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
		3 December 2002

Mr Justice Jacob:

1. Synthon contend that SB's UK patent No. 2,336,364 is invalid. Synthon have a prior application No. WO98/56787. This was filed but not published before SB's priority date. In these circumstances s.2(3) of the Patents Act 1997 (corresponding to Art.54(3) of the EPC) comes into play. S.2(3) of the Act deems the Synthon application to have been prior published for a novelty attack. The applicable law of novelty is just the same as if there had been an actual publication - see the decision of the Court of Appeal of 31^st July this year, [2002] EWCA Civ 1162 upholding my decision of 19^th June [2002] EWHC 1172 (Pat). Obviousness is not available as a ground of attack, as it would have been if the Synthon application had actually been prior published.

Technical Background

2. Before I get to the patent or the Synthon application it is necessary to set out some basic science concerning polymorphism, crystals and crystallisation, IR spectra, and other matters. With some truncation I borrow from the primer and expert report prepared by Professor Sir Jack Baldwin FRS, the distinguished organic chemist who gave evidence for Synthon.

Crystalline and amorphic forms

3. Crystalline solids have a high degree of internal order. The atoms or molecules are arranged in a precise and regular way - a distinctive lattice. The composition is regular. Crystals generally have sharp melting points or melting ranges.

Amorphic forms, as their name implies, are shapeless, the molecules or atoms being largely randomly oriented. They generally have diffuse melting ranges.

Polymorphism.

4. Sometimes the same compound can exist in different crystalline forms - if this is so it is said to be polymorphic. The lattices of different polymorphs of the same compound differ.

Solvates and Hydrates

5. Some compounds can form crystals which contain a solvent as part of the crystalline structure. If this happens then the proportions of solvent and compound are fixed - as follows from the fact that there is a regular lattice structure. If the solvent is water then the crystal is called a hydrate.

A compound may be able to form several sorts of solvate - different kinds of crystal containing solvent.

Crystallisation from solution

6. Crystallisation of both organic and inorganic compounds may take place from a solution of the compound. If a substance is dissolved its molecules are interspersed throughout those of the solvent. Even though they may have a tendency to cling together to form an ordered lattice, that tendency is defeated by the fact that the molecules in the solution all have energy - are jiggling about to put it crudely. The warmer the solution the faster the movement - which is why normally a given hot solvent will dissolve more of a given solute than will the same solvent when cold. A solution is said to be saturated when at a given temperature no more solute will dissolve in it.
7. The processes by which crystallisation is initiated are (and were in October 1998) far from precisely understood. However, it was generally accepted that at least two processes are involved, namely formation of nuclei followed by their growth into small crystals. Nuclei formation can occur spontaneously but in many cases is encouraged by rough surfaces, small particles of heterogeneous materials, intended or otherwise. Nuclei appear to be small, ordered, aggregates of molecules, continuously forming and dissolving, which can develop in size. Once nuclei have achieved a certain size they may continue to grow to produce small crystals. These small crystals can grow and divide to form the macroscopic crystals that are obtained.
8. Crystallisation is usually effected by making a hot and saturated solution of the substance in a suitable solvent and allowing this to cool. As cooling proceeds the solution becomes supersaturated due to the lower solubility at lower temperatures. Crystallisation may begin. When crystallisation is complete the crystals can be filtered off, washed and dried.
9. Supersaturation may also be achieved by the addition of an antisolvent, which is a miscible solvent having a much lower solubility for the substance. However, this method can sometimes lead to such rapid separation of the solute from solution that the solute separates as an amorphous or oily phase (sometimes referred to as "oiling out").
10. Too rapid cooling of a supersaturated solution may result in crystals which contain entrapped solvent molecules in various amounts according to the conditions. This solvent is usually removable by vacuum drying techniques to a greater or lesser extent. Too rapid cooling may also result in crystals that contain oily phase. Thus, in general the best conditions for crystallisation arise when the rate of cooling permits the growth of pure and well-formed crystals, free of entrapped solvent, impurities and oily phase.
11. If analytical tests show that the crystals are insufficiently pure then the process of crystallisation would be repeated (recrystallisation). The purification obtained by recrystallisation arises from the exclusion of impurities from the regular packing in the crystal. This is the most powerful and convenient method of purifying organic solids. A pure crystallisable form of a pharmaceutical compound is a major advantage for its acceptance and production as a drug.
12. Crystallisation is routinely initiated using an added pre-existing "seed" crystal. This "seeding" technique is widely used in industry where reproducible batches of crystals are required. But of course the very first time someone seeks to make crystals of a substance they do not have a seed to use.
13. In the absence of a seeding crystal the initial crystallisation of a newly prepared compound is often difficult, especially if the compound is impure. The following methods are (and were in October 1998) standard techniques suitable for inducing crystallisation in the absence of a seed crystal: scratch the oil with a glass rod: add small amounts of a suitable solvent which dissolves the oil and scratch whilst cooling or warming; triturate the oil with a solvent which doesn't dissolve the oil (i.e. try to use a solvent which does not dissolve the oil to dissolve out impurities in the oil); attempt to "seed" with isomorphous crystals or crystals which are believed to be isomorphous (i.e. use "seeds" consisting of crystals of some other compound but which are believed to be of very similar crystal form to the desired crystals); evaporate by warming under a high vacuum or by placing in a vacuum dessicator with a drying agent (This technique is based on the fact that traces of solvent and water can inhibit crystallisation); and leave the oil to stand at room temperature for as long as possible.
14. It will be gathered that getting a substance to crystallise out is something of an art - part of the skills of the organic chemist. To this I must return. However I should at this stage add this: it was common ground that once a particular polymorph had been prepared in a given laboratory, it is very possible that minute traces of that polymorph will be in the general laboratory environment and that they may be enough to seed crystallisation of that same form. In the present case, for example, once a laboratory has been exposed to what I will call the SB form, that form has always been produced on subsequent occasions - and with great ease. For that reason when it came to the experiments done for this case, both sides used "virgin" laboratories.

IR Analysis

15. When an organic molecule is irradiated with infrared light having a range of frequencies, bonds within the molecule absorb radiation of a particular frequency associated with that bond. The absorbance can be detected. The curve of absorbance verses wavelength is the IR spectrum of the substance. Absorbance can run from 0 to 100% at a given frequency. The wavelength is usually expressed as "wavenumbers" - the reciprocal of actual wavelength in cm. They normally run from 4000 (i.e. 1/4000 of a cm.) to 500 (i.e. 1/500 of a cm.). IR spectrometers in 1998 were widely available. The machine provided the spectrum on graph paper either via the use of a plotter or by the use of a printer connected to a computer. They could, if so adjusted, print peak positions (in wavenumbers) directly onto the spectrum when printed or plotted out. If peak positions were not printed directly by the machine they could be measured off manually from the graph paper on which the spectrum had been printed.
16. Because molecules in a crystal are held rigidly within a lattice their vibrations are modified. So the solid state IR spectrum of a crystal is characteristic not only of the substance itself but of the crystalline form. This last point is important for this case. It is worth emphasising in different words. When you have made a crystal you can get its IR spectrum. That spectrum will be a characteristic of that particular kind of crystal. You can tell whether another crystal of the same substance is the same polymorph by comparison of the IR spectra.

Nuclear Magnetic Resonance - NMR

17. This is another analysis technique, based on hydrogen or carbon 13 - ¹H NMR or ¹³C NMR. The nucleus of each of these atoms behaves like a small magnet whose direction can be flipped by radiant energy (microwave in practice). When this happens there is absorption which can be detected. ¹H NMR works like this. In organic molecules hydrogen atoms are bound to carbon and other bonds. Chemical bonds are comprised of electrons. These affect the energy of the flipping nuclei. If you apply radiant energy you get flipping at different peak positions depending on the particular environment of hydrogen atoms within the molecule. Peak areas depend on the number of hydrogen atoms in the chemical environment concerned. In principle you can assign every peak in the spectrum to specific hydrogen atoms. Moreover the peak area (obtained by integration) enables you to determine the number of hydrogen atoms in that location.

The SB Patent

18. In the jargon of the chemists in this case "methanesulfonate" is often abbreviated to "mesylate" or even "mes" and "paroxetine" to "pot." I shall use "paroxetine mesylate".

The SB patent begins by pointing out that paroxetine is an "especially important" antidepressant and anti-Parkinson product and that it is used in therapy as its hydrochloride salt. The patent goes on to say:

"We have now surprisingly discovered a novel salt of paroxetine which may be used as an alternative to the currently marketed hydrochloride, or as an intermediate in the preparation of the hydrochloride.

According to the present invention there is provided paroxetine methanesulfonate as a novel compound.

In one aspect the novel salt of this invention is provided in non-crystalline form, which may be a solid or oil. The oil is preferably absorbed on a solid carrier, especially a carrier that is usable as a component of a pharmaceutical composition.

In another aspect the novel salt of this invention is provided in crystalline form. When the crystalline form exists as more than one polymorph, each polymorph forms another aspect of this invention."

19. So in its widest form the SB invention is just the compound itself, whether crystalline or not and whether polymorphic or not. It may be noted that the patent is entirely neutral as to whether or not paroxetine mesylate exhibits polymorphism - all that last quoted sentence is saying is that if there are polymorphs, they are all within the invention. It adds that solvates may be obtained too, again without actually saying they do exist.
20. The patent then turns to the process for making the product. It contains a general teaching thus:

"In a further aspect the present invention provides a process for the preparation of a paroxetine methanesulfonate by precipitation from a solution of a paroxetine methanesulfonate, spray drying or freeze drying a solution of a paroxetine methanesulfonate, evaporating a solution of a paroxetine methanesulfonate to a glass, or by vacuum drying of oils of a paroxetine methanesulfonate, or solidification of melts of a paroxetine methanesulfonate.

Preferably such process provides crystalline paroxetine methanesulfonate by crystallization or re-crystallization from a solution of a paroxetine methanesulfonate, and especially on a commercial scale in a reproducible manner.

Paroxetine methanesulfonate may be prepared by chemical modification of a precursor methanesulfonate salt. Suitable precursors are those which may be converted to the methanesulfonate salt by hydrogenation. For example, the N-benzyl derivative of paroxetine methanesulfonate in a suitable solvent (such as a C_1-4alkanol) may be hydrogenated using a catalyst such as palladium on charcoal to generate a solution of paroxetine methanesulfonate.

Alternatively paroxetine methanesulfonate may be prepared by treating paroxetine free base or a labile derivative thereof with methanesulfonic acid or a labile derivative thereof. For example paroxetine methanesulfonate may be prepared by contacting stoichiometric amounts of the acid and paroxetine base, alternatively an excess of the acid may be used. Preferably the base is in solution and the methanesulfonic acid is used as a solid, liquid, or as a solution, for example in water, ethers, or lower alcohols such as methanol, ethanol, and propan-2-ol, or a mixture of solvents. There is no need for a pure form of paroxetine base to be used as a starting material in the preparation of the methanesulfonate salt."

21. I quote this in full so that two things can be noticed. First the patent says that a wide variety of recipes can be used to make the compound itself. And secondly that so far as crystallisation is concerned, no particular difficulties are suggested to exist. "Precipitation" from solution is how the crystals are preferably made. Very broadly what the patent then goes on to teach is that you make the paroxetine base by any method (a number are specified) and react it with methanesulphonic acid in solution.
22. Nor does the patent attach any particular importance to the solvent. This is what it says about suitable solvents:

"The paroxetine base or labile derivative thereof may be formed by evaporation of a solvent or solvent mixture in which the base or labile derivative is solubilized. Such a solvent or solvent mixture may for example be a solvent or solvent mixture medium in which paroxetine has been formed in situ e.g. in a preceding reaction step in the medium. The paroxetine base may be produced in an organic solvent or mixture such as those discussed herein, such as toluene or a medium containing toluene, which is then evaporated to leave a residue e.g. an oil, oily or solid or semi-solid residue. The unpurified paroxetine residue may be used in the preparation of paroxetine methanesulfonate. Alternatively the residue may be resolubilized in a suitable solvent such as a medium comprising an alcohol e.g. as discussed above, suitably propan-2-ol. The solvent may be heated and optionally agitated in order to effect complete dissolution of the residue.

In addition to the above-mentioned solvents, most commonly used solvents are suitable for mobilising, e.g. dissolving or suspending, paroxetine base, for example aromatic hydrocarbon type solvents such as alkylbenzenes e.g. toluene, xylene; alcohols such as C_1-8 alkanols which may be straight or branched chain e.g. methanol, ethanol, propan-2-ol; esters such as C_1-5 alkanoate esters such as ethyl acetate; ketones e.g. di- C_1-5 alkyl ketones such as acetone and butanone; amides such as C _1-5 alkyl substituted acetamides e.g. dimethyl acetamide; heterocyclic amines e.g. pyridine; halogenated hydrocarbons such as fluoro and/or chloro C_1-10 alkanes e.g. dichloromethane; nitriles such as C_1-10 alkyl nitriles e.g. acetonitrile, and ethers e.g. di- C_1-5 alkyl ethers and cyclic ethers such as tetrahydrofuran and diethyl ether.

In particular the following solvents are suitable for mobilising paroxetine free base: toluene, alcohols such as methanol, ethanol, propan-2-ol, esters such as ethyl acetate, ketones such as acetone and butanone, halogenated hydrocarbons such as dichloromethane, nitriles for example acetonitrile, and ethers such as tetrahydrofuran and diethyl ether.

Suitably mixtures of solvents may also be used e.g. mixtures of the above mentioned solvents. The paroxetine base may be provided in solution in one solvent and then the solution diluted with another solvent, miscible with the first solvent. The second solvent may be added to a solution of the paroxetine base or alternatively the solution of paroxetine base in a first solvent may be added to the second solvent, in both cases optionally with stirring in the first solvent. The mixing of the paroxetine solution and a second solvent may occur at any convenient working temperature between e.g. - 20°C and the boiling point of the solvent, preferably between 15 to 80°C under an inert atmosphere such as nitrogen."

There is evidently nothing special about the solvent. It may be noted that ethanol is one of the solvents particularly mentioned after toluene.

23. Having got the paroxetine base in solution the next stage is reaction with methanesulphonic acid. Again the teaching is general.

"Methanesulfonic acid is commercially available. It may be used as a neat liquid, or as a solution, for example in water, ethers, or lower alcohols such as methanol, ethanol and propan-2-ol, or a mixture of solvents. More generally it may be added as a neat liquid or preferably in solution, for example in water, or a lower alcohol i.e. a C_1-5 alkanol e.g. methanol, ethanol, or propan-2-ol; esters such as C_1-5 alkanoate esters such as ethyl acetate; aromatic hydrocarbon solvents e.g. a C_1-5 alkylbenzene such as toluene; di- C_1-5 alkyl ketone such as acetone, butanone, isomethylbutyl ketone, or a mixture of such solvents. The methanesulfonic acid may also be added in the form of labile derivatives as discussed above, such as a soluble salt, for example ammonium methanesulfonate, or the methanesulfonic acid salt of an amine, for example a C_1-5 alkylamine such as ethylamine or diethylamine."

24. As to the relative proportions of base to acid, the patent teaches a wide range:

"The concentration of paroxetine base or labile derivative thereof in the paroxetine feedstock is preferably in the range 5 to 80% weight/volume e.g. 5 to 50% weight/volume, more preferably in the range 10 to 50%, particularly 10 to 30%. The concentration of methanesulfonic acid or labile derivative thereof in the acid feedstock, when added in solution, is preferably in the range 0.1 to 7 molar e.g. 0.1 to 3 molar or 0.5 to 1.5 molar, but more preferably between 1 and 5 molar. A high or low concentration of the acid may be added to a low or high concentration, respectively, of the base, preferably a concentrated solution of the acid is added to a dilute solution of the base. Suitably, depending on the solvent(s) used, the concentration of paroxetine methanesulfonate formed may be in the range 2 to 50% weight/volume, typically 5 to 30%. The concentration ranges of the reactants as defined herein are found to facilitate formation in solution and subsequent precipitation of the paroxetine methanesulfonic acid salt in crystallized form."

25. As to the reaction conditions, again there is considerable latitude:

"The reaction of methanesulfonic acid with paroxetine base is exothermic and results in a rise in temperature; typically by between 10 and 25°C, depending upon the concentration of the solution, unless controlled by cooling. Suitably the addition, in either order, is carried out above ambient conditions e.g. above 25°C such as between 30 and 80°C preferably above 30°C such as between 40 and 60°C and preferably under an inert atmosphere of nitrogen preferably with agitation e.g. stirring. Whilst temperatures above ambient suitably are used, so as to control the subsequent crystallisation process and to produce crystals having reproducible properties e.g. of uniform particle size distribution and habit, temperatures in excess of 90°C are preferably avoided since degradation occurs resulting in coloration and oil formation. Optionally seeds may be added to the paroxetine solution prior to the addition of the acid component."

26. Next there is a section on isolation of the salt from the solution. The disclosure, though detailed, does not purport to disclose anything special. On the contrary the section begins with the words:

"The salt may be isolated in solid form by conventional means from a solution thereof obtained as above"

27. The patent then goes into those techniques, beginning with production of a non-crystalline salt by known techniques. It adds that water may be removed by azeotropic distillation "to avoid the formation of hydrates or to obtain the product in anhydrous form."
28. Next it comes to production of a crystalline form:

"A crystalline salt may be prepared by various methods such as directly crystallizing the material from a solvent in which the product has limited solubility or by triturating for example with ethers such as diethyl ether or otherwise crystallizing a non-crystalline salt.

A number of solvents may be used for the crystallization process including those that are useful industrially; e.g. paroxetine methanesulfonate may be crystallized from a relatively crude feedstock such as is commonly produced during the final stage of the chemical synthesis of paroxetine. In particular solvent systems which are suitable for preparation of paroxetine methanesulfonates can also be used for recrystallization (including crystallization), for example toluene or lower alcohols followed by precipitation with ether or hexane. Alternatively, paroxetine methanesulfonate may be crystallized or recrystallized by cooling and optionally seeding a hot solution in a suitable solvent such as propan-2-ol. An improved yield of the salt is obtained by evaporation of some or all of the solvent or by crystallization at elevated temperature followed by controlled cooling, preferably in stages. Careful control of precipitation temperature and seeding may be used to improve the reproducibility of the production process and the particle size distribution and form of the product.

One method for preparing crystalline paroxetine methanesulfonate salt from solution comprises forming a supersaturated solution of the salt in a solvent and allowing the crystalline salt to precipitate from solution, for example by maintaining the solution in relatively quiescent conditions, e.g. under gentle stirring or leaving the solution to stand. Seeding of the solution is optional. By selection of a suitable solvent medium and concentration the present invention provides a process in which crystalline paroxetine methanesulfonate precipitates at temperatures above - 20°C e.g. above 0°C e.g. around ambient conditions of 10 to 25°C. Suitable solvent media for this method comprise C_1-5 alkyl benzenes such as toluene, alcohols e.g. C_1-5 alkanols such as 2-propanol, di- C_1-5 alkyl ketones such as acetone, ethers such as C_4-6 cyclic ethers such as tetrahydrofuran or mixtures thereof, and in particular mixtures of such alkyl benzenes with such alkanols or ketones e.g. toluene and 2-propanol or toluene-acetone mixtures.

Another method of preparing crystalline paroxetine methanesulfonate salt comprises forming a solution of the salt, for example as defined herein, and subsequently supersaturating the solution for example by evaporation of the solvent and/or the addition of an anti-solvent to precipitate the crystalline salt from solution. An "anti-solvent", as referred to herein, is a medium such as an organic liquid, which is miscible with a solvent for paroxetine methanesulfonate salt but in which the paroxetine methanesulfonate salt is less soluble than in the solvent. Preferably the solubility of paroxetine methanesulfonate salt in the anti-solvent is less than 1mg/ml, preferably less than 0.2 mg/ml, especially less than 0.1 mg/ml. Examples of anti-solvents include ethers, e.g. di- C_1-5 alkyl ethers and alkanes, such as C_5-10 alkanes which may be straight chain, branched chain or cyclic such as hexane. Solvent systems which are suitable for preparation of paroxetine methanesulfonate, e.g. those discussed above, e.g. with reference to the solvent systems used for the deprotection reactions discussed above, can also be used for recrystallization by precipitation with an anti-solvent."

29. Although a "preferred method" is given, this consists of no more than cooling a solution below a temperature at which it becomes saturated:

"A preferred method of preparing crystalline paroxetine methanesulfonate salt comprises cooling and optionally seeding a solution in a suitable solvent in which the paroxetine methanesulfonate salt has a greater solubility at higher temperatures than at lower temperatures so that as the solution cools the solubility at lower temperatures will be exceeded and the paroxetine methanesulfonate salt crystallises out.

Suitably the solubility of the paroxetine methanesulfonate salt at or immediately below the boiling point of the solvent is 5x or more, preferably 10x or more than that at ambient temperatures (e.g. ca. 20ºC) or lower. Suitable solvent systems include alkylbenzenes, e.g. C_1-3 alkylbenzenes such as toluene, alcohols such as C_1-5 alkanols such as methanol, ethanol, 2-propanol, and butan-1-ol, ketones such as di- C_1-5 alkyl ketones such as acetone, methyl ethyl ketone, methylisobutyl ketone, esters such as C_1-5 alkyl C_1-5 alkanoates such as methyl acetate, ethyl acetate, propyl acetate, isopropyl acetate, butyl acetate, and ethers such as methyl t-butyl ether and C_4-6 cyclic ethers such as tetrahydrofuran. Single and mixed solvent systems may be used as the solvent or co-solvent of choice.

The starting temperature of the solution containing the paroxetine methanesulfonate salt to be crystallized may vary depending upon the solubility of the reactants in the solvent system. Suitable temperatures are between minus 20°C and (+) 80°C, although temperatures between (+)10°C and (+)70°C are preferred and temperatures above (+)30°C e.g. between (+)40° and (+)60°C are most preferred. The solution is cooled to a temperature within the metastable zone in initiate crystallisation. Once crystallisation is underway, the temperature of the mixture may be reduced steadily or in stages in order to maintain a moderate degree of supersaturation and a controlled crystallisation at a high yield. The cooling rate is preferably within the range 0.1 to 5°C/minute and even more preferably is between 0.1 to 2°C per minute. The final temperature at the end of the crystallisation process is preferably around or below ambient e.g. 5 to 25°C even more preferably 10 to 20°C. Advantageously the methods provided herein do not require low temperatures i.e. less than 0°C in order to enable the crystallisation process. An improved yield and quality of the paroxetine methanesulfonate salt may be obtained by combining two or more of the aforementioned crystallisation methods. For example by evaporating some or all the solvent and/or by crystallisation at elevated temperature followed by controlled cooling, preferably in stages."

30. Nothing particularly turns, for the preferred method, on the solvent. Amongst a number of others toluene and ethanol are mentioned with no apparent difference between the two.
31. There is reference to seeding (which for present purposes does not matter) and then a little bit more about crystallisation. I omit a bit about seeding and come to this:

"In further aspects the present invention provides for the use of techniques such as insonation in the preparation of crystalline paroxetine methanesulfonate salt. Insonation and/or vigorous stirring may be used to initiate nucleation for example in addition to the use of anti-solvent(s), cooling, evaporation and/or seeding. Vigorous stirring is particularly useful when the crystallization vessel used has been used previously in the manufacture of the methanesulfonate salt."

32. Finally on the methods proposed for crystallization there is some reference to the apparatus to be used:

"The abovementioned processes for preparing the paroxetine methanesulfonate salt may be carried out in various types of reaction vessels. The crystallisation of paroxetine methanesulfonate salt suitable for use as a pharmaceutical may be carried out in a vessel such as a stirred tank reactor, which may be constructed from glass-lined or stainless steel, fitted with baffles and one or more jackets to control the temperature profile during crystallisation. Alternatively, the crystallisation may be carried out in a specially designed batch crystalliser, in which fine control of the crystallisation conditions can be maintained. Suitable batch crystallisers include draft tube baffled (DTB) crystallisers, double propeller (DP) crystallisers and fluidised bed crystallisers (Oslo cooling crystallisers). Various continuous crystallisers, such as draft tube cooling, direct contact cooling, scraped surface and turbulence crystallisers may also be employed.

Suitably crystallisation is carried out in a vessel provided with one or more high intensity ultrasonic horns, for example with titanium alloy resonant horns which enable acoustic energy to be coupled to the crystallizing medium at a frequency of 20 kHz and an amplitude of 12 microns or more, and with a device that modifies the power output according to the acoustic parameters of the load. Insonation may be intermittent, limited to part of the apparatus, or discontinued once sufficient nuclei have been generated."

33. On p.12 the SB patent indicates the powder X-ray diffraction pattern of the crystalline salt (upon which nothing turns). It then goes on to say:

"In a further aspect the present invention provides the paroxetine methanesulfonate salt in a crystalline form having an infra-red spectrum identical or substantially identical to that listed under Example 2 or Example 3 below. Suitably the crystalline paroxetine methanesulfonate has inter alia one or more of the following characteristic IR peaks: 1603, 1513, 1194, 1045, 946, 830, 776, 601, 554, and 539 ± 4 cm-1."

34. This is a forward cross-reference to two examples where a crystalline form of paroxetine mesylate was not only made (by the reaction of methanesulfonic acid and paroxetine base dissolved in toluene) but subsequently characterised by IR. The 9 IR peaks are most of those which were found. Nothing turns on the fact that more were actually found as reported in the experiments (10 in example 2 and 32 in example 3). I call the form with these IR peaks "the SB form".
35. The patent then goes on to say that the methanesulfonate salts of the invention may have a range of particle sizes and specifies the melting point (over a range). Nothing turns on this. The patent then says that the paroxetine mesylate may be obtained as a solvate.
36. There is then a lot about formulation and conditions which may be treated - upon which nothing turns. It is not suggested that any of this information is new. Then it comes to the Examples introducing them with this sentence: "The following Examples illustrate the present invention." There are 55 Examples though the last 2 are just tabletting. In all cases where paroxetine mesylate is actually made, it is done by the addition of methanesulfonic acid to paroxetine free base - an example of the classic general reaction between an acid and a base. Some make crystalline paroxetine mesylate without seeding for the crystallization stage, others involve seeding. Various solvents are used, the most common being toluene. In one example ethanol is used, the crystallization being achieved by seeding. A number of examples are of recrystallisation alone, the paroxetine mesylate having been made already by an unspecified method.
37. The quantities of starting materials of the examples are given by weight with no express indication of purity. The evidence indicates that the skilled man would not take the starting materials to be 100%. Anything from about 90% purity is consistent with the examples. That 100% purity is not essential is confirmed because some of the examples start with the making of the free base in a solvent followed by the addition of the methanesulfonic acid without any purification stage for the free base. Moreover the general teaching of the specification says there is "no need for a pure form of paroxetine base"
38. Some of the examples give IR peak lists, others do not. With the exception of acetonitrile examples (26-30 and 33-34) and a non-crystalline example, (43), they are all of the same crystalline form even though more details are given in some examples than others.
39. I now turn to the claims. They can be classified as follows:

(a) Claim 1 and those dependent on it (2-20). These are all tied to crystalline paroxetine mesylate having a particular IR spectrum - which means a particular crystalline form.

(b) Claim 21 and those dependent on it (22-30). These are all pharmaceutical composition claims, not tied to a crystalline form at all;

(c) Claim 31 and those dependent on it (31-36). These are process claims, not tied to any particular crystal form;

(d) Claim 37 A "Swiss-type" claim for use of paroxetine mesylate in the preparation of a medicament. This is not tied to any form of the compound

(e) Claims 38 - 45 - a set of independent pharmaceutical composition claims containing various amounts of paroxetine mesylate. None of these are tied to any form, crystalline or not.

40. There is no claim to paroxetine mesylate as such or even to paroxetine mesylate in crystalline form. This is at first sight surprising. After all the patentee had said his invention was "paroxetine mesylate" as a novel compound. And he had said all polymorphs were aspects of his invention (if there were more than one). But claim 1 is clearly limited to a single polymorph. It says:

"Paroxetine methanesulfonate in crystalline form having inter alia the following characteristic IR peaks: 1603, 1513, 1194, 1045, 946, 830, 776, 601, 554, and 539 ± 4 cm^-1.; and/or the following characteristic XRD peaks: 8.3, 10.5, 15.6, 16.3, 17.7, 18.2, 19.8, 20.4, 21.5, 22.0, 22.4, 23.8, 24.4, 25.0, 25.3, 25.8, 26.6, 30.0, 30.2, and 31.6 ±0.2 degrees 2 theta."

41. The reason for the narrower-than-foreshadowed claim is probably amendment following citation by the Patent Office of the Synthon application. That does not mean that more general teaching of the patent is irrelevant to the problem in this case - whether or not the SKB invention is "new" having regard to Synthon's application. It is to the latter I now turn.

The Synthon Application

42. This is entitled "4-phenylpiperidine compounds". In its most general form and claims it relates to a substantial class of these, as is common in many if not most chemical claims. But its emphasis is on paroxetine. Thus the disclosure begins with a generality:

"The present invention relates to a group of tri-substituted, 4-phenylpiperidines, to a process for preparing such compounds, to a medicament comprising such compounds, and to the use of such compounds for the manufacture of a medicament.

But the following paragraphs launch straight into paroxetine, its uses and problems with the prior art forms of paroxetine:

"The compound paroxetine … is known and has been used in medicaments for treating,amongst other ailments, depression.

Paroxetine has been used as a therapeutic agent in the form of a salt with pharmaceutically acceptable acids. The first clinical trails were conducted with the acetate salt.

A known useful salt of paroxetine is the hydrochloride. This salt is considered to be the active substance in several marketed pharmaceutical products, e.g. Paxil or Seroxat. A number of forms of paroxetine hydrochloride have been described:

¾ the anhydrous form in several crystalline modifications (…);

¾ the hydrated form – a hemihydrate (…..) and in the solvated forms.

…..

Most of these known salts of paroxetine have unsuitable physico-chemical characteristics for ensuring safe and efficient handling during production thereof and formulation into final forms, since they are unstable (acetate, maleate) and possess undesirable hygroscopicity.

Furthermore their formation by crystallization from both aqueous or non-aqueous solvents is generally low-yielded and troublesome as they usually contain an undefined and unpredicted amount of bound solvent which is difficult to remove.

The crystalline paroxetine hydrochloride hemihydrate approaches these problems, but as stated in WO 95/16448, its limited photostability causes undesired colouration during classical wet tabletting procedure.

Moreover, crystalline paroxetine hydrochloride hemihydrate exhibits only limited solubility in water.

It has been generally suggested that where the aqueous solubility is low, for example less than 3 mg/ml, the dissolution rate at in vivo administration could be rate-limiting in the absorption process. The aqueous solubility of the paroxetine hemihydrate at room temperature exceeds this threshold by a relatively small margin."

43. This sets the stage for Synthon's invention. Its first stated object is "to provide a compound with improved characteristics." The disclosure then goes on to the most general form of the compounds of the invention
[Diagram or picture not reproduced in HTML version - see original .rtf file to view diagram or picture]

- a typical (for a patent) class. The substitutents R, R¹, R², and X are each a specified range of chemical groups, the details of which do not matter. Synthon then go on to specify the advantages of their invention:

"The inventors have found that these compounds exhibit good stability and very high solubility. This yields the advantage that high concentrations of the compound are obtainable in small volumes."

44. Later on specific data for the solubility of paroxetine mesylate are given - which by common consent indeed indicate very high solubility indeed. For an obvious reason solubility is important in a pharmaceutical - it means ready and predictable bioavailability. [I notice in passing that SB's patent does not try to say what the virtues of paroxetene mesylate are, whether generally or in the case of the crystalline form having the specified IR - it is not clear why this salt is inventive over any other salt. All SB say is that they have "surprisingly discovered a novel salt which can be used as an alternative to the hydrochloride". The SB patent does not, in particular, contain any solubility data.]
45. Turning back to Synthon, the disclosure then narrows down to more "preferable" sub-classes of the general class. It continues to focus on solubility - for instance saying R² "most preferably represents a C1-C2 alkyl group in order to provide an optimum solubility." Paroxetine mesylate is within this preferred class. The focus continues:

"The compounds can have a solubility at about 20^oC of at least about 10 mg/ml water, preferably having a solubility in water of at least 100, for example 500 and most preferably of at least 1000 mg/ml water."

That is a very high solubility - about as much compound as water.

46. The disclosure then goes on to preparation, again going back to the very general. You mix a 4-phenylpiperidine compound or its salt or base with a sulfonic acid. The classes of these are specified. The mixing is "to form a solution" which is "followed by separating the compound formed from this solution." At this stage it is all perfectly general. The separation is not limited to any particular form. The teaching becomes a little more specific saying this:

"The compounds of the invention can be prepared from the free base of the 4 phenylpiperidine, having the formula II, this preferably being paroxetine, by treatment with a sulfonic acid as defined above in a suitable solvent to form a solution of the desired acid addition salt, whereafter this is precipitated out of the solution."

47. There are three points to notice here: "a suitable solvent", narrowing down to paroxetine, and "precipitated out of the solution" without specifying a crystalline form at all.
48. There is then more in general terms about preparation conditions: The disclosure indicates a wide range of conditions or methods. There is nothing critical involved. Thus, for instance this is what is said about temperature:

"The forming of a solution may preferably proceed at temperatures from about 0^oC to the boiling point of the solvent."

49. The disclosure then goes on to further "aspects of the invention:" the product of the process, provision of the compound for use as a medicament for specified conditions (a "Swiss" form claim) and use of the compound as a reagent for further synthesis. Nothing turns on any of this. Save that the specific advantage of high solubility is mentioned, nothing turns on the next parts of the disclosure until one gets to the examples. Oddly the first of these is called "Experimental" rather than "Example 1", but no matter. I must set it out:

"A seeding crystal of paroxetine methane sulfonate was made as follows:

2.7 g (8.2 mmol) of paroxetine was dissolved in 15 ml of hot ethanol.

1.0 g (10.4 mmol) of methanesulfonic acid in 15 ml of ethanol was added and the mixture was cooled to room temperature. When the mixture had reached room temperature the mixture was put in the freezer at -20^oC overnight. No crystalline compound was obtained.

The mixture was evaporated to dryness leaving an oil.

After 1 month at room temperature a waxy solid was obtained.

Part of this solid was taken apart and the rest was dissolved in 10 ml of EtOAc. The waxy crystals were added and the mixture was put in the freezer of -20^oC overnight. A white crystalline product was precipitated.

After filtration and drying in a vacuumoven 2.5g (5.9 mmol) of paroxetine methane sulfonate was obtained. Yield 72%"

50. The source of paroxetine used for this preparation is not disclosed. But for Example 1, which follows, the source is said to be "paroxetine prepared by the procedure disclosed in US 4007196" which is a patent called Ferrosan. Example 1 is to essentially the same reaction but with seeding. Nothing turns on the details of the other examples which are to other sulfonate salts and the use of paroxetine mesylate as an intermediate.
51. Next come some tables showing the analysis of the compounds made. The one that matters is that for paroxetine mesylate. A melting point of "142^O - 144^OC" is given. That is not suggested to be significantly different from the melting points given in the various examples of the SB patent, for instance example 2 of that gives 143^O - 146^OC. It is not suggested that the melting points in the two patents indicate different crystalline forms. Nor does the information about the DSC curve or the 1H-NMR and 13C-NMR indicate any difference. But the IR spectrum does I set it out:

531, 546, 777, 838, 931, 962, 1038, 1100, 1169, 1208, 1469, 1500, 1515, 1615, 2577, 2869, 2900, 3023

52. The Synthon disclosure then says this:

"The compounds of the invention are crystalline, with defined melting points, DSC curves and IR spectra. It cannot be excluded that, under different conditions of their formation and under specific conditions, they could exist also in other crystalline or polymorph modifications which may differ from those as described herein. The compounds of the invention are also generally very stable and non-hygroscopic."

53. This last point is made good by further examples comparing the hygroscopicity of some of the salts of the invention, including paroxetine mesylate, with that of the hydrochloride. Low hygroscopicity is an advantage for drug formulation – again indicating an advantage over the hydrochloride currently in use.
54. Finally there are the claims. These are to the general class of compounds in various degrees of generality. There are no claims to any particular crystalline form. The claims therefore cover any form. They cover, therefore, paroxetine mesylate as such.

A comparison between the Synthon and SB patents on their face

55. The major points of similarity and difference are thus:

(1) Synthon is concerned with a wide class of compounds whereas SB with just one;

(2) However Synthon is particularly concerned with paroxetine mesylate

(3) In their general teaching neither patent indicates any particular difficulty with either making the compounds or crystallisation from an unseeded reaction mixture. Indeed both suggest the contrary by referring to a wide range of reaction and crystalline conditions.

(4) In particular so far as general teaching goes both disclosures in effect say that the specific solvent used does not matter. Synthon says use a "suitable solvent", SB give a long list of "commonly used" solvents for paroxetine base.

(5) In the specific examples where paroxetine mesylate is made by the reaction between free base and sulfonic acid and crystallised out without seeding, Synthon use ethanol as the solvent whereas SB use toluene as well as other solvents.

(6) Ethanol is a solvent specifically mentioned by SB as suitable for "mobilising" (i.e. dissolving) paroxetine free base.

(7) The only hint of any difficulty in crystallisation is in the Synthon example - the fact that the reaction mixture took a month after evaporation to dryness to "obtain a waxy solid" would suggest to the skilled man that he may have problems in crystallisation.

(8) The reported IR spectra are different. The differences are, as is common ground, such that the skilled reader would take the two parties to have made different crystalline forms. He would not however know why there were different forms or what made one form rather than the other.

(9) Neither patent suggests that the particular crystalline form is what matters.

Is the Synthon example repeatable?

56. Synthon say that the IR spectrum reported in their application is as a result of inaccurate measurement by the experimenter. They say that actually what was got was a product with the same IR spectrum as that in the SB patent. But of course before you can get an IR spectrum you have to make crystalline material. SB say that the Synthon example "experimented"is not repeatable, or at least is not shown to be repeatable. I so hold. Although there was some fierce debate about the examples (starting material, purity and so on) I do not think the fine detail matters. Both sides did experiments. Neither side showed repeatability: it is not that crystals of differing form were produced - crystals could not be made at all.
57. As far as Synthon are concerned, experiments were conducted for the purposes of this case. Given the inevitable result branch of the law of anticipation (see below) one might have thought (a) that a team of ordinary ability might have been engaged (one is concerned with the ordinary skilled man or team, not world champions) and (b) that the team concerned would simply have been given the Synthon patent and asked to carry out its teaching to make paroxetine mesylate crystals. Synthon did neither. They asked Sir Jack Baldwin FRS, the Waynflete Professor of Chemistry at Oxford University and Dr Robert Adlington whom the Professor has described as the best practical organic chemist he has ever had working for him to conduct the experiments. They were not given the Synthon disclosure itself. The lawyers gave Dr Adlington a Protocol which said:

"This is the procedure under the "Experimental" heading in the Synthon Patent …."

58. That was just not so. The Protocol not only contains a lot of details not mentioned in the Synthon disclosure (e.g. use at least 99% pure paroxetine, evaporate to dryness using a rotary evaporator under 40^OC), but positively departs from it. The departure is potentially significant. Instead of leaving the reaction mixture for a month the team were told to observe for just 7 days. If no waxy solid had happened by then, then they were told to "use standard laboratory techniques to encourage crystallisation."
59. No explanation was ever offered for this Protocol. Nor why it was misleadingly headed. To me there is a fairly strong inference - that just giving the Synthon recipe is not enough to enable the skilled man to produce crystals and certainly not inevitably so. The trouble that the Adlington/Baldwin team then had in making crystals confirms this. I need not go into the detail. It is all in two helpful fold-out documents showing the efforts that were made. Perhaps the most revealing comment was by Prof. Baldwin. Halfway through their ultimately successful efforts he formed the theory that the problem was caused by using ethanol as a solvent. Since there was excess methanesulfonic acid he thought it possible that water was being formed by a reaction between these two components. And since the product is extremely soluble in water this might be the problem. Prof. Baldwin's evidence went like this:

A. If someone had told us or if he [Dr Adlington] had been smart enough to realise that ethanol and methanesulfonic acid give ether and water, we could have run the NMR spectrum on the following day after the (inaudible) experiment.

Q. If you had been smart enough to realise?

A. Yes. We were not smart enough to realise that.

60. What actually was done in the end was to use very fierce drying conditions and ultimately, perhaps because of that, crystals were obtained. SB dispute that water really was the problem and there was a lot of evidence about that - what could be deduced from NMR results. I do not think it matters - what is clear is that it is not shown that if the ordinary skilled man or team was just given the section of the Synthon application headed "experimental" and told to work it, crystals would result. [I would only add this, however, that SB suggest the reason that their experiment 1 works is because it used toluene – a non-polar solvent rather than ethanol, which is polar. Whether this is so or not does not matter – after all SB teach the use of ethanol as a solvent].
61. SB also did some work. Again, for the life of me, I cannot understand why the experimenters were not just given the unembellished disclosure. Some Israeli chemists were given an interpretation of it by SB's Mr Jacewicz. He says he "interpreted" the instructions to pick the most reasonable option - for instance ensuring that 100% pure ethanol was used so as to avoid the risk of water preventing crystallisation. Mr Jacewicz certainly expected the experimenters to make crystals. None of this detail matters. In the end no crystals were formed (though the attempts to do so had to be handed by the Israelis to some South Africans due to time constraints.) Synthon criticised what was done for instance by suggesting that the paroxetine was too impure. None of this matters.
62. Finally there is the question of the original Synthon "Experimental". Why it worked remains a mystery. It may have been some chance nucleating agent. It may be that it sometimes works and sometimes not for other reasons.
63. Neither side, after Synthon's initial experiment got "Experimental" to work as described. The nearest approach is by the Baldwin/Adlington team. They at least made crystals in the end, albeit following the Protocol they were given and using considerable skill on top of that. Their work does show one thing however, namely that the crystals they did produce have the IR spectra reported by SB. This is a point of separate importance to which I will return.

Does the Synthon specification enable the making of paroxetine mesylate crystals?

64. Mr Silverleaf rather accepted that he had no case on the inevitable result of performing Synthon "Experimental." What he said instead is that the Synthon specification as a whole was "enabling" in the sense that it would enable the skilled man to make crystals of paroxetine mesylate. This is an important step before one gets to the next stage of his argument. He submitted that notwithstanding the deficiencies of "Experimental", given the specification as a whole, the skilled man was enabled to make crystals. For this he relied not only on the general teaching of Synthon but also the general teaching of SB. Both indicate that the solvent is not critical and that crystallisation can be achieved using ordinary techniques. He accepted that a skilled man might well begin with Example 1, which of course involves first getting a seeding crystal from "Experimental", and use ethanol. But if that did not work he would try another solvent. Any solvent will do, as SB's patent said. Mr Silverleaf also relied upon the evidence of Mr Ward, the ex-SB employee who can fairly be described as having the abilities of the skilled man. Mr Silverleaf took him through the general teaching of Synthon pointing out its generality. He was then asked:

Q. If you were looking at this document and considering making paroxetine mesylate, you would certainly start with the examples. That is right, is it not?

A. Yes, yes, I certainly would.

Q. But you certainly would not feel bound by them, would you? You would expect to make paroxetine mesylate using a variety of other reactions and conditions?

A. Yes, I think I would.

Q. And crystallize it successfully?

A. That is perhaps another matter. I think I might be inclined to say, "Once I have actually got some sort of crystals, then I might investigate a better way of making crystalline material." In order to get your seeds, then it might be sensible to do what is in the document rather than go shooting off and doing your own thing.

Q. I think we agree you would start with the examples. You start with the examples. Let me assume that you perform "Experimental" and it does not work. What do you do next?

A. I think I would do the sort of thing that I have mentioned in my expert report. I would look at the chemistry that I was trying to carry out and say, "Well, maybe I need to make some changes." I think I have suggested the sort of changes I would make. I might well readjust the stoichiometry, the molar equivalents.

Q. And you would get a hint that that might be a good idea from looking at example 1, would you not?

A. You might and you might not. I am not necessarily convinced by that. In my experience of making salts of bases, it is sometimes actually advantageous to make the first sample with an excess of acid. This is frequently done with things like hydrochloride salts. I am not necessarily convinced that you would necessarily always go down in molarity.

Q. You might go up?

A. You might go up, yes. You might change the concentration; you might change the solvent.

Q. But these would all be straightforward changes, and you would expect, fairly quickly, to succeed, based on what is in this document?

A. I would certainly have an expectation of success, yes.

Q. Let me put this to you. Based on what is in the SB patent, you would come to exactly the same conclusion, would you not?

A. Yes. I think you are asking me a slightly different question. I think when we started on this, you said clear your mind of all the information you have and let us talk about this document.

Q. I am now asking you a different question.

A. Right.

MR. JUSTICE JACOB: I think he is now asking you to put the Synthon patent out of your mind and the SB patent and supposing you had not been the inventor and it was given to you.

A. Yes. I agree that in that position I would have the same expectation, yes.

65. What Mr Silverleaf gets from all this is that actually the general teaching of both patents is the same so far as manufacture and crystallisation is concerned. Neither patent helps on a crystallisation problem if one is encountered. They are not inventions about how to crystallise. On the other hand if such a problem is encountered, using the ordinary skilled man's techniques, the problem would be overcome within a reasonable time.
66. So, submits Mr Silverleaf, Synthon is enabling notwithstanding the deficiency of "Experimental." Putting it another way, if "Experimental" had been omitted from the specification, it would nonetheless have enabled a skilled man to make paroxetine mesylate crystals.
67. I accept those submissions. If SB's patent enables crystals to be made in all the ways it teaches, so does Synthon.

What sort of crystals are enabled?

68. The next step in Mr Silverleaf's argument is this: he submits that in fact there is only one crystalline form of paroxetine mesylate. It is a monomorph. The IR spectrum of this is as set out in the SB patent, not as set out in the Synthon patent. On a balance of probabilities I think that Mr Silverleaf is right. I turn to the detail of why I think this is so.
69. One starts from the fact that both lots of inventors actually thought there might well be both polymorphs and solvates. They both said so in their general teaching. Whether there were these different forms was, so far as the general teaching was concerned, a matter of indifference to each of them.
70. Next there is the undoubted fact that SB and the Baldwin/Adlington teams, each working in "virgin" laboratories produced the same form of crystal. They used different solvents and different degrees of purity in their starting materials. Their crystallisation techniques were, in detail different. SB's product crystallised without difficulty but Baldwin/Aldington had trouble.
71. The only "evidence" for a different form are the different IR measurements reported in the Synthon application. Given those reported measements, the onus clearly lies on Synthon to prove that they are in error - that in fact what was made by Experimental had the "SB" spectra. Have they done this? I think so on the balance of probabilities.
72. Mr Benneker, the actual experimentalist was called. I found him honest - indeed there is no suggestion otherwise. The suggestion is that his memory is fallible and that it is possible that he did indeed measure a different crystal form. To give his evidence he was of course assisted by his notebooks, computer records and documents made at the time. He was a less than meticulous worker. Thus he frequently not only recorded experiments by hand in a notebook but also, at the same time or perhaps later, made entries in his computer. The entries for the same experiment differ. He did not record times. In some cases more detail is put into the computer than the notebook, in others the position is reversed. Occasionally sentences have been copied by paste and stuck in the computer record from other experiments not wholly aptly.
73. As regards the record of "Experimental" itself, it will be recalled that the patent does not disclose the source of the paroxetine free base. Mr Beneker's notes (both handwritten and computer) show he made it from the hydrochloride, which of course was on the market at the time. Actually he was given the material by a Mr Peters and did not himself know its source. The notes contain details of how he made the base from the hemihydrate. He then went on to what is reported in the patent as "Experimental". This is the crucial experiment since it made the seeding crystals used thereafter. The experimental write-up contains more than is in the Patent. The latter omits an attempt at crystallisation following the first overnight freezer step. This was to warm the mixture up, add ether to make a precipitate, dissolve that by heating, let the mixture cool, add ethanol, freeze overnight which resulted in an oil. It is not suggested that this is a crucial step to obtain crystals, it is just that Mr Benneker left it out when telling Mr Peters what he had done.
74. Even odder is what happened about Experiment 1. In the Synthon application the paroxetine starting material is said to have been made by the procedure in the Ferrosan patent. Actually Mr Benneker used the hemihydrate as he had done before. I have no idea why the reference to Ferrosan appeared in the patent. It does not matter really (though it explains why SB's Mr Jacewicz told the Israeli experimenters to use the Ferrosan recipe to make the starting material - he assumed, not unreasonably, that the paroxetine base of "Experimental" was made in the same way as was stated in Experiment 1). The oddities do not stop there, however. Wholly inexplicable is that Mr Benneker's notebook shows a deleted passage. The deleted passage shows two crystallisations following seeding. First there was a seeding followed by crystallisation. This produced 38.8g of white solid. Then more product was obtained from the mother liquor by reducing it, cooling it and seeding. This produced a further 8.3g. The write-up of these two stages is crossed out and instead a passage appears suggesting just one crysallisation following seeding, the resulting product weighing 47.1g. The two crossed out stages have been combined into one.
75. Mr Benneker cannot remember why he crossed out the two stages. He is sure from other records (HPLC, NMR and DSC records) that he in fact carried out 2 stages. No-one can suggest any motive, sinister or otherwise, as to why there was a crossing out. Each stage had a seeding - the seed was the same - from Experimental. It is not as though the two stages could have had different crystalline forms. The whole thing remains a mystery. I do not have to resolve it. What SB seek to get out of the incident is no more than a general unreliability as to what was done - to cast doubt on whether or not in fact a different crystalline form existed.
76. Now it was the product of Example 1 (not Experimental) which was the subject of IR analysis. What Mr Benneker says he did was to use the University of Nijmegen IR spectrometer at a fairly coarse resolution. He did not know how to get the machine to assign peaks automatically and did not ask, though he could have done – perhaps with a little loss of face. Nor did he save the data electronically. He got a print-out, labelled it "pot.mes" and, by hand using a ruler measured the peaks. Now this print-out has got the "SB" peaks. What Mr Benneker thinks he must have done is to mis-measure them. He did not write them on the document but gave the peaks on a piece of paper to Mr Peters who put them into the Patent. The differences between the two patents are the result of mis-measurement - no more no less. SB say that is unlikely. Prof. Niemczyk gave evidence as to this - but of course he could only give indirect evidence.
77. SB make the point that the document itself (which has the "SB" peaks) is not related to any partticular experiment on its face. It has no date. They submit that a lot depends on Mr Benneker's memory. Partly his evidence depends on where he recalls the chart being placed in the files. So, say SB, it is just all too uncertain.
78. In the end I think the better view is that "pot.mes" is indeed the document used by Mr Benneker for the IR data in the patent.

I say that for a number of reasons:

(a) Synthon have, notwithstanding searches, not found any IR spectrum print out or result which has anything other than the "SB" spectra.

(b) Synthon have got a number of other IR spectra made from other, subsequent, batches of crystalline paroxetine mesylate. These all have the same IR spectra. Mr Benneker was indeed looking for evidence of polymorphism - but found none. This was well before anyone had any inkling of this dispute.

(c) Given the common ground as to the effect of seeding it is probable that all the later samples have the same crystalline form as the original. It was with experimental that Mr Benneker's laboratory lost its virginity – to the SB form.

(d) Mr Benneker was "quite sure" that "pot.mes" was the original chart of the IR for Experiment 1. He made two IRs and they were together, where they should be in his records.

(e) No one has ever made any other polymorph.

79. I part from this point with two observations:

(a) that it is a pity Mr Benneker was not more meticulous in his records. He himself agreed that was regrettable that was so.

(b) a scientist might well say there is not enough evidence here to "prove" that there is only one polymorph. But a judge's job is not that of a scientist. He has to come to a firm result. He works in 0's and 1's, choosing between them on a balance of probability no more.

The Law and its application here

80. The question posed by the statute is deceptively simple. It is whether the SB invention is "new" having regard to the Synthon disclosure. It is not permissible to take into account what would be obvious from that disclosure. Of course in the real world a scientist or engineer will read a document with all his common general knowledge - to him what is actually said and what he reads in as self-evident from what is said does not matter. Normally that is also the position as regards prior documents said to invalidate - it does not matter whether a document actually "anticipates" (to use the convenient English law term for lack of novelty) or makes obvious a later invention. However in the special case of s.2(3) objections the line between lack of novelty and obviousness matters - it is an artificial line drawn by the law. The problem in this case is on what side of this artificial line does the SB claim 1 stand?
81. Before I seek to answer the question in this case I must try and analyse the case law. I begin with the classic statement in General Tire ([1972] RPC 457 at p.485):

"To determine whether a patentee's claim has been anticipated by an earlier publication it is necessary to compare the earlier publication with the patentee's claim. The earlier publication must, for this purpose, be interpreted as at the date of its publication, having regard to the surrounding circumstances which then existed, and without regard to subsequent events. … If the earlier publication, so construed, discloses the same device as the device which the patentee by his claim, so construed, asserts that he has invented, the patentee's claim has been anticipated, but not otherwise.

The earlier publication and the patentee's claim must each be construed as they would be at the respective dates by a reader skilled in the art to which they relate having regard to the state of knowledge in such art at the relevant date.

If the prior inventor's publication contains a clear description of, or clear instructions to do or make, something that would have infringed the patentee's claim if carried out after the grant of the patentee's patent, the patentee's claim will have been shown to lack the necessary novelty, that is to say, it will have been anticipated. The prior inventor, however, and the patentee may have approached the same device from different starting points and may for this reason, or it may be for other reasons, have so described their devices that it cannot be immediately discerned from a reading of the language which they have respectively used that they have discovered in truth the same device; but if carrying out the directions contained in the prior inventor's publication will inevitably result in something being made or done which, if the patentee's patent were valid, would constitute an infringement of the patentee's claim, this circumstance demonstrates that the patentee's claim has in fact been anticipated.

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 ground of obviousness. To anticipate the patentee's claim the prior publication must contain clear and unmistakeable directions to what the patentee claims to have invented: Flour Oxidising v Carr ((1908) 25 RPC 428 at 457, line 34, approved in BTH v Metropolitan Vickers (1928) 45 RPC 1 at 24, line 1). 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."

82. This passage was expressly held by the Court of Appeal to apply under the Patents Act 1977 (Merrell Dow v Norton [1995] 233 at 245). The House of Lords in that case [1996] RPC 76 did not find it necessary to consider it.
83. In Inhale Therapeutic Systems v Quadrant [2001] IP&T 1229 Laddie J emphasised the strict nature of the novelty test:

[43] A claim can be anticipated in two ways. First, if the prior art describes something falling within its scope then, assuming that it is enabling, the claim is anticipated. In such a case it is not necessary to carry out experiments, or give evidence of what would have happened if the prior art was put into practice, because it already describes what it achieves. Expert evidence may be needed to explain the terminology used because the prior art may be written in technical language which is different to that used in the patent. The patentee may decide that the only way to avoid anticipation, other than by amendment of his claims, is to demonstrate that, notwithstanding its disclosure, the prior art is not enabling. He can do this by proving that what is described could not be achieved on the basis of the disclosure. Such proof would be secured by expert testimony, perhaps supported in some cases by experiments.

[44] The second way of proving anticipation is by showing that the inevitable result of carrying out what is described in the prior art would be a product or process falling within the scope of the claim. The classic statement on this issue is contained in the judgment Sachs LJ in General Tire [quoted]

[45] The second paragraph in this quotation points out the difficulty faced by someone trying to prove anticipation by this route. A disclosure which is 'capable' of being carried out in a manner which does not fall within the claim, does not anticipate - although it may be a basis for an obviousness attack".

84. So there are two ways of proving anticipation, the enabling disclosure and the inevitable result routes. To the former I must return, but before I do so it is helpful to recall the context in which the inevitable result branch was identified in General Tire. The claim under attack was to a process for compounding rubber to make tyres. In particular it was directed at the incorporation of oil into the rubber without significant mastication. Synthetic rubbers of a higher-than-previous molecular weight were thus obtained yet the rubbers were flexible enough for use in tyres. These tyres had improved wear. There were four alleged anticipations, each involving adding oil to rubber and mastication. The purposes of three of the prior art purposes were different from that of the patent: improved thermal properties, economising on use of rubber by use of extra softener, compounding with less use of power. The fourth involved some experiments which were simply outside the claimed process.
85. Thus the "inevitable result" route is really about anticipation by happenstance. The earlier proposal when carried out just happens to fall within the later claim, even though the earlier inventor had some different idea. Of course it is right that if it is an inevitable consequence of carrying the earlier man's proposal into effect the later claim is bad. Otherwise the later man will have a monopoly preventing the earlier man from carrying out his invention exactly as he proposed it should be done.
86. I turn to the other, more central, concept – enabling disclosure. This is not a rule about "inevitable result." It is about whether the two inventors have in substance reached the same invention. I think that is central to this case.
87. I begin by putting on one side the case where the attacked claim involves a variant from what was disclosed by the earlier application. Beloit v Valmet [1995] RPC 705 was such a case. I said, at p.738:

"Clearly the differences are fairly trivial. But obviousness is irrelevant. The problem is whether the differences are on the lack of novelty side of the borderline between the two conceptions. Mr. Thorley says they are not, relying on the well-known statement of principle in General Tire:

'To anticipate the patentee's claim the prior publication must contain clear and unmistakable directions to do what the patentee claims to have invented."

Mr. Kitchin, for Valmet, says they are, relying upon the equally well-known statement of Lord Westbury in Hills v. Evans (1862) 4 De G.F. & J. 288 at 301:

'The information given by the prior publication must, for the purposes of practical utility, be equal to that given by the subsequent patent."

Mr. Kitchin says the Japanese application discloses four sections. So, for practical purposes it discloses to the skilled man that he could have more if he wanted. Likewise it is clear that one can have the first inverted section after the first or second section (or indeed further sections). He is told all he needs to know by the Japanese document.

Both the General Tire and the Hills v. Evans formulations have recently been held applicable to anticipation (lack of novelty) under the 1977 Act, see Asahi per Lord Jauncey at 543. But that was in rather a different context: whether the information given had to be "enabling". Here I am really concerned with scope of claim. I think the test is whether the Japanese specific embodiment would fall within claims 12 and 13. If it would, the claims are not novel, if it would not, they are. Lord Westbury's statement, made in a context where the distinction between anticipation and obviousness did not matter, is not concerned with the case where an artificial line has to be drawn, as section 2(3) requires."

88. EPO case law is to the same effect. Thus in Nissan/Fuel Injector Valve T/167/84 [1987] 6 EPOR 344 at para. 6 the Technical Board of Appeal said:

"Neither does [the citation] which is to be taken into consideration with regard to Article 54 EPC , take away the novelty of the valve according to Claim 1. The appellant conceded that the whole content of this document including any features implicit to a person skilled in the art failed to disclose a valve which comprises completely the features mentioned in Claim 1. He is, however, of the opinion that the 'whole contents' of an older document within the meaning of Article 54(3) EPC comprise features which are equivalents to the features according to the document. In support of his view he refers to the EPC and Part C, chapter IV, 7.4, of the Guidelines.

The Board cannot agree with this point of view on the following grounds: In order to mitigate the harsh effects of the 'whole contents approach', its application is confined to novelty (cf. Article 56 EPC, second sentence). Further, in order to reduce the risk of 'self collision' it has always been considered justified for a strict approach to novelty to be adopted. For this reason, Part C, chapter IV, 7.2, of the Guidelines expressly states that when considering novelty, it is not correct to interpret the teaching of a document as embracing well-known equivalents which are not disclosed in the document; this is a matter of obviousness. This approach has been consistently followed in the practice of the European Patent Office and the appellant completely failed to satisfy the Board that it is wrong on any ground."

89. I do not think either of these cases help here – they are concerned with generalisations from an earlier specific disclosure. Take a simple example. Suppose the prior art says two items are to be "screwed together." There will not be anticipation of a claim in which the items are nailed or riveted together, however obvious it may be to substitute a screw with a nail or rivet. If the later claim were to a generalised conception ("fixed together") then there would be anticipation - but it could be cured by an amendment which excluded "screwed together."
90. More difficult is the case where it is the earlier document which has the generalised conception and the later the specific. Continuing with the example, suppose the earlier document said "fixed together". Would it anticipate a later claim to "screwed together" or indeed any other later claim which condescended to the obvious detail of how the fixing was done? There would be something odd if there were no anticipation here. In an extreme case the later claim could even specify every known method of fixing with the result that the later man could foreclose the earlier from all self-evident ways of implementing his invention. This suggests that one has got to examine the concept of novelty with care - it is not a mere mechanical application of an "inevitable result" rule.
91. I think the decision in Merrell Dow in the House of Lords helps towards the solution. They focussed on the informational nature of inventions under the modern law. Lord Hoffmann, giving the opinion in which the other members of the House concurred said this (p.89:

"Section 2(2) 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."

92. An important article by Judge Rogge equally focussed on the informational nature of an invention. Judge Rogge was for many years was the President of the Bundespatentsgericht. He argued for a rational concept of novelty in his lecture to the 8^th Symposium of European Patent Judges, 1996. It was entitled "The Concept of Novelty with Particular Regard to Conflicting Patent Applications", 28 IIC 6/1997 at p.794. I quote some important passages:

"3. A closer examination should begin by focusing on the wording of the relevant provisions. Under Art.54(2) EPC (and the relevant provisions of the national patent laws) any matter is novelty-destroying which has been made available to the public "by means of a written or oral description, by use, or in any other way". This clearly goes beyond a written or diagrammatical description; the content of the information, not the form in which it is imparted, is decisive. This allows a number of conclusions to be drawn, some of which also find expression in EPO jurisprudence.

(a) The information may to a certain degree be hidden, provided the skilled practitioner would ultimately be able to discover it without undue effort. The language used is irrelevant. It may even be a computer language. The chemical composition of a product that has been made available to the public also forms part of the state of the art if the skilled practitioner is able to analyse it. Conversely, disclosure of a process also discloses its inevitable result.

(b) Furthermore, matter will have to be viewed as also having been disclosed which the skilled practitioner is required to add from his technical knowledge as a matter of course or as being virtually essential to carrying out a given teaching.

Let us consider a simple example: A cook who finds that a soup being prepared from a recipe is lacking in flavour may well add salt, regardless of whether the recipe makes any mention either of adding any salt at all, or of adding a smaller amount than actually used. The addition of salt does not result in a new recipe. Quite the reverse: to a person skilled in the art it is so obvious as not to warrant particular mention."

"4. The practice of taking a broad view of the disclosed content of a particular item of prior art is supported by the fact that the content of an earlier, unpublished patent application is also viewed as comprised in the state of the art (Art. 54(3) EPC).

This meets the requirements of the Strasbourg Convention of 1963 (Arts. 4(3) and 6) and is intended to prevent matter from being patented that is already, or could become, the subject-matter of another patent. In that respect it does not genuinely contribute to the advance of the state of the art, and hence there is no objective justification for rewarding it with the grant of a patent. Added to which, double patenting clouds the legal position and is therefore undesirable.

It would be to ignore this legislative imperative if inventions exhibiting only the slightest of differences were to be described as "new". (Further) patents would then have to be granted for subject-matter which, although not literally anticipated, would predictably fall entirely or primarily within the scope of protection of an earlier patent, scope which should, however, remain the exclusive preserve of the owner of the prior right. Were this not so, we would have no way of effectively restricting double patenting.

It would therefore seem consistent to regard an earlier patent application as simultaneously anticipating, so as to be novelty-destroying, the entire range of subject-matter which it could conceivably cover, and in particular all possible equivalents. In fact, this view is encountered not infrequently in the relevant literature."

"6. One could argue about a number of points, and it will not always be possible to decide cases on logical grounds alone. We should not however lose sight of two things:

Examination as to novelty should not be restricted to a purely formal comparison with known prior art, but must include the actual information content which goes beyond the words used.

- Earlier applicants must be given reasonable scope to defend their inventions."

93. When it comes to the "enabling disclosure" branch Judge Rogge's two conclusions are in my view essential if one is not to have a purely formalist approach to anticipation. The earlier document must certainly be "enabling." The key question is whether the two inventors have for practical purposes disclosed the same invention and told you how to make it. This does not involve a claim comparison exercise as in the old law of prior claiming. It involves the question of whether the prior inventor has in substance disclosed what the later inventor claims. It is no good, because it is not enabling, if he discloses as the invention a particular chemical compound with a novel use but the skilled man does not know how to make it. Asahi was such a case. On the other hand if he discloses the compound and the skilled man would know how to make it, it does not matter if the earlier man does not supply the detail. It is not detail material to his invention (the compound itself).
94. Perhaps conscious of the danger that this case might turn on the enabling disclosure aspect of anticipation, Mr Waugh submitted that although it was a necessary it was not a sufficient condition for anticipation that an earlier disclosure be enabling. For, he said, a disclosure not only enables just what it specifically describes. It also enables anything which a skilled man would read in as obvious. Going back to the examples of Valmet and Nissan the earlier disclosures enabled the minor variants or mechanical equivalents but that did not make these anticipated. Enablement, submitted Mr Waugh, is a wider conception than novelty.
95. To that Mr Silverleaf responded by citing Laddie J in Evans Medical v Chiron [1998] RPC 517 (a case involving the question of whether a priority document was enabling):

"I do not accept [the petitioner's] argument that there is a double test of enablement. In the passage from his speech in Asahi quoted above, Lord Jauncey was not purporting to set such a test. He was only saying that describing a result as being achieveable was not enough to prove it to be achievable in fact. It is the latter which amounts to enablement. Nor was such a test being proposed in Schering or Beloit. As Lord Hoffmann said at page 47 in Asahi, the requirement of an enabling disclosure in a patent application is a matter of substance not form. Therefore, to this extent, I accept [the patentee's] argument. The reader will not necessarily be misled by incorrect nomenclature or other errors of description in the priority document. I think [the patentee] is right when it says that it is sufficient if the document sets out the useful properties of the technical advance and identifies a practical route which achieves it. That does not mean that the general description in the priority document is to be ignored and only the examples looked at. The priority document must be read as a whole. Its general description is no less a part of the technical content than the examples it contains."

96. Mr Waugh's submission I think only arises when the earlier disclosure is specific and the later claim more general. I think the answer to it is this: where that is the case then what is enabled may not be determinative – Laddie J was simply not concerned with that sort of case. Nor am I. What I have is a general disclosure of a wide variety of methods of making crystalline paroxetine mesylate. And the later disclosure, for all practical purposes, is an equivalent wide disclosure. So Synthon got there first. All that can be said for SB is that their disclosure is to a readier way of making the crystals, but Synthon's is clear enough.
97. Mr Waugh tried to force Synthon into the "inevitable result" branch of anticipation. He submits that there is no express disclosure of the same thing in the SB patent claim and Synthon's application. Ostensibly they are talking about different crystalline forms. So Synthon must, if they are to succeed, show that it is an inevitable result of doing what is taught in Synthon that the SB claimed crystalline form will be produced. That they cannot do. This is because Example when carried out does not lead inevitably to any crystalline form, still less that characterised by SB's IR.
98. Elegant though this argument is, I think it is inadequate. I accept the second part (about inevitable result from Example) but not the first. I think one must concentrate on the essential information in the two documents. I do not think the IR spectra of the crystals is relevant to either specification. Neither specification says so – both indicate that the invention is paroxetine mesylate as such and particularly the crystalline form. Synthon actually goes further than SB in saying why this form is useful (much greater solubility). The IR spectra merely relate to characterisation of the crystals and no more. It is rather like saying they are blue. It in no way relates to the crystals as an invention. From the point of view of use of a medicine, the IR does not matter. In this connection I bear in mind that regulatory authorities of course always insist on a particular crystalline form (and indeed formulation). But that has nothing to do with the inventive concepts of these two patents.
99. I also recognise that Synthon is directed to a substantial generality of compounds. That is a distinction without a difference. The alternatives to paroxetine mesylate (or indeed alternatives to paroxetine mesylate crystals) are just that. There is a sufficiently specific distinct disclosure of the crystals as such. SB does not contain any new relevant information about these over Synthon.
100. Mr Waugh also points out that even if there is only one crystalline form of anhydrous paroxetine mesylate, there are at least three hydrates (crystals containing solvate). So, he submitted, there is nothing inevitable about the production of the SB form from the general teaching of Synthon. Even if you make crystals you may get a hydrate. This is true but I do not think relevant. Both patents say that. They are both at the same level of generality. SB is not teaching the reader anything new in relation to hydrates.
101. In the end I conclude that this is indeed a case of anticipation. It was Synthon who first invented paroxetine mesylate, including the crystals of the SB form. And they gave enough information on how to make it. If that were not so, then SB would have a valid patent covering the very heart of the Synthon invention. That is not the purpose of s.2(3) or Art. 54(3).