is then revisited:

‘It should be understood that the present invention comprising paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol may contain unbound water, that is to say water which is other than water of crystallisation.’

12. This passage is concerned, it should be observed, not with organic solvents in general but with propan-2-ol in particular. The difference between unbound water and water of crystallisation is that the water of crystallisation occupies specific points in the crystal structure. Unbound water does not. Unbound water may be located elsewhere in the crystal. What matters is that in an infra-red spectrum of the crystalline material a peak characteristic of water free from any molecular interaction is detected, and that is what is meant by the presence of unbound water.
13. The specification then returns to the amount of bound organic solvent:

‘Typically the amount of bound organic solvent on a weight for weight basis would be less than 2.0% preferably less than 1.8%, more preferably less than 1.5%, even more preferably less than 1.0%, yet more preferably less than 0.5% and most preferably less than 0.1%.’

I do not read this passage as limiting the scope of the monopoly set out by claim 1. The word ‘typically’ has the effect of giving an indication rather than setting a limit. The descriptions of the preferred crystalline forms is significant:

‘Preferred forms of paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol or substantially free of bound organic solvent include:

i) paroxetine hydrochloride anhydrate in Form A; (as hereinafter defined)

ii) paroxetine hydrochloride anhydrate in Form B (as hereinafter defined)

iii) paroxetine hydrochloride anhydrate in Form C; (as hereinafter defined)

iv) paroxetine hydrochloride anhydrate in Form D; (as hereinafter defined)’

14. The definitions of Forms A, B, C and D are lists of characterising features which follow the discussion of the characterising features of analytical data which distinguish the material ‘substantially free of solvent’ from other forms:

‘The forms of paroxetine hydrochloride anhydrate may be distinguished from each other and the material formed as a result of carrying out the procedures mentioned in [’403] and the International Journal of Pharmaceutics 42, (1988), 135 to 143 by crystalline shape, solvent analysis, or techniques such as IR, melting point, X-ray diffraction, NMR, DSC, microscopy and any other analytical techniques which differentiate one form from another.’ (page 3 lines 2–6)

15. The paper referred to is by Buxton Lynch and Rowe and was referred to as ‘Buxton’. The analytical techniques listed in this passage are conventional. The abbreviations are for Infra-red spectroscopy (IR), nuclear magnetic resonance (NMR), and differential scanning calorimetry (DSC). Results from each are shown in the specification, and certain of the claims are defined in terms of the results of these analytical methods. IR spectroscopy shines a beam of infra red light of many frequencies through the specimen, and detects which frequencies are absorbed by it. The result is expressed in wave numbers, the units of which are numbers per centimetre (cm^–1). The X-ray diffraction patterns are obtained by shining a beam of X–rays through the sample and detecting the intensity of the transmitted X–rays at angles from the axis. Relative intensity of the diffracted beams is expressed in counts (like Geiger counter counts). The angle measured is conventionally referred to as 2q. For present purposes, NMR is a technique which exploits the magnetic moment of certain atomic nuclei, including the proton. Placed in a magnetic field and exposed to microwave radiation, these nuclei will resonate at certain frequencies and absorb the incident radiation, re-radiating it in all directions. The frequencies at which absorption takes place are determined by the magnetic environment of the nuclei, which is in turn affected by (among other things) the crystalline structure in which the atoms of which they form part find themselves. The frequency of the microwaves is generally held constant, and the magnetic field varied, and the spectra shown in the Figures of the specification are all NMR spectra for the carbon-13 atom (¹³C) in the solid state. The magnetic field is expressed in parts per million (ppm).
16. Buxton discloses crystallisation of paroxetine hydrochloride from isopropanol (propan-2-ol). Buxton was originally relied on by BASF in support of their allegations of anticipation and obviousness of the claimed invention, but that attack has been abandoned. It also clearly discloses the hemihydrate as a stable form of paroxetine hydrochloride.
17. Having described the characterising features of Forms A, B, C and D, the specification proceeds to describe the method of preparing paroxetine hydrochloride substantially free of propan-2-ol (page 4 line 28). The method as described has two distinct aspects:

‘The present invention also provides a process for the preparation of paroxetine hydrochloride anhydrate substantially free of propan-2-ol which comprises crystallising paroxetine hydrochloride in either

i) an organic solvent or mixture of organic solvents which form a solvate with the paroxetine hydrochloride and which are not removable by conventional drying techniques; or

ii) an organic solvent or mixture [of] organic solvents which do or do not form a solvate but which are removable by conventional vacuum oven drying;

thereafter in the case of i) displacing the solvated solvent or solvents using a displacing agent and in the case of ii) by removing the solvent.

…

The present invention also provides a process for the preparation of paroxetine hydrochloride anhydrate substantially free of bound organic solvent which comprises displacing the solvated solvent or solvents from a paroxetine hydrochloride solvate using a displacing agent.’

18. Process variants i) and ii) are both covered by the corresponding claim (claim 10) and both need to be considered. It will be observed that variant ii) appears, on the disclosure of the specification, to present no difficulty. The paroxetine hydrochloride is prepared in solvents which can be removed by drying, and are removed by drying. The use of these solvents enables the novel crystal forms to be obtained. The second variant is directed to a different problem: how to remove solvents which cannot be removed by drying alone. What is being proposed, reduced to its essentials, is exemplified in example 1. Paroxetine hydrochloride is prepared in propan-2-ol solution under anhydrous conditions. The result is crystals of the propan-2-ol solvate. The crystals of propan-2-ol solvate are then stirred in a beaker with water. Some will dissolve, but most will not. The crystals which do not dissolve are robbed of their propan-2-ol of solvation but it is not replaced by water: the crystals become ‘paroxetine hydrochloride anhydrate substantially free of propan-2-ol.’ This, it is said, is remarkable. No-one would expect a step of washing in water to result in an anhydrate from which any solvent of solvation has been removed. The displacing agents identified are water and supercritical carbon dioxide (which is a liquid when maintained under suitable conditions of temperature and pressure).
19. The examples need to be classified. Example 1 relates to paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol. Examples 2 and 3 are directed to the propan-2-ol solvate. Example 4 shows the preparation of the pyridine solvate and its washing in water to give an anhydrate substantially free of pyridine. Example 5 is a preparation in acetic acid and 2% propan-2-ol. The resulting acetic acid solvate (13.5% w/w acetic acid) is washed, giving an anhydrate with 0.4% acetic acid. Example 6 is concerned with paroxetine hydrochloride anhydrate substantially free from acetonitrile. Each of examples 1–6 is said to result in the production of crystalline Form A.
20. Example 7 shows the production of Form B in butan-1-ol without any displacement step, the anhydrate rather than the butan-1-ol solvate being produced by a drying at high vacuum over phosphorus pentoxide.
21. Form C is produced in Example 8. This involves azeotropic distillation of a hemihydrate in toluene, and the removal of the toluene by distillation with butan-1-ol. This preparation (like Example 7) involved no separate displacement step, but results in the anhydrate. This is a method falling within claim 10(ii).
22. Examples 9, 10, 11, 13, 15, 16, 19 and 20 all demonstrate the production of paroxetine hydrochloride anhydrate substantially free from bound acetone, ethanol, chloroform, propan-1-ol, tetrahydrofuran, propan-2-ol, propan-2-ol and acetone respectively. All relate to Form A, and all involve a displacement step. Examples 9, 10, 11, 13, 15 and 20 displace the solvent by means of a water wash. In Example 16, carbon dioxide at 75°C and 2,500 psi is used. In Example 19 the displacing agent is nitrogen saturated with water at 40°C.
23. In Example 14, the anhydrate substantially free of toluene is produced in form D, after crystallisation from the hemihydrate dissolved in toluene. There is no displacement of the solvent of solvation (present at 10%w/w) which is removed by drying at 80%C in vacuum over phosphorus pentoxide.
24. It is not necessary at this stage to describe the other Examples in detail.

The claims in issue

25. BASF seek to revoke claims 1, 2, 3, 7, 10, 11, 12 and 13. It is conceded that claims 12 and 13 do not have independent validity. I set out the claims in issue below, and I underline the phrases which give rise to issues on construction.

‘1. Paroxetine hydrochloride anhydrate substantially free of bound organic solvate.

2. Paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol with the proviso that it is other than Form Z.

3. Paroxetine hydrochloride anhydrate substantially free of bound propan-2-ol in Form A, which is characterised in that

(a) it has a melting point of about 123-125°C when obtained in similar purity to the material described in Example 1

(b) and has significant IR bands at about 513, 538, 571, 592, 613, 665, 722, 761,806, 818, 839, 888, 906, 924, 947, 966, 982, 1006, 1034, 1068, 1091, 1134,1194, 1221, 1248, 1286, 1340, 1387, 1493, 1513, 1562, 1604, 3402 and 3631cm^–1

(c) and the DSC exotherm, measured at 10°C per minute shows a maximum at about 126°C using an open pan and a maximum at about 121°C using a closed pan;

(d) it also has a substantially similar X-ray diffractogram to that shown in Figure 4, including characteristic peaks at 6.6, 8.0, 11.2 and 13.1 degrees 2q

(e) and a substantially similar solid state NMR spectrum to that shown in Figure 7 including with characteristic peaks at 154.3, 149.3, 141.6 and 138.5 ppm.

7. Paroxetine hydrochloride anhydrate as defined in claim 3, which is in the form of needles.

10. A process for the preparation of paroxetine hydrochloride anhydrate substantially free of propan-2-ol which comprises crystallising paroxetine hydrochloride in either;

(i) an organic solvent or mixture of organic solvents which form a solvate with the paroxetine hydrochloride and which are not removable by conventional vacuum oven drying techniques; or

(ii) an organic solvent or mixture or organic solvents which do or do not form a solvate with the paroxetine hydrochloride but which are removable by conventional vacuum oven drying;

thereafter in the case of (i) displacing the solvated solvent or solvents using a displacing agent and in the case of (ii) by removing the solvent.

11. A process for the preparation of paroxetine hydrochloride anhydrate substantially free of bound organic solvent which comprises displacing the solvated solvent or solvents from a paroxetine hydrochloride solvate using a displacing agent.’

Construction—principles, and the skilled man

26. A patent specification is to be construed as of its date by the court who must approach it in the same way as its notional addressee would. The notional addressee of a specification is the person skilled in the art, likely to have a practical interest in the subject matter of the specification. This patent is directed to persons who are interested in the preparation of dosage forms of the known pharmaceutical paroxetine hydrochloride, and in particular in the manufacture of the active ingredient. Such a person will be a process chemist, and such persons are well known in the pharmaceutical industry. Dr Lee (for SB) and Dr Fishwick (for BASF) both attempted to identify the individual’s level of attainment. Dr Lee said that the addressee was a ‘practising process chemist assisted where necessary by analysts and working with a chemical engineer/formulator/pharmacologist.’ Dr Fishwick identified him as ‘a competent graduate organic chemist, probably working in the pharmaceutical industry, familiar with handling crystalline organic compounds and with access to analytical support staff.’
27. It is important in a case such as this not to pitch the skills of the addressee too high. Although the specification refers to a number of analytical techniques like NMR and X–ray crystallography which possess in themselves a high degree of complexity and sophistication, I think that I am here concerned with a person who can perform the analysis, or ask others to perform it, and interpret the results in an intelligent manner. He or she will be used to characterising crystal structure, and in analysing materials using techniques which are, in fact, the normal analytical techniques of a well-equipped laboratory.

The experts

28. Dr Lee gave evidence on behalf of SB. Dr Fishwick and Professor Frampton gave evidence on behalf of BASF. Dr Lee has extensive experience in the pharmaceutical industry. Dr Fishwick is an academic organic chemist, who has substantially less experience of the pharmaceutical industry, but who is knowledgeable and reasonable in his approach. He is heavily criticised by Mr Waugh QC for SB for not taking into account realities and practicalities. I shall deal with this objection in its place as it is not maintained in respect of all his evidence, but in general I consider it to be overstated. Professor Frampton had extensive knowledge both of crystallography and solid state chemistry and of solid state pharmaceutics. It is often helpful, and was in this case, to hear from an expert who can indicate with precision the ‘correct’ meaning of words and phrases in a technical context, since the scope of departure from that strict meaning in current usage and in the specification is easier to identify. Professor Frampton was obviously not representative of the skilled addressee. Mr Ward, one of the inventors, might be taken as more representative of the addressee of the specification, but his evidence was restricted to questions of fact. It no doubt injected a breath of reality into the proceedings, since it demonstrated again that real people do not necessarily act in the way that patent lawyers conclude they should have acted, even after making all possible allowances for the use of hindsight. But for reasons I explain below, his evidence, although highly informative on the history of the invention, was in the end of limited assistance on the central issue of obviousness. All the witnesses did their best to help me in coming to a conclusion, in a case which proved to be far from straightforward.

Construction—particular words

29. This case is about crystals and their forms, and I must set out my understanding of certain of the terms used at the outset. As is the case of many molecules, PHA exists in a number of forms. The term ‘form’ is ambiguous, and I can take the following unchallenged evidence from the report of Professor Frampton:

‘Crystalline State

8. The crystalline state is one in which there is a very high degree of internal order. The atoms or molecules of which the crystal is composed are arranged in a precise regular way that is repeated over and over in three dimensions forming a crystal lattice.

Crystal habit and morphology

9. Crystal habit and morphology are interchangeable terms that refer to the physical shape or form of the individual crystalline specimens, for example, the crystal habit of a specimen could be described as “needles”, “prisms” or “plates”. The term “form” in this context should not be confused with “form” as in polymorph. Crystal habit is not necessarily an indicator of different polymorphic forms, as one polymorph may exist in a range of different crystal habits. Also, different polymorphs may exist in the same crystal habit.

Polymorphism

Polymorphism occurs when a given compound of a specific molecular formula and chemical structure exists in two or more physical phases resulting from differing arrangements of the molecules in the solid state, i.e. exhibiting different crystalline lattices. These phases are often referred to as “forms”, “polymorphs” or “modifications” and can be characterised on the basis of unique analytical data. The interpretation of the terms “form” and “modification” must be made in the correct context, however, as these terms can also be used in reference to the crystal habit.

…

The consequences of polymorphism is that the different solid-forms can have different physicochemical properties such as solubility, bulk density and hardness. The pharmacokinetic profile e.g. the bioavailability of an active pharmaceutical ingredient is closely related to properties such as solubility and intrinsic dissolution rate, and is therefore dependent on the choice of polymorphic form. It is usual in the development stage of the drug discovery process to implement a polymorph screen on the active pharmaceutical ingredient both to optimise the properties based on the solid-form and to ensure that the selected solid-form is stable.

…

Solvate

14 A solvate is a single crystalline phase of a given compound of specific chemical structure in which a solvent molecule or molecules have been incorporated into the crystal lattice. In a solvate each solvent molecule will occupy a specific lattice point within the crystalline lattice, and a solvate is thus a discrete crystalline form that can be characterised as such on the basis of unique analytical data. The solvent molecule is often referred to as “bound” although this term can also be used in a different context.’

30. The other context in which the word ‘bound’ may be used is described by Professor Frampton in his discussion of ‘Bound Solvent’ and ‘Free Solvent’:

‘Bound solvent

A bound solvent molecule in a solvate is one that occupies a discrete lattice position in the crystal structure.

Free Solvent

Free solvent is that which is retained in the sample after crystallisation but prior to drying. This solvent does not occupy a discrete lattice position within the crystal structure and therefore does not make the sample a solvate. This solvent is present as agglomerated, occluded or clathrated solvent within the microcrystalline powder and exists as an impurity only, the amount of which may be determined from NMR spectroscopy performed on a solution of the sample. Such NMR spectroscopy would not, however, be capable of distinguishing such free solvent from solvent that is present as a solvate. The agglomerated, occluded or clathrated solvent can be removed by drying, although occasionally it may be hard to remove it completely using conventional vacuum oven drying techniques. In this case the remaining solvent is often confusingly described as being “bound”, i.e. in the context of its being adhesive and hard to remove from the sample.’

31. From this I take it that the word ‘bound’ can mean just difficult to remove, or it can have a strict scientific meaning. I shall deal with the alternatives below.
32. The term ‘pseudopolymorph’ was also used in the evidence. This term refers to the case in which solvates have a different structure from each other and from the unsolvated material. The solvent will be present in the solvate in a stoichiometric (fixed measurable molecule to molecule) ratio to the parent molecule. Thus one may refer to the hemihydrate, which possesses one water molecule for every two molecules of paroxetine hydrochloride in the crystal lattice.

‘Conventional vacuum oven drying conditions’

33. This phrase is nowhere enlarged on in the specification. The evidence of SKB’s expert, Dr Lee, was that ‘conventional conditions’ included 50°C ± 10°C. He said that these were the conditions which he had encountered in drying the vast majority of materials during his extensive experience in the pharmaceutical industry. Thus, 70°C or above would not be considered conventional. Dr Fishwick approached the question from a different direction (he did not have the experience of the pharmaceutical industry which Dr Lee had) preferring to regard temperatures which would be used in conventional apparatus to be found in the laboratory without damage to the pharmaceutical while seeking to remove as much solvent as possible by drying as ‘conventional’. Mr Waugh QC for SB criticised this approach strongly. He observed that 80°C is close to the melting point of anhydrous paroxetine hydrochloride. Other publications suggested a drying temperature some 30° to 50° below the melting point of the material for drying: how then, said Mr Waugh, could Dr Fishwick possibly justify heating the material to within a whisker of its melting point in order to dry it?
34. A properly drawn specification would never have left a question like this open. In fact, the enquiry is an objective one: what temperatures are generally used in the pharmaceutical industry. Dr Fishwick’s evidence is evidence of the obviousness of using conventional drying equipment to dry a material and bring it within the claim, but it does not identify the ‘conventional’ conditions. This means that the temperature employed in Example 8 is conventional, as is in all probability the temperature employed in Example 7. The latter says merely that the ‘product was dried under high vacuum in a desiccator containing phosphorus pentoxide.’ The evidence of Dr Lee was clear that the overwhelming likelihood is that if the conditions are not specified conventional conditions will be used. Example 7, therefore, is a conventional drying step.

The meaning of ‘substantially free of bound organic solvent’

35. This is a difficult question. It has caused me great difficulty and uncertainty both during and after the hearing, and I am far from confident that I have even now reached the right answer. The words are straightforward, and normally confronted with such a phrase one might interpret the word ‘substantially’ as requiring at least so low a level of bound organic solvent that it can be disregarded for all practical purposes. The further definition of the phrase, which I have quoted in paragraph 8 above, confuses the issue badly. Why is the criterion for the absence of an organic solvent stated in terms of propan-2-ol, which may not be the organic solvent in question? There is no straightforward answer to this question. Mr Waugh QC for SB suggests in effect that the draftsman obviously made a mistake. It is entirely plausible that what he meant to say is ‘the amount of organic solvent which would remain solvated … under conventional drying conditions.’ Unfortunately, this interpretation is confronted with a difficulty. As I have indicated above, there may be two classes of product which the claim is intended to cover. The first comprises those materials like the propan-2-ol solvate which have been subjected to the displacement step using water or supercritical CO₂ leaving the anhydrate. But it seems from the passage relating to the ‘preferred forms of paroxetine hydrochloride anhydrate substantially free of propan-2-ol or of bound organic solvent’ that I have quoted above in paragraph 13 that the claim is intended to cover the product of the only examples which relate to the preparation of these forms, that is, Examples 7, 8, 12, 14, 17, 18 and 21.
36. Mr Waugh submitted that these examples were directed to the making of new forms of paroxetine hydrochloride anhydrate not necessarily free of bound organic solvent. The method is reflected in claim 10 (ii). But this is an interpretation itself confronted by certain obvious difficulties. Claim 10, which I have set out above in paragraph 25, is directed to the making of paroxetine hydrochloride ‘substantially free of propan-2-ol’. Obviously a material cannot be substantially free of propan-2-ol without also being substantially free of bound propan-2-ol. But propan-2-ol is an organic solvent, and thus the material will only fall within claim 10 if it contains less than the amount of propan-2-ol ‘which would remain under conventional drying conditions.’ If one substitutes the definition of ‘substantially free of a bound organic solvent’ in this claim, the preamble becomes

‘A process for the preparation of paroxetine hydrochloride anhydrate [having] less than the amount of propan-2-ol which would remain solvated, i.e. bound, within the crystal lattice of the product under conventional vacuum oven drying conditions.’

While this approach to the construction of claim 10 is attractive, it is confronted by claim 11, which is to

‘A process for the preparation of paroxetine hydrochloride anhydrate substantially free of bound organic solvent which comprises displacing the solvated solvent or solvents from a paroxetine hydrochloride solvate using a displacing agent.’

37. It is very difficult to see how on the construction of the preamble of claim 10(i) I have proposed it covers anything not covered by claim 11. It would differ from claim 11 only in that it is limited to propan-2-ol. It is also true that those of the examples which use a drying step only, rather than a displacement step followed by drying, do not describe their product as being ‘substantially free of bound [organic solvent]’, but merely as the anhydrate. Example 7 (butan-1-ol) does not describe the solvent content. Example 8, which describes the successive removal of water (using toluene) and of toluene (using butan-1-ol) without a displacement step results in a product containing only a ‘trace’ or about 0.1% by weight, of the solvent. This looks to me to be ‘substantially free’ of the solvent. Example 14 is consistent with this comparatively straightforward approach. It gives two drying steps, one apparently conventional (dried in vacuum over phosphorus pentoxide) and one unconventional (80°C in vacuum over phosphorus pentoxide). The latter step is described as ‘desolvation’, a term reinforcing my impression that this example discloses a material ‘substantially free of bound organic solvent’. But in my view consideration of claims 4–6 puts the matter beyond doubt. These claims are to paroxetine hydrochloride anhydrate ‘substantially free of propan-2-ol’ and employ the products of examples 7, 8 and 14 as the comparison. It seems to me to be an inevitable inference that these products fall within claim 1.
38. Tables setting out the salient features of the Examples as aids to construction were produced by the parties. The examples are not unequivocal, and I think one must be thrown back on what they actually describe to ascertain whether on the specification’s own showing their products fall within claim 1 or not. That brings me back to the problem of the appropriate criterion. Mr Wyand submits that ‘less than the amount of propan-2-ol which would remain solvated…under conventional vacuum oven drying conditions’ means what it says. He points out that the specification describes two items of prior art, both of which it says disclose methods of crystallising paroxetine hydrochloride from propan-2-ol to yield a product from which the solvent could not be removed. It follows, he submits, that to use propan-2-ol content as the touchstone makes good sense: the invention is a product containing less organic solvent than the propan-2-ol-containing prior art material.
39. It seems to me that Examples 7, 8 and 14, which form the substrate of claims 4, 5 and 6, equally provide a strong pointer here. Since propan-2-ol has never played any part in the preparations they describe, they are obviously free of bound propan-2-ol. It makes no sense to me that these materials are not also to be considered as substantially free of bound organic solvent, as the absence of propan-2-ol would be irrelevant if there were an excess of some other solvent present. So I accept SB’s submissions to this extent, that the explanatory passage is to be read so as to relate to the organic solvent or solvents as appropriate, rather than merely propan-2-ol.
40. The approach to the word ‘bound’ elaborated by Mr Wyand QC was as follows. He submitted that in the light of the words ‘solvated, i.e. bound, within the crystal lattice’ the word ‘bound’ was to be given the narrower meaning identified by Professor Frampton (paragraph 30 above). He observes that ‘bound’ is used of the toluene solvate, from which all of the toluene may be removed by conventional vacuum oven drying. Therefore the criterion for ‘bound’ cannot depend on ease of removal by conventional drying. This is right.
41. He then turned to the general teaching of the patent. It is said that the forms of the anhydrate may be distinguished from each other by the battery of analytical techniques to which I have referred (page 3 lines 2–6, paragraph 14 above). For this reason, he suggests that in fact the criterion for ‘bound’ is the amount which is present in the isopropanol solvate when it converts to the anhydrate polymorph, which is on the evidence something in excess of 3%. Another way of expressing this point is to say, well, the patent discloses the existence of an isopropanol solvate which it is desired to convert to an anhydrate. When the desired transition has taken place, there may be isopropanol remaining. The desired transition takes place at about 3%w/w propan-2-ol, and this is what the claim means by ‘bound’ isopropanol.
42. I would like to be able to accept this submission. But the specification does not in my judgment permit such an approach. One can test the contention by asking whether the ‘paroxetine hydrochloride anhydrate’ of Example 14 satisfies the criterion of page 2 line 1 for a material ‘substantially free of bound organic solvent’. It is certainly free of bound propan-2-ol, a material which has never been involved in its preparation. It is also the anhydrate. It has 0.3% w/w toluene, a small quantity, and is well within the ranges of page 2 lines 25-30, which place it in the ‘yet more preferable’ class. One returns to page 2 line 33, where the patent says that the preferred forms of the anhydrate substantially free of bound propan-2-ol or substantially free of bound organic solvent include paroxetine hydrochloride anhydrate in Form D ‘as hereinafter defined’. The subsequent definition (page 4 lines 9–19) characterises the material by reference to Example 14, and the definition is circular. So this material is defined as falling within claim 1. To import Mr Wyand’s definition would mean that this material, according to which the claim for crystal form D of the invention is defined, cannot be put in the class of materials falling within claim 1 for lack of data. I would not accept this construction.
43. Overall, I think that the best answer to the conundrum is that the products of methods which fall within claim 10 fall within claim 1. My principal reason for reaching this conclusion is the opening words of claim 10 itself, taken with the two truisms: (1) that a compound cannot be substantially free of a solvent without being free of that solvent bound in the lattice and (2) there is no point in being substantially free of propan-2-ol without also being substantially free of other solvents. A subordinate reason is that this construction is consistent with the passage quoted in paragraph 13 above and is not actually inconsistent with the Examples and their titles. It does mean that it is not possible to determine whether some of the Examples, particularly 12, 17 and 18 are within claim 1, but the method of bringing them within the claim if their solvent cannot be removed by a conventional drying step is clearly disclosed. These examples are not used to characterise the crystal forms. On the other hand, Examples 7 and 14 (forms B and D) do contain sufficient data to show the product is within claim 1, and Example 8 discloses a combination of solvent and crystalline form where only conventional drying is needed to remove all but a trace of the solvent (about 0.1% w/w) and is equally within claim 1.

Anticipation

The law of anticipation

44. Section 2(1) of the Patents Act 1977 provides that an invention shall be taken to be new if it does not form part of the ‘state of the art’. By section 2(2), the state of the art is to be taken to comprise all matter (whether a product, a process, information about either, or anything else) which has at any time before the priority date of the invention been made available to the public (whether in the United Kingdom or elsewhere) by written or oral description, by use or in any other way. So far as written or oral description is concerned, the law has two aspects. First, the disclosure must be enabling: that is, it must enable the skilled man to work the invention. It must be a disclosure which in the words of Lord Westbury in Hills v Evans (1862) LJ (Ch) 457 approved by the House of Lords in Asahi Kogyo KK’s Application [1991] RPC 485 at 538 is ‘equal to that given by the subsequent patent’. Second, it must be a description of the invention, and not of some near approach. The modern law is as stated by the Court of Appeal in General Tire v Firestone [1972] RPC 457 at 484. For a claim to be anticipated by a prior disclosure, the prior disclosure must contain a clear description of, or clear instructions to do or make, something that would infringe the patentee’s claim if carried out after the grant of the patentee’s patent. If, on the other hand, the prior publication contains a direction which is capable of being carried out in a manner which would infringe the patentee’s claim, but would be at least as likely to be carried out in a way which would not do so, the patentee’s claim will not have been anticipated, although it may fail on the grounds of obviousness. A signpost, however clear, upon the road to the patentee’s invention will not suffice. The prior inventor must be clearly shown to have planted his flag at the precise destination before the patentee. This formulation is sometimes glossed as requiring that the invention be the ‘inevitable result’ of carrying out the directions of the prior publication. This is the test applied in the European Patent Office as well:

‘7.5 In the case of a prior document, the lack of novelty may be apparent from what is explicitly stated in the document itself. Alternatively, it may be implicit in the sense that, in carrying out the teaching of the prior document, the skilled person would inevitably arrive at a result falling within the terms of the claim. An objection of lack of novelty of this kind should be raised by the examiner only where there can be no reasonable doubt as to the practical effect of the prior teaching…’ (Guidelines for Examination Part C chapter IV)

GB 85-26407

45. Claims 1, 2, 3, 7, 10 and 11 are alleged to be anticipated by the disclosure of published patent application number GB 85-26407, an application made by Beecham Group plc, a predecessor of SB. This application was never published by the United Kingdom Patent Office, but was published in the files of ’403 when that application was published by the European Patent Office on 27 May 1987.
46. ’407 is concerned with the preparation of crystalline paroxetine hydrochloride. After acknowledging that paroxetine hydrochloride is old, and that it has therapeutic properties, the specification continues:

‘However for commercial use it is also important that the solid product should have good handling qualities.

We have found that amorphous paroxetine hydrochloride is a hygroscopic [water absorbing] solid of poor handling qualities.

It has now been discovered that paroxetine hydrochloride can be produced in crystalline form in a manner reproducible on a commercial scale.

Accordingly the present invention provides crystalline paroxetine hydrochloride as a novel material, in particular in pharmaceutically acceptable form.

It has been discovered that crystalline paroxetine hydrochloride can exist in at least two different pseudo-polymorphic forms,

1) a hemihydrate

2) an anhydrate

It has also been discovered that paroxetine hydrochloride can form crystalline solvates with certain solvents such as certain lower alcohols and acetone, in particular isopropyl alcohol.

Accordingly the present invention provides as novel forms of crystalline paroxetine hydrochloride:

1) paroxetine hydrochloride hemihydrate

2) paroxetine hydrochloride anhydrate

3) paroxetine hydrochloride isopropanol solvate.’

47. It is with the disclosure relating to the anhydrate that I am particularly concerned. The specification describes the physical characteristics of the hemihydrate and the anhydrate, and characteristic infra-red absorbtion spectra, X-ray powder diffractogram and DSC results are shown in the figures. At page 4 line 6 appears the following passage, speaking of the isopropanol (propan-2-ol) solvate:

‘The solvent is fairly weakly bound and may be removed by heating under vacuum. The solvate contains approximately 1 mole of isopropanol per mole.’

48. SB now say, of course, that this passage is simply wrong, and all the information in the case confirms it. It is not possible to remove the propan-2-ol merely by heating under vacuum.
49. The preparation of the anhydrate is described in more detail at page 5 line 26:

‘The crystalline anhydrate form of paroxetine hydrochloride may be prepared via the initial formation of a crystalline solvate e.g. propan-2-ol or acetone solvate, of the hydrochloride and followed by removal of the solvating solvent. The IPA [propan-2-ol] solvate may be conveniently obtained by crystallisation from propan-2-ol, ideally under anhydrous conditions, by adding gaseous or concentrated hydrochloric acid to a solution of the free base or acetate in propan-2-ol., or by crystallising or recrystallising preformed paroxetine hydrochloride from propan-2-ol solution. The solvent of solvation may be removed by drying, typically under vacuum at high temperature e.g. 60ºC, to give the hygroscopic anhydrate.

…

(page 6 lines 19-25) In practice the earlier described procedure for producing the anhydrate may result in the formation of some hemihydrate. The proportion of anhydrate to hemihydrate can be increased by drying at elevated temperatures .

…

(page 6 line 35) To obtain the anhydrate by crystallisation/recrystallisation the solvent of choice is anhydrous isopropanol.’

50. It should be noted that in this general description a number of alternatives are presented to the skilled man. In particular, a choice between the use of gaseous hydrogen chloride, which is substantially anhydrous, and hydrochloric acid, which is certainly an aqueous solution, is offered. I can turn to Example 1, upon which the discussion turned. It is headed ‘paroxetine hydrochloride as anhydrate’, giving the full chemical name. After describing the preparation of paroxetine acetate from about 350g of base, the example continues:

‘The acetate salt was dissolved in isopropanol (2.4 litres) and treated with a mixture of concentrated hydrochloric acid (75ml) and more isopropanol. After standing at 0ºC for about 16 hours, the crystals of the hydrochloride salt containing isopropanol were filtered off and dried. The salt was stirred in distilled water (0.5 litres) for about 20 minutes, filtered off and dried, giving paroxetine hydrochloride anhydrate (m.p. 118ºC).’

51. I should also refer to Example 5. This is an example of the preparation of the isopropanol solvate, and it proceeds by dissolving ‘paroxetine hydrochloride’ in propan-2-ol close to reflux, i.e. near boiling, filtering and allowing to cool to 20ºC overnight. The crystals recovered are dried at 20ºC under vacuum. The example concludes with this observation:

‘The solvate of solution was fairly weakly bound and could be removed by drying at high temperatures.’

52. BASF say that Example 1 of ’407 discloses the method of claim 10(i) of the patent in suit, and that performing it results in a material within claim 1. It is said that the washing step (stirred in distilled water (0.5 litres) for about 20 minutes) is in fact a solvent displacement step, and that the result will be precisely that described in the patent in suit. SB say that on the contrary a fair repetition of the example does not give an anhydrate substantially free of propan-2-ol, and there is not proper disclosure of an appropriate displacement step.
53. This disclosure must be read in its context in the specification. Dr Lee was inclined to suggest that it was not clear that the product containing propan-2-ol which is filtered off and dried was the propan-2-ol solvate. In paragraph 10.4 of his first report, for example, he says that ‘it is not stated whether it is a solvate or not’. The same point is made in paragraph 10.7. The passages quoted in paragraph 49 above, which are specifically concerned with production of the anhydrate, suggest that this is not the case. On a careful reading of the specification, the material to be dried prior to the washing step in Example 1 might be the solvate. On the other hand, a specific example of production of the propan-2-ol solvate is included (Example 5) and it seems slightly odd that the draftsman did not also present the material of Example 1 as the propan-2-ol solvate. There is no clear and unambiguous disclosure that the material being washed in water and dried is the propan-2-ol solvate.
54. Dr Lee’s evidence generally is that certain of the conditions are surprising (see paragraph 10.5 of his report) and he continues:

‘Although I would suspect that there may be a problem with the example 1 procedure, this is not stated clearly in the 407 application. There is a suggestion along these lines on page 6 lines 19-25 [see paragraph 49 above] …This passage would ring a warning bell in my mind, that there were two forms of paroxetine hydrochloride, and that we need to do more work to investigate the stability of each form to make sure that it could be manufactured reliably. There is an indication that this may be a problem in particular with the anhydrate.’

These considerations are not relevant to anticipation. Indeed, none of Dr Lee’s doubts and worries based on what he considered to be difficulties in carrying the experimental technique over to volume production are in any way relevant to the question of anticipation. However, Dr Lee’s further evidence, which is that the function of the water wash in Example 1 of ’407 is not disclosed, is correct as a matter of language. The skilled man would have to appreciate that he had started with the propan-2-ol solvate, washed the crystals, and come out with the anhydrate. There is no doubt that this is very surprising. The skilled man would certainly have appreciated that this is what the example appeared to be teaching: Dr Fishwick described it as curious, and in cross-examination he said his reaction to the passage was ‘Oh, what on earth is going on there?’ My firm impression was that the skilled man reading this passage would have the same reaction as Dr Fishwick. I do not think that he would appreciate that what was happening (if the experiment was an accurate account of an experiment which had taken place) was that the solvent of solvation was being removed. My view is that this surprise at finding a water washing step at all in the context of a preparation which is supposed to produce an anhydrate would provide a context in which the skilled man would be left in real doubt whether he was in fact washing the propan-2-ol solvate regardless of the passage on page 5, or merely washing the anhydrate already prepared.

55. I conclude, therefore, that there is no clear and unambiguous disclosure of a distinct displacement step for removing either bound organic solvents in general or bound propan-2-ol in particular. The question thus becomes one of inevitable result. It does not matter if the skilled man does not know what he is doing, if in following the directions in the patent he carries out the invention. In Merrell Dow Pharmaceuticals v H. N. Norton & Co. Ltd. [1996] RPC 76 at 89, Lord Hoffmann said:

‘Section 2(2) [of the Patents Act 1977] does not purport to confine the state of the art about products to knowledge of their chemical composition. It is the invention which must be new and which must therefore not be part of the state of the art. It is therefore part of the state of the art if the information which has been disclosed enables the public to know the product under a description sufficient to work the invention.

…

…If the recipe which inevitably produces the substance is part of the state of the art so is the substance as made by that recipe. CPC/Flavour Concentrates Decision T 303/86 [1989] EPOR 95 was a case about actual recipes for cooking. The application was to patent a process for making flavour concentrates from vegetable or animal substances by extracting with fat solvents under pressure in the presence of water. The claim specified certain parameters for the ratio between the vapour pressure of the water in the meat or vegetables and the vapour pressure of the free water. Opposition was based upon two cook-book recipes for pressure-frying chickens and making stews which in non-technical terms disclosed processes having the same effect. The Technical board of Appeal said (at page 98):

“It is sufficient to destroy the novelty of the claimed process that this process and the known process are identical with respect to starting material and reaction conditions since processes identical in these features must inevitably yield identical products.”

Furthermore, it did not matter that the cook did not realise that he was not only frying a chicken, but also making a “flavour concentrate” in the surplus oil. It was enough, as the Board said, that “some flavour of the fried chicken is extracted into the oil during the frying process even if this is not the desired result of that process.”’

56. The principle is that if the prior disclosure does not contain a teaching which is of equal utility to the teaching of the patent, it may nonetheless anticipate if carrying out its directions the skilled man carries out a process or makes a product falling within the claim.

The experiments

57. If it is desired to make paroxetine hydrochloride anhydrate, Example 1 of ’407 has a problem. The production of the hydrochloride in propan-2-ol solution is carried out using concentrated hydrochloric acid, which as I have indicated is aqueous. There is obviously a risk of making a hydrate or hemihydrate rather than the paroxetine hydrochloride anhydrate propan-2-ol solvate. SB say that if you carry out the experiment fairly, that is what you get. In my judgment, the experimental evidence as a whole failed to demonstrate that production of the anhydrate substantially free of bound isopropanol was an inevitable result of following these directions. The facts are as follows.
58. BASF performed six experiments. Experiment 1 was said to be ‘in accordance with example 1 of 407 as read through the eyes of a person skilled in the art, alternatively differs therefrom only by obvious variations’. The same was said to be true of Experiments 2 and 3. Experiment 4 was said to repeat the teaching of page 5 of the specification (see paragraph 49 above), the solvent used being acetone. Experiment 5 is not said to reproduce the teaching of the specification, and as reported in the Notice of Experiments results in a mixture of anhydrate substantially free of propan-2-ol (20%) and hemihydrate (80%). On repetition, it produced only hemihydrate, and since the earlier experiment is not proved I think that Mr Waugh QC is right when he says that this is the only admissible result of BASF’s Experiment 5.
59. The common feature of BASF’s experiments are variations designed to enhance the anhydrous nature of the conditions under which the procedure is conducted. The direction to use concentrated hydrochloric acid, a bench reagent, necessarily involved the addition of some water. The instruction to add with it a further unspecified quantity of isopropanol (‘more isopropanol’) was followed by the addition of a very substantial quantity of isopropanol indeed (Dr Lee, with some hyperbole, described it as a swimming-pool-full), the effect being to dilute the water present substantially. In Experiment 3, the problem of the concentrated hydrochloric acid is overcome by varying the reagents so as to add HCl gas, which can be obtained substantially anhydrous.
60. BASF Experiment 3 persuaded me that in the production of anhydrate, the problem is really the use of concentrated hydrochloric acid. I have to say that the instruction at page 5 of the specification in respect of the crystalline anhydrate, ideally to use anhydrous conditions, seems to me a clear teaching that the anhydrate is achievable, and that if a hydrate is obtained the right thing to do is to change the conditions to make them more anhydrous. But this is not anticipation. I am quite satisfied that there is no question of inevitable result in this case. If Example 1 of ’407 is repeated, the hemihydrate, not the anhydrate, will be obtained.
61. SB also relied on certain experiments which had been put in by Generics before the action between them and SB settled. These also, Mr Waugh QC said, had been carefully ‘worked up’ so as not to represent a fair repetition. The clear evidence is that the experimenter responsible for working up these experiments had found that he obtained the hemihydrate if concentrated hydrochloric acid was used. He appears to have started by using comparatively small quantities for ‘more isopropanol’. There was also evidence of experiments conducted in Germany by associates of Generics which had the same result.
62. In my judgment, the objection of anticipation on the basis of Example 1 is not made out. I should mention that I was convinced by Professor Frampton’s evidence that the actual experimental results in ’407 showed that at the time it had been possible to produce the anhydrate rather than the propan-2-ol solvate. But the rest of the evidence satisfies me that it is no longer possible to do so. I shall return to this topic below.

Anticipation by ’407: acetone

63. The passage at page 5 line 26 which I have quoted in paragraph 49 above suggests that the anhydrate can be prepared via the acetone solvate. There is a suggestion that drying may be carried out at 60ºC. BASF’s Experiment 4 is to repeat these directions. There is no displacement step. The acetone content was 1.7% and the unchallenged analytical evidence was that this produced paroxetine hydrochloride anhydrate Form A in the form of needles. There is more than a trace of acetone, and it is bound, but it is not solvate.
64. There was no real challenge to the conditions which were adopted by BASF for this experiment, nor to Dr Fishwick’s reasons to support them.
65. I think that this example anticipates claim 3. Professor Frampton identified it as the Form A material, and Dr Lee accepted that. As I understand it, the same goes for the material produced at the repeat experiments. It anticipates claim 10(ii) but not claim 11. Claim 7 is also anticipated, since it was in needle form.

Obviousness in the light of 407

66. Before turning to the objection of obviousness, I should spend a short time on the law, and I should also make certain observations about the scope of the objection in this case. The statutory test is stated in section 3 of the Patents Act 1977:

‘An invention shall be taken to involve an inventive step if it is not obvious to a person skilled in the art, having regard to any matter which forms part of the state of the art…’

67. An approach to assessing what is a pure question of fact was described by the Court of Appeal in Windsurfing International Inc v Tabur Marine(Great Britain) Ltd. [1985] RPC 59. A passage in the judgment of Oliver LJ (as he then was) is often adopted as providing a helpful approach to deciding the statutory question:

‘There are, we think, four steps which require to be taken in answering the jury question. The first is to identify the inventive concept embodied in the patent in suit. Thereafter, the court has to assume the mantle of the normally skilled but unimaginative addressee in the art at the priority date and to impute to him what was, at that date, common general knowledge in the art in question. The third step is to identify what, if any, differences exist between the matter cited as being “known or used” and the alleged invention. Finally, the court has to ask itself whether, viewed without any knowledge of the alleged invention, those differences constitute steps which would have been obvious to the skilled man or whether they require any degree of invention.’

68. This is an analysis of an attack on the ‘inventive concept’ of the patent in suit. When one is dealing with a particular disclosure, however, it can often be of assistance merely to identify the various ways that the skilled man could put that disclosure into effect exercising only his common general knowledge. Such an analysis can be helpful, provided that it is always remembered that the ultimate statutory question is whether something falling within the claim was obvious at the date. In Windsurfing, for example, the concept of the patent in suit was the ‘free sail’ concept (no stays and a freely movable mast). This was disclosed in the prior article by Darby — see [1985] RPC at page 74 line 16. But the difference lay in the requirement of the claim for a wishbone boom, which is possible as part of a Bermuda rig but not as part of Darby’s square rig. The finding was that ‘all the evidence suggests that [the skilled man] would immediately see, by application of his own general knowledge, the adoption of a Bermuda rig as an obvious way of improving the performance of the Darby vehicle.’ So the inventive concept is normally what is set out in claim 1 of the patent. Because it is the law that anything obvious falling within the claim will invalidate the patent, it is usually incorrect to attempt to generalise from the claim’s express terms, at least unless it contains arbitrary or meaningless limitations.
69. The principal evidence on which a finding of obviousness or inventiveness will be reached is that of the expert witnesses. However, Mr Waugh QC relied upon the actual history of the developments leading up to the patent in suit as indications of non-obviousness. He said that the evidence of Mr Ward demonstrated that the invention was not obvious, and he also suggested that that Mr Ward should properly be viewed in many ways as representative of the skilled man. What was not obvious to Mr Ward should not be taken to be obvious to the notional skilled person. This kind of approach introduces a subjective element into the investigation which is both unwelcome and dangerous. The Court of Appeal makes it clear in Mölnlycke v Procter & 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 past, evidential criteria may have been useful to help to elucidate the approach of the common law to the question of inventiveness. Now that there is a statutory definition, evidential criteria do not form part of the formulation of the question to be decided.

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. For instance, many people may have been industriously searching for a solution to the problem for some years without hitting upon the allegedly obvious invention. When this type of evidence is adduced, the court can quickly find itself caught up in an investigation of what was or was not obvious to certain identified individuals at certain dates during the history of the development of the product or process involved. This gives rise to complications because the state of knowledge of these individuals, though skilled, may not correspond to the statutory definition of the state of the art. A particular inventor may have been unaware of some aspect of the state of the art as defined in section 2(2), and may therefore have genuinely taken what was actually an inventive step, but nevertheless be unable to claim a patentable invention since the step was, in the terms of the statute, obvious. Further, this type of evidence invites the court to speculate whether particular individuals were of an inventive disposition, because the earlier making of the same invention by another others does not necessarily mean that at a later date the invention was obvious Yet again, evidence of the commercial success of the invention can lead into an investigation of the reasons for this success; there may be commercial reasons for its success unrelated to whether the invention was or was not obvious in the past.

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.’

70. I wish to emphasise the last paragraph of this quotation. Evidence of what actually happened can also be used merely to provide confirmation of a view reached on the primary evidence alone. The skilled person is a legal construct, like the reasonable man and their various relations. To the extent that individuals in the art followed the line that appears to have been obvious to the skilled person, or failed to appreciate matters which on the primary evidence were not obvious to the skilled person, that provides some comfort. But to use this material as part of the primary investigation is in many cases very difficult, for the reasons which Nicholls V-C (as he then was) sets out.
71. As will become apparent, I consider that it is possible to dispose of one aspect of the case of obviousness without reference to what Mr Ward and the team of which he was part actually did. But Mr Waugh QC relied upon it extensively, and for this reason I will summarise it here.

The history of the invention

72. Mr Ward’s evidence covers in seventy-five paragraphs and seventy pages of extracts from his own and others’ notebooks the history of the problems which the claimants had with paroxetine hydrochloride and the attempts which were made to solve them. I can summarise it as follows. Paroxetine hydrochloride was first produced by SB at the start of the 1980’s as a hygroscopic anhydrate. The method was worked up to a chemical pilot plant batch scale. Problems arose with the reliability of production, and work was done in 1982 to ensure reliable production of a consistent crystalline form. In 1985, the crystalline form produced in the pilot plant changed. The same change appeared at the same time in the laboratory syntheses and the material now produced turned out on analysis to be the hemihydrate. It proved impossible to induce the existing syntheses to revert to producing anhydrate. This problem is described in an entertaining article entitled ‘Disappearing Polymorphs’ by Dunitz & Bernstein exhibited by Dr Lee, in which it is attributed largely to problems of seeding with the unwanted crystalline form. It is a problem which has appeared in the cases before: see Bristol Myers (Johnson’s) Application [1972] RPC 137 at 147 line 1. SB shifted production from anhydrate to hemihydrate, but applied in 1985 for a patent (’407) covering both.
73. The method of preparation of the anhydrate failed, and all reference to the anhydrate was removed from the applications for granted patents derived from ’407. (An example is EP(UK) 0223403). Mr Ward traces the various statements in ’407 to various experimental work done at the time.
74. The product was launched on the United Kingdom market in 1991. The active ingredient in the tablets was paroxetine hydrochloride hemihydrate. In 1992 the chemical development department of SB was asked to carry out a program of experiments related to the different solid forms of paroxetine and its salts, including paroxetine hydrochloride. A search was conducted for different polymorphs by three scientists, Mr Jacewicz, Mr Ward and Mr Hanson.
75. The main feature of their work is the difficulties which they experienced in preventing formation of the hemihydrate. As Mr Ward freely acknowledges, to make a hemihydrate requires one molecule of water for every two of paroxetine hydrochloride. This is not a negligible quantity. If there is no water, there will be no hemihydrate. In performing this work, Mr Ward at least was not initially familiar with the IR spectrographic analysis of these materials. As he says, ‘…early in our investigations, I did not know how to distinguish a solvate from an anhydrate by reference to IR spectra.’ He obtained some success in producing a solvent free material using pyridine.
76. In 1993, impurities began to show up in the output of the SB factory producing the hemihydrate at Irvine in Scotland. Production was stopped, since the product no longer complied with the various approvals it had, and the US Food and Drug Administration would not longer be satisfied that the production process was under control. An urgent effort was made to overcome the problem, resulting in the production of the propan-2-ol solvate which, after drying, retained about 2.4% solvent. The washing step was suggested, and resulted in a reduction of solvent to about 0.8% on the second attempt. Mr Ward was left with the impression that in the propan-2-ol system, the presence of water would mean the hemihydrate (‘Form 1’) would be produced, but if the conditions were sufficiently anhydrous the propan-2-ol solvate would be produced. The anhydrate would not be produced without the washing step. Thus the amount of water in the process at Irvine was increased, and the propan-2-ol solvate in the product disappeared. Mr Ward returned to attempting to crystallise the anhydrate directly. His first attempt was with toluene, from which a toluene solvate was directly crystallised. This could be converted to the anhydrate by vigorous drying. His experiments with n-butanol (butan-1-ol) were initially unsuccessful as the conditions were not sufficiently anhydrous. Eventually he directly crystallised the anhydrate from n-butanol. These crystals were used to seed other solutions, with the results described in the patent. With this introduction, I can return to the case of obviousness based on ’407.
77. It must be remembered that the object of Example 1 of ’407 is to produce the anhydrate. The specification appears to distinguish this from the isopropanol solvate, and would not describe the anhydrate as such were it to be an isopropanol solvate. But there is certainly no express recognition of the fact that the isopropanol solvate may after drying ‘under conventional conditions’ contain a residue of ‘bound’ propan-2-ol.
78. The starting point is to assume that the skilled man has performed Example 1, and has obtained the hemihydrate. What would he do? The evidence (not least of BASF’s experiments) is that he would take all steps to remove the water, following the clear teaching at page 5 line 32 that the ideal conditions are anhydrous. Principally this means that he would use gaseous hydrogen chloride rather than concentrated hydrochloric acid, and BASF’s Experiment 3 shows, I think, that his efforts would be crowned with success. I am less impressed by Experiments 1 and 2. I do not think he would consider that the natural thing to do would be to increase the quantities of propan-2-ol so as to swamp the water unless lawyers were telling him that he had to use concentrated hydrochloric acid and the only freedom left to him was in the words ‘more isopropanol’. The crucial question is whether the final washing step would survive the drive to anhydrous conditions, or whether it would be retained.
79. Dr Lee was satisfied that it would not. Dr Fishwick was consistent with Dr Lee in his criticisms of the SB experiments purporting to repeat Example 1 in pointing to their deficiencies in maintaining anhydrous conditions, but he thought the skilled man would retain the washing step even though he agreed that its employment was bizarre in the production of the anhydrate.
80. Although a great deal of evidence was directed to the point, I think the answer is rather clear. Unless the skilled man were to appreciate that the particular significance in the washing step was that it was converting the propan-2-ol solvate, obtained by the hydrochlorination and crystallisation in propan-2-ol, to the anhydrate, he would obviously drop it. The specification provides him with other suggestions for the elimination of solvents in general and propan-2-ol in particular. The passages I have quoted from pages 4 and 5 of the specification (paragraphs 47 and 49 above), and the concluding words of Example 5, are clear teaching that the propan-2-ol can be removed by drying, albeit not necessarily at conventional temperatures but at high temperatures. It was not suggested that in point of fact this was true, in the sense that the propan-2-ol remaining would be in sufficiently small quantities to satisfy the claim. Thus the skilled man is taught how to remove the solvent if present, and has no need to retain the odd step of washing a material in water with a view to obtaining an anhydrate.
81. I conclude, therefore, that the invention of claims 1 and 10 of the patent in suit are not invalid in the light of ’407.

Obviousness: other solvents

82. A distinct attack is mounted against claim 10(ii) and on claim 1 to the extent that it covers the products of claim 10(ii). This is that it is perfectly obvious and routine to investigate paroxetine hydrochloride and its crystallisation characteristics with a view to finding new crystal forms. In the pharmaceutical field, particular crystal forms may have advantages from the point of view particularly of stability, and thus there is always an impulse to carry out such an investigation. A further subsidiary attack is based on the use of a range of solvents in the hydrochlorination step, using gaseous HCl.
83. The fact that particular crystal forms of active ingredients may have particular advantages in the formulation of dosage forms is common general knowledge. Dr Fishwick’s evidence in chief was that toluene, butan-1-ol, ethyl acetate and butan-2-one are ‘completely standard solvents within the repertoire commonly used by the organic chemist seeking to crystallise an organic compound.’ He points to the standard textbook on the subject, Vogel, Textbook of Practical Organic Chemistry, which plainly supports his view. Vogel describes in some detail how to conduct a solvent screen to check for suitable solvents of recrystallisation.
84. Dr Fishwick was cross-examined at considerable length on this. The thrust of the cross-examination was an attempt to show that a solvent screen would not, in fact, reveal the solutions from which solvates will crystallise that will dry to the anhydrate. This attempt was really based upon the problems that Mr Ward had had in his butan-1-ol work, particularly in maintaining anhydrous conditions. In my judgment, this cross-examination wholly failed to displace the clear evidence of Dr Fishwick on this topic. Of course, there would be problems along the way. Of course, the conditions might be insufficiently anhydrous: but this is a problem which can, and would, be solved without invention. Butan-1-ol gives the product directly without forming a solvate under anhydrous conditions. This is the start, and is sufficient in any event to invalidate claim 10(ii). But toluene also gives a solvate which does not need to be seeded and can be dried to be toluene-free.
85. Mr Waugh submitted that Dr Fishwick’s analysis plainly did not take into account the realities and practicalities of what would happen during the work that Dr Fishwick described as straightforward, and that the evidence was a classic exercise in hindsight. On the contrary, I consider that Dr Fishwick’s evidence was straightforward and based on a sound premise clearly set out in Vogel. The problems which would arise along the way, which on the evidence of what Mr Ward actually did arise from a failure to maintain anhydrous conditions, were soluble without invention (by purchasing anhydrous solvents and carefully maintaining anhydrous conditions) and the results would be obtained. Nothing in the evidence of what Mr Ward and his colleagues did suggests to me that there was any invention in this approach to obtaining the anhydrate.
86. I think that this conclusion is supported by the evidence of Dr Lee. In considering the solvent screen, he said this:

‘MR. WYAND: You look at example 7, what I suggest to you is that if you were carrying out a solvent screen, this is just the sort of thing that you would do. There is nothing in this that is unusual, out of the ordinary, unconventional. A. I would reiterate my earlier point that if I were doing a solvent screen, I would start from preformed hydrochloride rather than the practical difficulties of using the free base and the hydrogen chloride gas. A simple solvent screen - that would be my starting material.

Q. So are you saying that you regard this as unconventional because they have decided to take the free base? A. It is not unconventional. You asked me how I would approach it.

Q. I am sorry. I thought that I said that this is exactly what you would expect the skilled person .... I am sorry if you understood that to be you personally. A. We are talking around ----

Q. You might not choose to use this method, but there is nothing unconventional. A. There is nothing unconventional. It would be a less preferable approach.

Q. But it would be something that you would expect the skilled man to try. A. If his earlier approach had been unsuccessful.’

87. As there is no issue as to the obviousness of a solvent screen, this is close to an admission of obviousness of the approach taken in those examples of the patent in suit. It goes to reinforce my conclusions on the solvent screen.
88. It follows that claim 10(ii) is invalid. Claims 1, 2 and 3 are also invalid. Claim 10(i) and claim 11 are valid, and it seems to me as at present advised that all product claims must be limited to products of this process, that is, a process in which the solvent of solvation is removed by displacement.

Further matters

89. In his opening, Mr Waugh QC suggested further amendment of the specification by the addition of two new claims, 14 and 15:

‘14. Paroxetine hydrochloride anhydrate crystallised from propan-2-ol having less than 2% bound propan-2-ol characterised in that [as claim 3]

15. Paroxetine hydrochloride anhydrate crystallised from acetone having less than 1.2% bound acetone charaterised in that [as claim 3]’

90. These are stated to meet the case if SB are wrong on the construction of the claims. I will hear counsel on whether there should be further amendment and on further steps. At present it seems to me that any proposed new claims must be limited in the manner which I have indicated. Acetone is a solvent which would be considered in any solvent screen. There is some suggestion that it can be removed by drying alone, and if this is so then it will not support a new claim.