Generics (UK) Ltd & Ors v H Lundbeck A/S

This is the trial of three claims for revocation of European Patent (UK) No 0,347,066 (“The Patent”) in the name of the defendant (“Lundbeck”). As a result of an earlier direction, the evidence in each claim stands as evidence in the others and the three claimants have provided a single set of expert reports. For practical purposes I can therefore treat the three claims as a single claim for revocation.

The Patent has a priority date of June 1988 and concerns an antidepressant drug called escitalopram. It is sold by Lundbeck which is a comparatively small research based pharmaceutical company located in Denmark. Lundbeck specialises in diseases of the central nervous system. Escitalopram was launched in 2002 and accounts for approximately 60% of Lundbeck’s turnover. It is currently the world’s top selling branded antidepressant in terms of volume.

The challenges to the validity of the Patent are founded upon the prior art drug called citalopram. This was first synthesised by Lundbeck in 1972 and launched as an antidepressant in Denmark in 1989. Citalopram is a racemate and so comprises (+) enantiomers and (-) enantiomers, as I shall explain. Escitalopram, on the other hand, comprises the pure (+) enantiomer. The Patent has seven claims of which claims 1, 3 and 6 are alleged by Lundbeck to have independent validity. Claim 1 is a product claim and is directed to the (+) enantiomer (and salts thereof). Claim 3 is to a pharmaceutical composition in unit dosage form containing the compound of claim 1. Claim 6 is to a method of preparing the compound of claim 1 which comprises converting the (-) enantiomer of an intermediate made during the synthesis of citalopram to the (+) enantiomer, which is isolated as such or as a salt.

The attacks on the Patent can be summarised as follows:

Lundbeck relies upon commercial success to counter the obviousness allegation. It points to the very large sales of escitalopram and to unexpected technical benefits. This is answered by the claimants who say that the commercial success is not attributable to the invention, that escitalopram is not technically superior to citalopram and, in any event, the alleged unexpected technical benefits are not described in the Patent and so cannot be relied upon.

Before describing the disclosure of the Patent, I must set out some of the technical background.

Depression and, in particular, Major Depressive Disorder (“MDD”) is one of the most common psychiatric disorders with a very wide distribution in the population. It has an early age of onset, beginning in the teenage years and twenties, and in almost all cases tends to become recurrent or chronic over time. According to the World Health Organisation, MDD is the second most common cause of disability by a chronic disorder. The underlying biochemical processes of depression are far from fully understood, however the basis of depression is thought to be a disruption of normal neural transmission. Serotonin (“5-HT”) and noradrenaline (“NA”) are neurotransmitters which allow messages to pass from nerve to nerve in the body. By June 1998, it was understood that neurotransmission by these agents was poorer in people suffering from MDD than in non-depressed people. The thinking was that if the reuptake of 5-HT and NA by nerve cells could be inhibited, this would leave more free 5-HT and NA in the synapses between nerve cells, and neurotransmission would be improved.

Antidepressant treatments alleviate the symptoms of an episode of depression rather than effecting a cure of the underlying condition. By June 1988, the tricyclic antidepressants (“TCAs”) were the most widely prescribed class of antidepressants. These were introduced in 1955 and act by inhibiting NA or 5-HT reuptake. It is thought that they lead to an increase in the level of NA or 5-HT in the brain and that this accounts for their therapeutic action. However, it was found that they are not selective and have a number of other actions at postsynaptic receptors which cause serious side effects. They therefore tend to be prescribed in low doses.

Another group of molecules developed soon after the TCAs were introduced were the monoamine oxidase inhibitors. These inhibit the enzymes involved in the metabolism of 5-HT and NA. Although effective in treating in MDD, this class of drugs also has very severe side effects.

Escitalopram belongs to another class of drugs called the selective serotonin reuptake inhibitors (“SSRIs”). These are compounds which are selective in blocking only 5-HT reuptake and have little or no NA inhibitory or other action. By 1988 there were no SSRIs on the market in the UK. Two had been launched prior to that date but withdrawn due to toxicity problems. Nevertheless a number of SSRIs were in development. Fluoxetine (sold under the brand name Prozac) was the first successful SSRI to be launched into the UK and it was closely followed by paroxetine (sold under the brand name Seroxat) and citalopram. Fluoxetine and citalopram were launched as racemates and paroxetine as a single enantiomer. Of these, citalopram was known as the most selective SSRI. Since 1988, a number of other SSRIs have been introduced, namely sertraline and, most recently, escitalopram. There is evidence of the efficacy of all SSRIs in the treatment of moderate to severe depression. Most of the side effects of SSRIs are short term and occur immediately on starting treatment. The drugs are generally well tolerated and significantly fewer patients discontinue treatment due to side effects than is the case with the TCAs.

Alternative approaches have included the development of compounds of which are highly selective NA reuptake inhibitors, such as reboxetine, and compounds that are selective inhibitors of both 5-HT and NA reuptake, such as venlafaxine and duloxetine.

The basic principles of stereochemistry necessary to understand the Patent were explained by Dr Newton, one of the experts called on behalf of the claimants. Stereochemistry is the science of structures in three dimensions. Isomers are different compounds that have the same molecular formula. If isomers differ from each other only in the way the atoms are orientated in space they are called stereoisomers. Of particular relevance to these proceedings is the stereochemistry of the carbon atom. When a carbon atom is bonded to four other atoms its bonds are directed towards the corners of a tetrahedron. Two simple molecules illustrating this stereochemistry are shown below:

Sometimes, as in the case of the illustration of the methane molecule, the direction of the bonds is depicted. By convention a carbon-hydrogen shown by a dotted line or broken wedge goes into the page and away from the viewer (as in the case of the bond marked d); and one shown by a solid wedge comes out of the plane of the paper and towards the viewer (as in the case of the bond marked c).

Molecules that are not superimposable on their mirror images are chiral. If a molecule is superimposable on its mirror image it is achiral. This is illustrated by the molecule depicted below. Each of the atoms bonded to the central carbon atom is different. It can be seen that the stereoisomer on the left cannot be superimposed on the mirror image stereoisomer shown on the right. Such stereoisomeric molecules that are not superimposable are called enantiomers and the central carbon atom to which the four different atoms are attached is the chiral centre. These molecules are sometimes said to have a “handedness”. They are designated R or S depending upon their chiral configuration according to certain rules which it is not necessary to explain for the purposes of this case. It is one of the characteristics of enantiomers that they rotate the plane of polarised light in opposite directions and so are examples of optically active substances. A molecule which rotates the plane of polarised light to the right is dextro-rotatory and this characteristic is often indicated by the symbol (+). A molecule which rotates the plane of light to the left is levo-rotatory and designated by a (-). This characteristic apart, the physical properties of enantiomers are generally the same. So, for example, they have identical melting points, boiling points and densities.

A racemate is a mixture of equal parts of enantiomers. In such a mixture the rotation caused by a molecule of one enantiomer is cancelled by the opposite rotation caused by a molecule of the other enantiomer. It is therefore optically inactive. Racemates are commonly designated by the prefix (±) or rac-. The identity of most physical properties of enantiomers means that racemates cannot be separated by ordinary methods such as fractional distillation because their boiling points are the same, or fractional crystallisation because their solubilities in a given solvent are the same. So to obtain a pure enantiomer a chemist therefore has to synthesize it using chiral starting materials or using chiral reagents or, alternatively, separate the enantiomers from a racemate by using rather special resolution techniques such as those involving diastereoisomers.

Diastereoisomers (otherwise known as diastereomers) are stereoisomers with two chiral centres and which are not related as mirror images. An illustration of two diastereoisomers is shown below.

Frequently diastereoisomers have similar chemical properties but, more importantly, they sometimes have different physical properties such as different melting points, boiling points and solubilities in a particular solvent. This has led to their employment in techniques for attempting to resolve racemic mixtures. For example, the reaction of a racemic amine with a single enantiomer of an optically active acid will give diastereoisomeric salts. Another route is to react a racemic amine with the acid chloride of the single enantiomer of an optically active acid to form a mixture of diastereoisomeric amides. In the case of racemic mixtures of alcohols, these can be converted to diastereoisomeric esters by using an optically pure acid chloride. In each case it may then be possible to separate the diastereoisomers by conventional techniques, such as fractional crystallisation, and then convert the separated diastereoisomers back to the enantiomers of interest.

I would emphasise that this is something of an empirical art. There is no guarantee in any case that separation will be possible. For example, if diastereoisomers are to be separated by fractional crystallisation there are two requirements. First, it must be possible to obtain crystals, and this is by no means certain. Second, the diastereoisomers must in fact have different degrees of solubility in a particular solvent for preferential crystallisation to occur and this may or may not be the case. I will return to consider the known methods for separating racemates in more detail later in this judgment in considering the issue of common general knowledge.

Organic compounds undergo a variety of different reactions. Those known as nucleophilic reactions are of particular relevance to these proceedings. Many covalent bonds have a polar character with the result that the atoms joined by the bond carry some positive or negative charge. This makes them susceptible to attack by another charged reagent. Electron rich sites are readily attacked by reagents which can accept electrons, called electrophiles. The commonest electrophiles are positively charged ions, or molecules containing an atom which can act as an electron acceptor. If, on the other hand, the atom subject to the attack is negatively charged then it will readily be attacked by a nucleophilic reagent called a nucleophile. The commonest nucleophiles are negative ions or molecules containing an atom which can act as an electron donor. Typical of these is the OH ^– (hydroxyl) group.

It has been known for a very long time that nucleophilic substitution reactions, as illustrated by the conversion of an alkyl halide into an alcohol under alkaline conditions, can proceed by two different mechanisms. These are called S_N1 and S_N2 reactions.

An S_N2 reaction is illustrated below. During this type of reaction the new bond between the incoming nucleophile (the hydroxyl group) and the carbon atom is formed at the same time as the original bond (the carbon-bromine bond) is breaking. A transition state is formed in which the leaving group and the incoming group are half bonded to the carbon atom that is being substituted.

The most energetically favourable approach for the hydroxyl group is along the line of centres of the carbon and bromine atoms. During the course of the reaction the carbon atom is turned “inside out”, a process known as the Walden inversion. The illustration above shows the reaction occurring with a chiral carbon atom. The result is the inversion of the stereochemistry. This reaction therefore has the capacity to turn one enantiomer into its mirror image. This was illustrated as early as 1923 in the case of an optically active alcohol. It was first activated for nucleophilic substitution by conversion into a toluene t-sulphonyl ester or tosyl ester, again as illustrated below:

It is and has for very many years been well known that the hydroxyl group of ordinary alcohols is not a good leaving group and so to make an aliphatic alcohol undergo a nucleophilic substitution it is necessary to activate it. A well understood way of achieving this is to convert it into an ester such as a sulphonyl ester. This is the first step of the sequence shown. At this point the configuration at the carbon atom remains unchanged. Inversion takes place during the course of the nucleophilic attack. In the final stage of the reaction, the ester is hydrolysed to the corresponding alcohol. The overall result of the reaction is to change the (+) enantiomer to the (-) enantiomer.

By contrast, in S_N1 reactions the bond to the leaving group is broken before the new bond is formed. This is shown below. The intermediate in this case is known as a planar carbocation and may be attacked by the nucleophile from either side. As a result, any stereochemistry in the starting material may be partially or completely lost:

With this background I can now turn to consider the Patent.

The Patent is entitled “New enantiomers and their isolation”. The specification explains that the invention relates to the two novel enantiomers of citalopram and to the use of the enantiomers as antidepressant compounds. The chemical structure of citalopram is shown below and is identified as formula I. It can be seen it has one chiral centre, namely the carbon atom marked * which forms part of the furan ring and to which a side chain bearing an amino group and a further side chain bearing a fluorophenyl group are attached.

At page 2, lines 36-41, the specification explains that the invention is also concerned with a method to resolve the racemate of citalopram into the individual isomers. It states that citalopram has been disclosed in, for example, the cited 193 patent and has proved to be an efficient antidepressant but that all work in the development of the compound has been made with the racemate. It points out that citalopram has been shown pharmacologically to be a very selective inhibitor of 5-HT reuptake but that previous attempts to crystallize diastereoisomeric salts of citalopram enantiomers have failed.

The specification continues at page 2, lines 43-48 that surprisingly it has now proved possible to resolve an intermediate racemic diol (referred to as compound II) into its enantiomers and finally, in a stereoselective way to convert these enantiomers into the corresponding citalopram enantiomers. The diol, it points out, has been disclosed in the cited 884 patent as a racemic mixture. It is shown below with the primary and tertiary alcohol groups identified:

Page 3, line 13 of the specification states that, to the surprise of the inventors, almost the entire 5-HT uptake inhibition has been found to reside in the (+) citalopram enantiomer.

The specification continues at lines 16-19 that the invention also includes a new method of synthesising the compound of formula I from the diol. Two reaction schemes are described, namely Reaction Scheme I (referred to as method (a)) and Reaction Scheme II (referred to as methods (b) and (c)).

Reaction Scheme I involves reacting the racemic diol with an enantiomerically pure acid derivative as an acid chloride, anhydride or labile ester. One particular derivative is identified, namely either the (+) or (-) forms of an agent called Mosher’s Acid (α-methoxy-α-trifluoromethylphenylacetyl chloride). This produces a mixture of two diastereomeric esters on the primary alcohol which can be separated by HPLC or fractional crystallisation. The separated diastereoisomers can then be ring closed in the presence of a strong base (such as alkoxide) in an inert organic solvent such as toluene. These conditions effect the ring closure while preserving the stereochemistry, so yielding the pure citalopram enantiomers. Scheme I is shown below:

Scheme II is described on page 4 and involves making diastereomeric salts of the diol. The specification explains that optically antipodes of tartaric acid, di-benzoyltartaric acid, di-(p-toluoyl)tartaric acid, bisnaphthylphosphoric acid, 10-camphorsulfonic acid and the like are conveniently used. The enantiomers can then be separated and converted back to the alcohol. In the second stage, ring closure is effected by making a labile ester with the primary alcohol, such as methansulfonyl acid, and stereoselective ring closure using a strong base such as triethylamine (“NEt₃”) in an inert organic solvent at low temperature. Scheme II is shown below:

The enantiomers of citalopram obtained by the method of the Patent were tested for their ability to block 5-HT reuptake in two models. The first was an in vivo test to evaluate the ability of the compounds to potentiate the effect of 5-HTP. Citalopram and each of the enantiomers were administered to a test group of mice followed by 5-HTP. A control group of mice was treated with 5-HTP alone. Those test animals in which 5-HT uptake had been inhibited exhibited a 5-HTP syndrome characterised by excitation, tremor and abduction of the hind limbs. The control animals were unaffected. The second model involved an in vitro study of the inhibition of the uptake of the 5-HT in rat brain synaptosomes. The tests show that virtually all of the biological activity resides in the (+) enantiomer.

Finally, pages 8-9 of the Patent describe the formulation of tablets, syrups and solutions containing (+) citalopram.

As I have indicated, claim 1 of the Patent is to (+) citalopram. Claim 3 is to a pharmaceutical composition comprising as its active ingredient (+) citalopram. Claim 6 is directed to a method for the preparation of (+) citalopram which comprises converting the diol in a stereoselective way to (+) citalopram, which is isolated as such or as a non toxic addition salt.

A patent specification is addressed to those likely to have a practical interest in the subject matter of the invention, and such persons are those with practical knowledge and experience of the kind of work in which the invention was intended to be used: Catnic Components v Hill and Smith Ltd [1982] RPC 183 at 242-243. Such persons are collectively known as the “person skilled in the art” or “the skilled addressee”. This skilled addressee comes to a reading of the specification with the common general knowledge of the art and he reads it on the assumption that its purpose is both to describe and demarcate an invention, that is to say a practical idea that the patentee has had for a new product or process: Kirin-Amgen Inc v Hoechst Marion Roussel [2005] UKHL 59; [2005] RPC 9 at [33]. He is unimaginative and has no inventive capacity: Technip France SA’s Patent [2004] RPC 46 at [6]-[15].

The skilled addressee is relevant to all aspects of the dispute I have to decide. The claims of the Patent must be understood as if read by him and he is the person who must be considered for the purposes of assessing obviousness and sufficiency.

It is well established that the subject matter of the specification may be such that a range of skills would generally be employed in the art to which it is directed and that such skills are not to be found in a single individual. In such a case the skilled addressee may comprise a team of persons, all with different basic skills, equipped with the common general knowledge but, at the same time, unimaginative.

In this case the parties were in agreement that the skilled addressee would be a team of persons working in the pharmaceutical industry including, primarily, medicinal chemists. Their role is to find new drug candidates and they usually start by identifying a target disease, a biological mechanism for treating the disease, for example, by selective inhibition of an enzyme and finally, identifying molecules which may be effective in that mechanism. Such medicinal chemists would characteristically be PhDs with two or three years experience or with less formal qualifications but 10-15 years of practical experience.

The parties were, however, unable to agree as to the further members of the team. The claimants contended it would include analytical chemists responsible for the verification of the structures and the determination of the purities of molecules identified by the medicinal chemists. Lundbeck contended the team would include clinicians able to identify areas of unmet need and so guide any line of research undertaken by the medicinal chemists. In my judgment, both these submissions are, in substance, correct.

I am satisfied on the evidence that an analytical chemist would be part of the skilled team and that such a person would have a chemistry degree and be skilled in basic chromatographic techniques. His role would essentially be to analyse what the medicinal chemist produced. Similarly, I am satisfied on the evidence that a clinician would generally be engaged in the outset of any project to explain to the medicinal chemists what drugs were available, whether they were adequate and whether there was room for improvement.

There was some suggestion that process research chemists and pharmaceutical chemists would also be part of the team. Process chemists are concerned with the development of methods of synthesis suitable for large scale production. Pharmaceutical chemists are concerned with the development of suitable formulations. Such persons are not normally part of a drug discovery research team and become involved once a drug has been identified as worth pursuing and has moved from the research phase into drug development. The Patent is not concerned with large scale manufacture. Nor does it describe a formulation invention. I therefore reject the suggestion that process research chemists and pharmaceutical chemists would be part of the team.

The parties each called three experts to assist me in relation to the issues arising in the fields of medicinal chemistry, analytical chemistry and clinical psychiatry. They were all subject to criticism of one form or another, some mild and some extreme. The claimants called Dr Newton, Dr Collicott and Professor Reid. Lundbeck called Professor Davies, Dr Pochapsky and Professor Montgomery.

Dr Newton was employed by Glaxo from 1971 to 1996. His areas of research included drugs for the treatment of cardiovascular, central nervous system and infectious diseases. For eight years he directed the company’s global research into respiratory diseases. He joined Glaxo as a senior research chemist and eventually became the director of the Chemical Research Division. During his time as a director of Glaxo, the company discovered and successfully launched a number of very successful pharmaceuticals including Zofran and Serevent. Dr Newton is clearly a very accomplished and knowledgeable medicinal chemist. He was well placed to assist me as to the approach of the skilled team, being part of the medicinal chemistry team at Glaxo at the priority date. The claimants described him, quite rightly in my judgment, as a lively and robust individual with an extremely thorough, knowledgeable and practical view point. Nevertheless, there is one aspect of his evidence which I think it is right to approach with some caution. He gave evidence on the issue of obviousness but was told of the solution described in the Patent before he was invited to consider the problem itself. This inevitably made it harder for him to consider the issue without the wisdom of hindsight. Nevertheless, it is right to record at this stage that I found Dr Newton’s evidence under cross-examination to be clear and cogent. In my judgment his opinions carry considerable weight.

Dr Collicott is an analytical chemist. He began his career in analytical chemistry with Glaxo in 1970. He graduated with a degree in chemistry in 1982 which he completed while working with Glaxo. As part of his studies he prepared and tested a chiral stationary phase based upon the work of Professor Pirkle, a pioneer in this field, who developed the chiral HPLC columns known as “Pirkle columns” to which I refer in more detail later in this judgment. In 1989 he was invited to speak at the International Symposium of Chiral Separations and assisted with compilation of a literature survey of chiral separations which was published by the Chromatographic Society in 1991. Since 1980, he has used non-chiral HPLC to resolve enantiomers via diastereomeric derivatives and since 1982, he has used chiral HPLC columns to resolve enantiomers on an analytical scale. Lundbeck accept, rightly in my view, that Dr Collicott endeavoured to give his evidence in a fair and helpful manner. He was clearly knowledgeable about the state of the art at the priority date and had practical experience of relevant analytical techniques at that time. I found his evidence valuable. Dr Collicot was, however, put in some difficulty by the fact that he was given a series of papers dated many years after the priority date which showed how resolution of the enantiomers of citalopram had been achieved, at least on an analytical scale. As in the case of Dr Newton, this made it harder for him to avoid using hindsight in considering what would have been obvious to the skilled addressee at the priority date.

Professor Reid is the Head of Department of Mental Health at the University of Aberdeen. He has been a member of the Royal College of Psychiatrists since 1987 and was elected to the Fellowship in 2005. He is Honorary Consultant Psychiatrist in General Adult Psychiatry for NHS Grampian. At the priority date he was a Wellcome Trust Research Fellow at the Department of Pharmacology at the University of Edinburgh. He has had considerable experience of the use of antidepressants to treat MDD. Professor Reid gave his evidence with great care and authority. It is his opinion that the evidence that escitalopram has therapeutic advantages in terms of effectiveness in clinical practice is of questionable value and that even if these advantages do exist they are so small as to be clinically irrelevant. The opinions which Professor Reid expressed were considered and measured. They are, in my judgment, entitled to a good deal of respect. Nevertheless, I also have in mind that Professor Reid has never prescribed escitalopram so he has no first hand experience of it; further, he did not deal in his report with many of the papers relied upon by Professor Montgomery as illustrating its efficacy. I think it is fair to say he approached the issues in this case from a position of considerable scepticism.

Professor Stephen Davies is the Chairman of Chemistry at the University of Oxford, and the Waynflete Professor of Chemistry. He is an outstanding organic chemist. He has published widely and received a large number of prestigious awards and prizes. His research interests since 1973 have been generally in the area of the stereochemical aspects of organic and organometallic chemistry, and the preparation of enantiomerically pure (that is to say, homochiral) materials. His particular interests are asymmetric and stereoselective syntheses of homochiral compounds. After completing his D.Sc. degree in Paris in 1980 he returned to Oxford as an academic. He then began acting as a consultant to pharmaceutical companies such as ICI Pharmaceuticals (now AstraZeneca), Roussel (now Sanofi Aventis) and Fisons, who consulted him on a wide range of problems. In addition, many companies funded studentships with him which meant frequent visits and interactions with their scientists. His particular expertise through the 1980s was in all aspects of the control of stereochemistry. In the early 1990s he perceived that there would be an increasing need in the pharmaceutical industry for a centre of excellence in the area of the preparation of homochiral compounds. Along with others he therefore founded Oxford Asymmetry Ltd in 1992. The object of this company was to provide pharmaceutical companies with homochiral compounds of interest on any desired scale. They achieved this by using single enantiomer starting materials, asymmetric synthesis and chiral separations.

The claimants do not dispute that Professor Davies is an extremely eminent academic organic chemist. They make no criticism of his chemical knowledge. On the contrary, they say that his knowledge is so advanced he had great difficulty in putting himself in the place of the notional skilled addressee and that I should treat his evidence with caution as a result. I do not think that there is anything in this criticism. In the course of his cross-examination, Professor Davies was at pains to point out the extent to which the various matters to which he was referring would have been taught at undergraduate level. I think he was fully conscious of the characteristics of the notional skilled addressee that I have to consider. Two more serious criticisms were advanced. First, it was said that he was unable to keep himself from stepping into the realm of the advocate, as illustrated by his steadfast refusal to accept that by 1988 there was any serious motivation by pharmaceutical companies to resolve chiral drugs into their enantiomers. This is a significant criticism and is one that I address in detail in considering the issues of common general knowledge and obviousness later in this judgment. The other criticism concerned the treatment by Professor Davies of a paper by Cannone et al (J. Org. Chem., 1980, Vol 45, pages 1828-1835). Professor Davies referred to this paper, amongst a number of others, to support the proposition that the particular reaction mechanism disclosed in the Patent was not discussed in the literature, when in fact it is. This, it was said, illustrated his evidence had not been prepared with the full care the court was entitled to expect. I do not accept this criticism. The paper was one of a number found on a search and I do not think that Professor Davies can be criticised for his failure to identify the particular reaction mechanism upon which the claimants rely. This paper must be seen in context, as I elaborate later in this judgment. It was one of a number of papers found by or on behalf of Professor Davies on a search and he reviewed it as he would any other paper. When the particular reaction mechanism upon which the claimants rely was drawn to his attention in cross-examination he readily accepted what it disclosed. In general, and subject to the one matter to which I have referred, I believe that the opinions of Professor Davies must be accorded considerable weight and I have found his evidence of great value.

Dr Pochapsky is Professor of Chemistry and Biochemistry and member of the Rosensteil Basic Medical Science Research Institute at Brandeis University in Waltham, Massachusetts. He is a specialist in analytical chemistry and was awarded his PhD at the University of Illinois under the supervision of Professor Pirkle in 1986. Following the completion of his PhD, Dr Pochapsky continued to write review articles with Professor Pirkle on chiral separation from 1986 through to 1988. I have no doubt that Dr Pochapsky was very well experienced in the relevant analytical techniques at the priority date. However, there are aspects of his first report which he quickly abandoned during the course of his oral evidence. Further, he sought in that report to cast doubt on the availability of a particular chiral HPLC column called Chiralcel OD, to which I refer to later in this judgment, without mentioning various matters and documents which had come to his attention during the course of the US trial relating to this same invention. I accept his explanation that he overlooked these matters, but they reveal a certain lack of care in the preparation of his statement which I think must therefore be approached with caution. Nevertheless, I found Dr Pochapsky’s evidence under cross-examination to be fair and frank and he readily apologised for inconsistencies with his written report. I have no hesitation in relying upon the evidence he gave under cross examination.

Professor Montgomery is Emeritus Professor of Psychiatry at the Imperial College School of Medicine, University of London. He was a Member of the Council of the Royal College of Psychiatrists from 1980 to 1992 and a Member of the Committee on Safety of Medicines in the UK from 1987 to 1993. In 1978 he developed the MADRS (Montgomery Åsberg Depression Rating Scale) to measure levels of depression. This is one of two pivotal scales of efficacy measuring the severity of depression and the effect of antidepressants approved by the Licensing Authorities. He has completed a total of some 40 expert reports to support regulatory applications for CNS medications, including for SSRIs (fluoxetine, paroxetine, citalopram and escitalopram) and other antidepressants (mirtazapine, reboxetine and moclobamide). The claimants accept that Professor Montgomery was honestly doing his best to help the court but submit that his long standing involvement with escitalopram has made him overly enthusiastic about it and that it was plain his views were not representative. I accept that Professor Montgomery’s views are not representative of all psychiatric clinicians; indeed Professor Reid would be one such. Nevertheless, I am entirely satisfied that Professor Montgomery has an immense knowledge and understanding of the efficacy of escitalopram. He gave his evidence in a balanced and fair way and I have found it of very great assistance.

The claimants contended that claims 1 and 3 of the Patent are anticipated by the disclosures of the 193 and 884 patents. It was common ground these are enabling disclosures of the citalopram racemate and that 193 discloses a pharmaceutical composition containing that racemate.

The law of novelty was recently explained by the House of Lords in Synthon v Smithkline Beecham [2005] UKHL 59; [2006] RPC 323. There are two requirements. First, the matter relied upon as prior art must disclose subject matter which, if performed, would necessarily result in an infringement of the patent. Second, that disclosure must have been enabling, that is to say the ordinary skilled person would have been able to perform the invention if he attempted to do so by using the disclosed matter and common general knowledge.

In the present case the claimants did not suggest the 193 or 884 patents disclose the (+) enantiomer, escitalopram. That must be right. The disclosure of a racemate does not of itself amount to a disclosure of each of the enantiomers. This was confirmed by the Technical Board of Appeal of the EPO in decision T1046/97, Optically active triazole derivatives and compositions of 2 December, 1999. Instead, the claimants contended that claim 1 of the Patent has a scope which extends to the (+) enantiomer when in the racemate. If the racemate falls in the scope of claim 1 then, said the claimants, the claim must be invalid because the racemate was disclosed in the 193 and 884 patents and that disclosure is accepted to be enabling.

Lundbeck, on the other hand, contended that claim 1 of the Patent is limited to the isolated or pure (+) enantiomer (and non-toxic acid addition salts thereof) and so excludes the racemate. The dispute is therefore one of construction and turns on whether or not claim 1 covers the (+) enantiomer in the racemate.

As I have mentioned, claim 1 merely sets out the molecular formula of citalopram with a (+) beside it to identify one particular enantiomer. The claim contains no express limitation as to purity, whether as measured in relation to the other enantiomer or in relation to other organic or inorganic material.

The correct approach to construction was explained by Lord Hoffmann in Kirin-Amgen at [27] – [35]. The question is what the person skilled in the art would have understood the patentee was using the language of the claim to mean. As he emphasised, the meaning of words is highly sensitive to context.

I have already explained in some detail the substance of the description of the Patent. For present purposes I need only emphasise certain aspects of it. First, the title of the invention is “New enantiomers and their isolation”. This indicates at the outset that the inventors were concerned with the provision of individual enantiomers and a way to isolate them. The body of the specification continues with the same theme. Page 2, lines 1-20 emphasise the invention relates to the two novel enantiomers and to the use of these enantiomers as antidepressant compounds, that is to say as individual enantiomers. At page 2, line 36 the specification explains the invention is also concerned with the method to resolve the racemate into the individual enantiomers.

Page 3 explains that, to the inventors’ surprise, it was found that the activity resides in the (+) enantiomer and continues with the description of the two schemes of synthesis which I have detailed earlier in this judgment, each of which resulted in the production of the pure enantiomers (see page 3, line 28; page 4, line 54). Example 1 then describes the resolution by method (a) and resulted in the isolation of 1.1g of enantiomerically pure compound (page 5, line 56). After reaction with potassium t-butoxide in dry toluene, this yielded 0.6g of the (+) enantiomer with an optical purity of 99.6%. (page 6, line 5). In an analogous way the (-) enantiomer was synthesised and produced with an optical purity of 99.9% (page 6, line 8). Example 2 describes resolution by methods (b) and (c), which again resulted in the production of the pure enantiomers.

The examples are followed by the description of the in vivo and in vitro tests carried out on the enantiomers of example 1 and show, as I have indicated, that the activity lies in the (+) enantiomer. Finally, I would refer to the formulations which all contain the (+) enantiomer, not the racemate.

Against this background I come to consider claims 1 and 3. On the face of it they are directed to the (+) enantiomer and a pharmaceutical composition containing it. Absent the description, it seems to me the person skilled in the art might well regard the language of the claims as ambiguous. However, when read in the context of the specification as a whole, that ambiguity is, I believe, quite clearly resolved. In my judgment it would be apparent to any skilled person reading the Patent that it is directed at the isolation of the individual enantiomers of the citalopram racemate and the use of the isolated (+) enantiomer to make a pharmaceutical composition. As Lord Hoffmann explained in Kirin-Amgen at [34], the skilled person would recognise that the patentee is trying to describe something which, at any rate in his opinion, is new. The Patent not only recognises citalopram, that is to say the racemate, as old but also is plainly directed at the isolation of its individual enantiomers. It would be improbable that the patentee intended to cover that which he has expressly described as being old: See Ultraframe v Eurocell Building Plastics [2005] EWCA Civ 761; [2005] RPC 36 at [47] and Beloit Technologies Inc v Valmet Paper Machinery [1995] RPC 705 at 720.

The claimants submitted that the conventional approach to a product claim in a patent is that it covers the product wherever found. As Lord Hoffmann said in Merrel Dow Pharmaceuticals v Norton [1996] RPC 76 at 82, lines 36 to 44:

“The scope of the monopoly conferred by a product claim is defined by section 60(1)(a), which provides that where the invention is a product, a person infringes the patent if, without the consent of the proprietor, he “makes, disposes of, offers to dispose of, uses or imports the product or keeps it whether for disposal or otherwise.” For this purpose it does not matter how the product is made or what form it takes. The monopoly covers every aspect of manufacture and every form which comes within the description in the claim. So claim 24 includes the making of the acid metabolite in one’s liver just as much as making it by synthetic process; in the body as well as in isolation.”

This is undoubtedly so, but it depends upon the scope of the monopoly in issue. If the claim, properly construed, is simply to the product as such then the monopoly will indeed cover that product wherever it may be found. If, on the other hand, the claim is to the isolated product or to a pharmaceutical composition comprising as an active ingredient that isolated product then the monopoly will not extend to that product whatever form it takes. In my judgment the citalopram racemate does not come within the description of claim 1 and a pharmaceutical composition comprising as an active ingredient the racemate does not come within the description of claim 3.

Lundbeck advanced two further contentions which, in the light of my conclusions, I can deal with quite shortly. First, Professor Davies said in evidence that the (+) designation is a macroscopic property of a material. He said this could not cover the enantiomers in a racemate because the racemate will not rotate polarised light. Those in the art do not use (+) or (-) to refer to the specific orientation of separate molecules. I do not think there is anything in this point. Professor Davies exhibited to his report extracts from the fifth edition of the well known textbook Organic Chemistry by Morrison and Boyd, 1987 as containing what would have been part of the common general knowledge of the skilled addressee. It contains two clear examples of the (+) designation being used to identify an enantiomer as one of the components of a mixture, being in one case a racemate and in the other, a mixture containing more of one enantiomer than the other (see, particularly, Chapter 4, sections 4.25 and 4.27). Moreover, the Patent itself uses the (+) and (-) designations to refer to the two components of a racemic mixture in Reaction Scheme I. I think that Morrison and Boyd is consistent with the conclusion that I have reached, namely that the skilled addressee would understand the meaning to be attributed to any particular use of the designation (+) from the context in which it appears.

The second contention advanced by Lundbeck was that when the absolute purity of a “pure” compound is not specified then, as a matter of established convention, it is understood to be a least 95% pure. This was supported by Professor Davies who said in his first report that this is so because the standard method for measuring purity (including optical purity) is NMR and this can only reliably detect impurities if they are present at a level of 5% or greater. However, in the light of the evidence he gave in the course of cross examination I do not accept there is any such convention. I believe medicinal chemists consider purity on a case-by-case basis. Once again, all depends upon context. The precise scope of the term pure in the context of the Patent is not a matter that requires determination in these proceedings. It is sufficient to say that claims 1 and 3 are directed to the isolated (+) enantiomer and to compositions containing that isolated enantiomer. They do not extend to the racemate.

Lundbeck has made a conditional application to amend claims 1 and 7 of the Patent in the event I concluded, contrary to its submission, that the claims were not limited to the isolated (+) enantiomer. In summary, Lundbeck proposed adding to the body of the specification a passage making clear that it did not claim any racemates or enantiomers as part of any such racemates. It also proposed adding to claims 1 and 7 a specific limitation that the particular enantiomers identified were claimed “as the individual enantiomer”. In the light of my conclusion it is not necessary for Lundbeck to pursue this application. Nevertheless, I would observe that had I reached the opposite conclusion I would have been minded to allow some amendment which had the effect of limiting the claims to the isolated (+) enantiomer. It is abundantly clear from the passages in the body of the specification to which I have referred that the invention is concerned with isolated (+) enantiomer and methods of making it.

As I have mentioned, the claimants contend that claims 1, 3 and 6 are invalid for obviousness. The obviousness case has two limbs which are essentially distinct. The claimants attack all three claims on the basis that it was obvious to produce a pharmaceutical composition containing escitalopram by the method disclosed in the Patent and claimed in claim 6 (the amino diol case). They have a separate and self standing attack against claims 1 and 3 on the basis that it was obvious to separate the enantiomers of citalopram using chiral HPLC and to use the escitalopram so separated to make a pharmaceutical composition (the chiral HPLC case). In particular they say that in the course of a systematic but uninventive study, the person skilled in the art would be likely to try certain chiral HPLC columns and be able to separate the enantiomers of citalopram. It is to be noted that the claimants do not say that that the claims are obvious because they claim an obvious desideratum, although this does form the basis of the insufficiency case, as I shall explain. I will deal with each of the obviousness attacks in turn. But before doing so it is convenient to mention some general principles which have a particular bearing on this case.

Both parties approached the question of obviousness using the structured approach explained by the Court of Appeal in Windsurfing International Inc. v Tabur Marine (Great Britain) Ltd [1985] RPC 59. This can be summarised as follows:

In considering the vital fourth question it is always important to avoid hindsight, as Jacob LJ re-emphasised in Ferag v Muller Martini [2007] EWCA Civ 15 at [13]. The question must be considered without assuming knowledge of the invention. Once an invention has been made it is often all too easy to postulate how it might have been arrived at by a series of apparently obvious steps from something that was known. This is particularly important in a case such as the present for two reasons. First, the invention was made many years ago and the state of the art has advanced considerably in the meantime. Second, the attacks of obviousness are based in large part upon what is said to have been common general knowledge. As Aldous LJ noted in Coflexip v Stolt Connex Seaway, 31 July 2000, unreported on this point, at [45], the advantage of starting an attack upon this basis may well be that it is unencumbered by any detail.

The primary evidence to be considered is that of properly qualified expert witnesses who can explain to the court whether the relevant step would have been obvious to the skilled person having regard to the state of the art. All other evidence is secondary to that primary evidence. Evidence of contemporaneous events can be of assistance when testing the expert’s primary evidence. So also can evidence of commercial success. Secondary evidence has its place and the weight to be attached to it will vary from case to case. However, such evidence must be kept firmly in its place as Sir Donald Nicholls V.C. explained in Molnlycke v Procter & Gamble Ltd (No 5) [1994] RPC 49 at 113.

If a particular route is an obvious one to take, it is not rendered any less obvious from a technical point of view merely because there are a number, and perhaps a large number, of other obvious routes as well: Brugger v Medic-Aid Ltd [1996] RPC 635 at 661.

This is a case in which it is said that the method of claim 6 was one which was obvious to try. The “obvious to try” doctrine was considered by the Court of Appeal most recently in Angiotech Pharmaceuticals v Conor [2007] EWCA Civ 5:

“[39] Given that position, what Mr Waugh endeavoured to do was to say that the Judge had wrongly applied an "obvious to try" test. It is necessary to say a little about this. The expression got into the law of obviousness by virtue of the Johns-Manville case, [1967] RPC 479. The facts were simple: there was a known process. The patent was for the old process using the new agent. It was held obvious as being "well worth trying out". Diplock LJ said:

"It is enough that the person versed in the art would assess the likelihood of success as sufficient to warrant actual trial"

[40] More recently, in this court I, with the concurrence of Peter Gibson and Scott Baker LJJ said:

"Mere possible inclusion of something within a research programme on the basis you will find out more and something might turn up is not enough. If it were otherwise there would be few inventions which were patentable. The only research which would be worthwhile (because of the prospect of protection) would be in areas totally devoid of prospect. The "obvious to try test really only works where it is more-or-less self evident that what is being tested ought to work", St Gobain v Fusion Provida [2005] EWCA Civ 177.

[41] Judge Rich in the US Court of Appeal for the Federal Circuit said (I did not know this when I wrote St Gobain ) much the same thing in Tomlinson's Appn (1966) 363 F 2d 298 at 931:

"Slight reflection suggests, we think, that there is usually an element of 'obviousness to try' in any research endeavour that is not undertaken with complete blindness but rather with some semblance of a chance of success, and that patentability determinations based on that as the test would not only be contrary to statute but result in a marked deterioration of the whole patent system as an incentive to invest in those efforts and attempts which go by the name of 'research'."

[42] Mr Waugh submitted that was the correct approach and that it was that approach which was also followed in Australia (Hässle v Alphapharm (2002) 312 CLR 411), Canada (Aventis v Apotex (2005) [2005] FC 1504) and the USA (Tomlinson and re O'Farell (1988) 853 F 2d 894 also per Judge Rich).

[43] I have to say that I do not discern a shift in the position in this country following the 1977 Act as the majority of the Australian High Court thought had happened. It is perhaps noteworthy that currently Australian courts seem to be taking a very pro-patent view of obviousness and that patents are being upheld there which are not upheld elsewhere. The Hässle case and the Viagra case, Pfizer v Lilley (held by the Federal Court of Appeal non-obvious though invalid on other grounds) are perhaps examples of this. Whether, if that is so, it is good for the Australian economy is not my concern.

[44] I also take the view that one can overelaborate a discussion of the concept of "obviousness" so that it becomes metaphysical or endowed with unwritten and unwarranted doctrines, sub-doctrines or even sub-sub-doctrines. This can be coupled with a massive citation of authority (the opinions in the 84 printed page, 203 paragraph judgment, in Hässle have 307 footnotes, many of which are citations of authority); Diplock LJ warned against this in Johns Manville saying:

"I have endeavoured to refrain from coining a definition of 'obviousness' which counsel may be tempted to cite in subsequent cases relating to different types of claims."

I interpolate to say, he failed there! Continuing:

"Patent law can too easily be bedevilled by linguistics and the citation of a plethora of cases about inventions of different kinds. The correctness of a decision upon an issue of obviousness does not depend upon whether or not the decider has paraphrased the words of the Act in some particular verbal formula. I doubt whether there is any verbal formula which is appropriate to all classes of claims."

[45] That reminder cannot be repeated too often. The words of the law are simply:

"An invention shall be considered as involving an inventive step, if, having regard to the state of the art, it is not obvious to a person skilled in the art" (Art 56 EPC).

In the end the question is simply "was the invention obvious?" This involves taking into account a number of factors, for instance the attributes and cgk of the skilled man, the difference between what is claimed and the prior art, whether there is a motive provided or hinted by the prior art and so on. Some factors are more important than others. Sometimes commercial success can demonstrate that an idea was a good one. In others "obvious to try" may come into the assessment. But such a formula cannot itself necessarily provide the answer. Of particular importance is of course the nature of the invention itself.”

Paragraph [45] is, to my mind, key. The question of obviousness must be considered on the facts of each case. The court must consider the weight to be attached to any particular factor in the light of all the relevant circumstances. These may include such matters as the motive to find a solution to the problem the patent addresses, the number and extent of the possible avenues of research, the effort involved in pursuing them and the expectation of success.

Lundbeck contended that the inventive concept of each of the claims in issue is straightforward:

I am content to adopt this formulation for the purpose of considering obviousness. It provides a convenient framework in which to consider whether it was obvious to make any product or carry out any process within the scope of the claims. However, as I shall explain in considering the issue of insufficiency, I do not accept that it reflects with adequate precision the technical contribution that Lundbeck has actually made.

I have identified the skilled addressee of the Patent in paragraphs [36]–[42] above. I need say no more about him. The law as to what constitutes the common general knowledge of the addressee was explained by the Court of Appeal in Beloit Technologies Inc v Valmet Paper Machinery Inc [1997] RPC 489 at pages 494-495:

“It has never been easy to differentiate between common general knowledge and that which is known by some. It has become particularly difficult with the modern ability to circulate and retrieve information. Employees of some companies, with the use of libraries and patent departments, will become aware of information soon after it is published in a whole variety of documents; whereas others, without such advantages, may never do so until that information is accepted generally and put into practice. The notional skilled addressee is the ordinary man who may not have the advantages that some employees of large companies may have. The information in a patent specification is addressed to such a man and must contain sufficient details for him to understand and apply the invention. It will only lack an inventive step if it is obvious to such a man.

It follows that evidence that a fact is known or even well-known to a witness does not establish that that fact forms part of the common general knowledge. Neither does it follow that it will form part of the common general knowledge if it is recorded in a document. As stated by the Court of Appeal in General Tire & Rubber Co. v. Firestone Tyre & Rubber Co. Ltd. [1972] R.P.C. 457, at page 482, line 33:

"The two classes of documents which call for consideration in relation to common general knowledge in the instant case were individual patent specifications and widely read publications'. As to the former, it is clear that individual patent specifications and their contents do not normally form part of the relevant common general knowledge, though there may be specifications which are so well known amongst those versed in the art that upon evidence of that state of affairs they form part of such knowledge, and also there may occasionally be particular industries (such as that of colour photography) in which the evidence may show that all specifications form part of the relevant knowledge.

As regards scientific papers generally, it was said by Luxmoore, J. in British Acoustic Films (53 R.P.C. 221 at 250):

"In my judgment it is not sufficient to prove common general knowledge that a particular disclosure is made in an article, or series of articles, in a scientific journal, no matter how wide the circulation of that journal may be, in the absence of any evidence that the disclosure is accepted generally by those who are engaged in the art to which the disclosure relates. A piece of particular knowledge as disclosed in a scientific paper does not become common general knowledge merely because it is widely read, and still less because it is widely circulated. Such a piece of knowledge only becomes general knowledge when it is generally known and accepted without question by the bulk of those who are engaged in the particular art; in other words, when it becomes part of their common stock of knowledge relating to the art."

And a little later, distinguishing between what has been written and what has been used, he said:

"It is certainly difficult to appreciate how the use of something which has in fact never been used in a particular art can ever be held to be common general knowledge in the art."

Those passages have often been quoted, and there has not been cited to us any case in which they have been criticised. We accept them as correctly stating in general the law on this point, though reserving for further consideration whether the words 'accepted without question' may not be putting the position rather high: for the purposes of this case we are disposed, without wishing to put forward any full definition, to substitute the words 'generally regarded as a good basis for further action'.”

As the court noted, the notional skilled addressee is the ordinary man who may not have the advantages that some employees of large companies may have and information does not form part of the common general knowledge simply because it is known to some persons in the art. It must be generally known and generally regarded as a good basis for further action by the bulk of those engaged in that art before it becomes part of their common stock of knowledge relating to the art, and so part of the common general knowledge. Further, it follows that it is difficult, but certainly not impossible, to show that the use of something which has in fact never previously been used in a particular art can be held to be a part of the common general knowledge. As the court later explained (at page 497), when deciding whether something forms part of the common general knowledge the first and most important step is to look at the sources from which notional skilled addressee could have acquired his information. Such sources can vary from instruction at university to description in obscure patents and specifications. Whatever the source, it is necessary to have in mind the observations of the Court of Appeal in General Tire.

There was a measure of agreement between the parties as to matters which formed part of the common general knowledge and I can deal with these relatively quickly. But there were two major areas of disagreement, namely:

First, it was accepted that the matters I have related as to the nature and causes of depression were all common general knowledge. So too were the methods of treatment available in 1988.

Second, it was agreed that the fundamental principles of chirality, enantiomers, stereoisomers, racemates and optical activity which I have summarised were common general knowledge and taught to students very early on.

Third, there was agreement as to the nature of diastereoisomers and the fact they are not mirror images and therefore can sometimes be separated using classical methods. I have also summarised these matters and again it was accepted that they formed part of the common general knowledge. As I elaborate below, it was also generally known that resolution by the use of diastereoisomers is not always successful.

Fourth, there was agreement as to the basics of the reactions that take place in carbon atoms in organic chemistry and how they proceed. I have described the nucleophilic reactions that take place at a carbon atom. These reactions and, in particular, the nature of the S_N1 and the S_N2 substitution reactions were matters of common general knowledge. The skilled addressee would have known that S_N2 reactions invert the stereochemistry at the chiral centre but that S_N1 reactions do not.

Dr Newton explained various factors that have an effect on the rate and mechanism of nucleophilic substitution reactions and these were also accepted to be matters of common general knowledge. Generally, factors that stabilise carbocations favour the S_N1 over the S_N2 mechanism. So benzylic carbocations were known to be stabilised by delocalisation of the positive charge over the benzene ring. Similarly, in the case of alkyl halides, the alkyl groups exert an inductive effect that increases on traversing the series methyl, ethyl, isopropyl and t-butyl so making the carbon atom to which the halide is attached progressively more negative and therefore less susceptible to attack by a nucleophile. In addition, as the series is traversed the carbon atom becomes progressively more sterically hindered, again favouring the S_N1 mechanism. So, for example, methyl bromide undergoes S_N2 reactions rapidly whereas t-butyl bromide does not undergo S_N2 reactions but will undergo S_N1 reactions.

The effect of the leaving group was also known to play an important part in determining the rate of reaction and, indeed, whether substitution will take place at all. As I have mentioned, it had been recognised for many years that the hydroxyl group of ordinary alcohols is not a leaving group. In order to make an aliphatic alcohol undergo a nucleophilic substitution it is necessary to activate it in some way. One well known way of achieving this is to convert the alcohol into a sulphonyl ester. This dramatically increases its leaving group ability and permits a nucleophilic substitution reaction. An alternative method of activating an alcohol for nucleophilic substitution was known to be by protonation of the hydroxyl group, for instance, by using a strong acid. This results in the formation of a carbocation upon the loss of a water molecule.

It was generally known that nucleophile strength also plays a part. A nucleophile with a negative charge is more powerful than its conjugate acid. Tertiary alcohols were generally known to be poor nucleophiles compared to primary alcohols. Steric factors were also known to be important.

Finally, the existence of what have become known as Baldwin’s rules were well known to organic chemists in 1988. How the skilled person would apply these rules was, however, hotly disputed. I return to this issue in considering the expert evidence on the issue of obviousness.

This was a major area of dispute and it has an important bearing on the issues of obviousness and insufficiency. The claimants contended that by June 1988 it was common general knowledge that the biological activity of enantiomers was likely to be different and that for any racemate under development as a potential therapeutic agent it was highly desirable, if not necessary, to resolve the racemate to determine the properties of the enantiomers it contained.

Lundbeck contended that it was not possible to say either that the biological activities of the enantiomers would be different or that the skilled person would expect that only one of the enantiomers would be responsible for most of the biological activity. Further, in June 1988 there was little interest in the preparation of single enantiomeric compounds and there was little interest in the resolution of racemates. It accepted, however, this attitude began to change in the early 1990s, primarily because the US Food and Drug Administration (“FDA”) and, sometime later, the European regulatory authorities were becoming interested in racemic drugs from a regulatory point of view and it was apparent that regulatory guidelines and requirements concerning such drugs were likely to be introduced. In the early 1990s it was recognised those requirements were likely to include an obligation to provide some data relating to the individual enantiomers of a racemic drug.

Timing in this case is therefore critical. The parties agreed that in the early 1990s there was a motive to resolve the enantiomers of a racemate but Lundbeck disputed that any such motive existed in June 1988.

I am satisfied on the evidence that it had been recognised for very many years prior to 1988 that the vast majority of drugs exert their activity by binding to a protein receptor to form a drug-receptor complex.

Protein receptors are composed of a complex array of chiral amino acids and it follows that the binding site of any drug is in a chiral environment. In circumstances where the drug substance has an asymmetric carbon atom or atoms, the binding efficiency to a given receptor and so also the biological activity of the enantiomers may well be different. This was well understood by the priority date and illustrated by the drug Thalidomide. It was prescribed as a mild sedative and anti-emetic. It has an asymmetric centre but was marketed as the racemate. It was found that one enantiomer was a non-mutagenic sedative, while the other was mutagenic and caused widespread deformities amongst those children whose mothers took the drug during pregnancy.

Dr Newton maintained that as a result of the way in which most drugs exert their activity it was generally known that very often the majority of the biological activity observed for a racemate resided within a single enantiomer. I am satisfied that this was so. However, I also accept that no sure prediction could be made in any particular case. As Professor Davies explained, there were cases where both enantiomers were equally active, others where one enantiomer was active and the other inactive and a spectrum of cases in between.

In the mid 1980s, the FDA began to request information, not only in relation to racemates, but also on the constituent enantiomers, in order to evaluate the relative benefits and practicability of developing single enantiomers. In February 1987, the FDA published a “Guideline for Submitting Supporting Documentation in Drug Applications for the Manufacture of Drug Substances”. I consider this a very important document. It made clear that it did not impose mandatory requirements but offered guidance on acceptable approaches to meeting regulatory requirements. In relation to the requirements for a new drug application (“NDA”), it stated at paragraph III.D.

“ There are two principal reasons for requesting a detailed presentation of the synthetic pathway and/or manufacturing process. First, a particular synthetic pathway will typically be uniquely associated with a set of impurities (actual and potential), and also a specific solid-state form. The impurities may have significant clinical or toxicological effects. It should be noted that (even in racemates) enantiomers may be considered as impurities. Proper control of the synthetic process monitors impurity levels during the process and in the final bulk drug substance. When a change is made or proposed in a synthetic process, a different ratio or even different set of impurities may arise, and the control testing may need modification. The “United States Pharmacopeia” (USP) also states that tests, besides those provided in drug substance monographs, may be necessary when a change in the source of material or processing occurs.”

Later, in addressing the requirements for an investigational new drug (“IND”), it stated that present regulations required the submission of manufacturing and controls information for the new drug substance along with the name and address of the manufacturer. Depending on the type of IND and the phase of investigation, information should be submitted as described in the various subsections which followed. One of these, section III.A.1.a, concerned physical and chemical characteristics and provided:

“When the NDS [new drug substance] is asymmetric (e.g., contains one or more chiral centers, or has cis-trans or other types of isomers), the sponsor should ideally (and prior to the submission of an IND) have either separated the various potential stereoisomers of the NDS or synthesized them independently. Physical/chemical information about each stereoisomer should be provided (in detail) or may be requested. Individual stereoisomers may need to be studied for pharmacological and toxicological properties (and/or for safety and efficacy).

Appropriate specifications and tests to control the ratios of any admixtures (e.g., ratios of enantiomers, and/or solid-state forms) for batches of drug substances used in toxicological and/or clinical studies should be established, so that results can be extrapolated to the drug substance prepared for marketing.”

Dr Newton was aware of these issues because they were relevant to his involvement in the development at Glaxo of Zofran and Serevent in 1985 and 1986.

In the case of Zofran, this has an asymmetric carbon atom and Glaxo studied the racemate and the individual enantiomers. Ultimately the drug was marketed as a racemate because it was found that when the enantiomers were separated and the more active one administered on its own, it was racemised in vivo to give a mixture of the two enantiomers. Serevent was also marketed as a racemate because at the time it was technically difficult to develop an industrial scale synthesis of the more active enantiomer. In the case of both drugs, however, the enantiomers were separated and their individual biological activities were determined during the course of development.

Dr Newton’s experience was that four out of four chiral molecules under development at Glaxo in the 1980s were resolved. It was made clear to the medicinal chemists working on these projects that if the company wanted to launch a drug in America they should provide biological data on the enantiomers, and that is what they did. Any compounds with chiral centres which looked as if they were likely to progress were routinely and as a matter of course, resolved. Glaxo took the 1987 FDA guidelines as being what the FDA required, and Glaxo attempted to meet them. It was Dr Newton’s opinion in the light of these matters and what was known about the activity of enantiomers and mechanisms of drug action that any medicinal chemist looking at citalopram at the priority date would have considered it highly desirable to determine which enantiomer was effective.

It is also clear the FDA was not standing alone. Indeed, Japan was some way ahead. The Japanese “Requirements for Drug Manufacturing Approval” were amended in 1985 to add a sentence in the section “Test Data Concerning Absorption, Distribution, Metabolism and Excretion” stating:

“When the drug concerned is a racemate, it is recommended to investigate the absorption, distribution, metabolism and excretion of each optical isomer.”

This would of course require resolution of the enantiomers and the administration of each enantiomer separately.

The position in the UK was much the same. At a symposium on stereoisomerism at the Winter Meeting of the British Pharmacological Society held in London on 6 January 1988, the issue of the pharmacological implications of stereoisomerism was discussed and a speaker from the Secretariat of the Committee of Society of Medicines outlined the regulatory aspects of the issues of racemates and enantiomers in this jurisdiction. Following that meeting, a paper by Robert Smith and John Caldwell published in 1988 (Trends in Pharmacological Sciences, (TIPS), March 1988, Vol 9, No. 3. page 75) included the following passage headed “Responses of the regulatory authorities”:

“The regulatory authorities are beginning to respond to the scientific and clinical maelstrom concerning the issue of racemates and enantiomers. In the United States the Food and Drug Administration has both an ongoing intramural discussion and an extramural debate with industry which is likely to lead to the promulgation of guidelines. Matters are further advanced within the EEC, where at present the Committee for Proprietory Medicinal Products is considering a draft guideline statement on isomerism for inclusion in its ‘Notice to Applicants’. If adopted, as looks likely, new submissions for drugs with chiral centres will have to provide information inter alia on the following points: isomer ratio and batch to batch consistency; a discussion of the toxicological and pharmacological properties of the isomers, enantiomer specific metabolism and kinetics and the extrapolation of preclinical data (particularly if species differences occur in the handling of stereoisomers); and a discussion of possible clinical problems that may arise in relation to stereoisomers. ”

The same emerges from a paper by Wilson De Camp of the FDA, published in 1989 (Chirality 1:2-6 (1989), pages 1-9) adapted from remarks delivered to the Midwest Regional Meeting of the American Association of Pharmaceutical Scientists in Chicago on May 16, 1988. He pointed out that in practice the decision about whether to develop the racemic or optically pure form of the drug is made well before the time that an NDA is made to the FDA. In the course of drug development, a manufacturer should consider both enantiomers, as well as the racemate, to be potential drugs. He observed that data would be needed to support an NDA and that in addition to physical, chemical and pharmacological studies, clinical studies to compare the safety and efficacy of the racemate with the enantiomers might be needed. He continued that whether the decision was to market a racemate in preference to a pure enantiomer, or the reverse, it should be justified by the submission of appropriate data. He pointed out that the FDA guidelines noted that even in racemates, enantiomers might be considered as impurities. Further, good science required that conclusions by the regulators should be based on experimental evidence derived from well planned experiments. Such planning should not neglect the potential for differences in properties for enantiomers of a chiral molecule in a chiral environment. Thus not only was it desirable to recognise the implications of stereochemistry for drug action, but it was also desirable that they be investigated. Either the enantiomers should be separated, or they should be synthesised. Good sense required that the hazards associated with the use of any substance, or its components, be identified. The same should be applied to enantiomeric molecules in a racemate. He concluded whenever a drug could be obtained in a variety of chemically equivalent forms (such as enantiomers), it was both good science and good sense to explore the potential for in vivo differences between these forms.

These publications all seem to me to present a consistent and powerful picture that by 1988 it was well understood by medicinal chemists that:

The conclusions I have drawn from the publications are supported by the actions of the various pharmaceutical companies involved in seeking to develop 5-HT reuptake inhibitors. The position established by the evidence was as follows. Fluoxetine (Prozac) was launched soon after 1998 by Eli-Lilly as a racemate. But by 1985 Eli-Lilly had resolved the racemate and found that (+) isomer was slightly more potent than the (-) isomer. Venlafaxine was a racemate first introduced by Wyeth in 1993. However, by December 1983 Wyeth had resolved the racemate into its two enantiomers. Sertraline was launched by Pfizer in 1991 as a single enantiomer. But by 1979 Pfizer had resolved the racemate in order to determine the properties of the individual enantiomers. Duloxetine was marketed as a singular enantiomer which had been resolved by Eli-Lilly by December 1987. Paroxetine was launched by SmithKline Beecham as a single enantiomer shortly after 1988. Finally and, as I shall explain, Lundbeck attempted to resolve citalopram through the 1980s. In summary, Eli-Lilly, Pfizer, Wyeth and SmithKline Beecham had resolved their 5-HT reuptake inhibitors by 1988 and, in the case of Lundbeck, had attempted to do so.

This picture is further confirmed by written observations of both Dr Pochapsky and Professor Davies before the priority date. Dr Pochapsky wrote a review article with Professor Pirkle in 1987 (Advances in Chromatography Vol. 27, Chap.3 (1987) at page 74) in which they said:

“ To discuss biological activity of a racemate without at least noting the possibility of differential or even opposing effects of the enantiomers is foolhardy, especially when many examples of such situations are known. Nowadays, enantiomeric impurities are often quite simple to monitor. ”

For his part, Professor Davies published a paper in 1987 on dimethylamphetamine (Tetrahedron, Vol.43, No. 19, pages 4463-4471) in which he stated:

“Whilst many pharmaceuticals are still produced in racemic form there is a growing appreciation of the different biological effects of enantiomeric molecules and as a result the preparation of optically pure β-amino-alcohols has attracted considerable attention. ”

Not surprisingly, all of these materials were put to Professor Davies in cross-examination. He maintained that industry was not motivated to resolve racemates or to study individual enantiomers in the late 1980s. He explained both in his reports and in cross-examination that he was not consulted on the preparation of single enantiomer compounds during the 1980s to any real extent. It was his experience that molecules that were first made as racemates tended to proceed as racemates through development, whereas molecules that were first made individual enantiomers proceeded through development as single enantiomers. It was perceived by industry that the additional expense and difficulty of resolution of racemic materials meant that the exercise was not worthwhile. Moreover, in the late 1980s there was no regulation which made resolution necessary. This attitude was, he said, reflected in the fact that when he presented to industry his idea of forming a company to provide pharmaceutical companies with homochiral compounds of interest on any desired scale he was met with a resounding lack of interest. Once a company had proceeded with the development of a racemate and shown it was both safe and efficacious then there was simply no incentive to go back and investigate the enantiomers. That course would involve both cost and delay.

As I have indicated, it was in giving this evidence that Professor Davies was accused of losing objectivity and stepping into the realm of the advocate. I think this is unduly harsh. In my judgment Professor Davies based his opinions primarily upon his own experience and there are aspects of his evidence which I have no difficulty accepting. For example, it was certainly the case that there was no mandatory requirement that enantiomers be investigated. Further, if a racemate had been shown to be safe and efficacious then there would clearly be less incentive for the company selling it to embark on the potentially costly course of developing a version based upon an individual enantiomer. Similarly, I have no doubt that pharmaceutical companies were concerned that some racemates might prove very difficult or impossible to resolve.

Nevertheless, I am entirely satisfied on the evidence that medicinal chemists working in the field of SSRIs were well aware of the desirability of resolving racemic drug candidates and testing the individual enantiomers. In so far as there was a difference in the opinions of Dr Newton and Professor Davies I prefer those of Dr Newton. He was engaged at the relevant time as a medicinal chemist and was head of chemistry at Glaxo. Although he was not particularly interested in the FDA and its regulations, he was aware of the February 1987 guidelines and that for any racemate under development after that time the FDA wanted the enantiomers to be separated or synthesised and chemically, physically and biologically examined. Professor Davies, on the other hand, was not engaged in industry as a medicinal chemist but was consulted for his expertise as an organic chemist. It seems to be therefore that on this particular issue Dr Newton was better placed to assist the court than Professor Davies.

I am confirmed in this opinion by the fact that Professor Davies gave evidence in his report that it was only in the 1990s that regulators became interested in racemic drugs from a regulatory point of view, and it was apparent that regulatory guidelines and requirements concerning such drugs were likely to be introduced. This was plainly not the case. Further, Japan was evidently in advance of the FDA and the European regulatory authorities were also engaged with the issues. In my judgment Professor Davies was not sufficiently close to the industry to be aware of these matters.

As to the difficulty and cost of resolving a racemate, I accept that it might be difficult or impossible, particularly from a scale up point of view. That was something of which those in the art were well aware. But I do not accept that this means the skilled addressee had no incentive to investigate the enantiomers. For the reasons I have given, I have no doubt that he did, as confirmed by the actions of all the companies that marketed 5-HT reuptake inhibitors.

To summarise, I must consider the notional skilled medicinal chemist considering the disclosures of the 193 and 884 patents. At the priority date citalopram was not on the market although its activity was well known. In my judgment any medicinal chemist in this position in 1988 would have appreciated that the enantiomers might well have different activities, that an inactive enantiomer was, at best, ballast but might be toxic or have some other negative effect, that the regulators considered that an investigation of the enantiomers was desirable and that such an investigation might in due course become mandatory. All of these matters provided a clear motive to isolate and test the enantiomers and this would have been well understood by the notional skilled addressee. Investigation of the enantiomers of citalopram was an obvious goal for the ordinary skilled medicinal chemist in 1988.

In reaching these conclusions I have taken into account the evidence of Professor Montgomery and Professor Reid. Their positions were very similar. Professor Montgomery believed that the most fruitful areas for new antidepressants were compounds which were selective for both 5-HT and NA reuptake inhibition, such as venlafaxine. He, and other practising clinicians, paid little attention to whether drugs were racemates or single enantiomers. He would not have advised a company to embark on the resolution of citalopram. He would not have thought that there was any prospect that an enantiomer of citalopram would exhibit any increased efficacy. Professor Reid confirmed that in 1988 he too would not have advised Lundbeck to develop an enantiomerically pure form of citalopram because he would have seen no benefit in doing so.

I can quite understand the perspective of these clinicians. They had little interest in whether drugs were sold as racemates or single enantiomers. So far as they were concerned citalopram was a highly selective SSRI. They were more interested in entirely new molecules with different modes of action. This was clearly an obvious direction to go. Nevertheless, this does not affect my conclusion that the notional medicinal chemist did have an incentive to resolve citalopram. It had the potential to double the activity of the compound (assuming all the activity lay in one enantiomer), to remove any ballast from the pharmaceutical, to minimise the risk of toxicity or pharmacological problems and to satisfy the guidelines issued by the regulatory authorities which might, in due course, become mandatory.

Professor Davies carried out a review of the specialist monographs which showed thirteen different techniques for resolving racemates and these are identified in paragraph 29 of his report. However, few medicinal chemists had experience of all of these in 1988. A better picture emerges from Morrison and Boyd in which the resolution of racemates is described as a special kind of job requiring a special kind of approach. It continues in section 4.27 that most resolutions are accomplished through the use of reagents that are themselves optically active and I am satisfied this is where the skilled person would start.

For amines, the person skilled in the art could use chiral acids to make corresponding diastereomeric salts and try and separate these by fractional crystallisation; in the case of acids, he could use chiral amines to make the diastereomeric salts. Once the enantiomerically pure salts were obtained upon crystallisation, subsequent basification or acidification respectively and extraction would generate the enantiomers of the amine or the acid. The skilled person would not know whether or not this technique would work. Many compounds do not crystallise and some that do, do not crystallise differentially. As Dr Newton accepted, the skilled person could try this technique with dozens of different salts for many years and end up with a nil return.

An alternative approach could be used if the compound had a suitable reactive centre permitting it to form a covalent bond with a chiral reagent and so produce a pair of diastereoisomers. Thereafter the skilled person could attempt to separate these by, for example, crystallisation or standard column chromatography and then convert the separated enantiomers back to the desired molecule using synthetic procedures that did not compromise the stereochemical integrity of the chiral carbon atom. Once again, the outcome was unpredictable.

Another technique known to the skilled person was to seek to develop a chiral synthesis of the drug. This involves either starting from a commercially available chiral material or using chiral reagents to introduce chirality.

A yet further technique known to the skilled person was to seek to make derivatives of the drug, try and separate the derivatives and then convert the separated derivatives back to the enantiomers.

If the classical methods for resolution failed, the skilled person could seek to resolve an intermediate into its constituent enantiomers and then form the desired enantiomers of the final product by a stereo-specific reaction step. This final aspect is important. It is vital to have a viable method to move from the resolved precursor to the final molecule while maintaining its stereochemistry.

I must begin with a brief description of HPLC and chiral HPLC and analytical and preparative chromatography (whether chiral or non chiral). The following description, which I take largely from the report of Dr Collicott, was not in dispute.

Liquid chromatography, also referred to as column chromatography, is a technique used to separate compounds. It works on the principle that the mixture to be separated is dissolved in a liquid mobile phase and then passed over an adsorbent (“stationary”) phase which has been packed into a column. Different compounds adsorb to the stationary phase to different extents, and therefore migrate through the stationary phase at different rates and are separated as they emerge at the end of the chromatography column. A typical high performance liquid chromatography (“HPLC”) system is shown below:

The heart of the system is the column. In HPLC this is a stainless steel tube packed with the stationary phase. A sample containing the mixture of compounds is dissolved and introduced into the column. It is driven through the column by the mobile phase. Each compound in the mixture potentially has a different affinity for the stationary phase. Components that are less retarded will emerge earlier at the end of the column and will be detected sooner in the detector. A series of separated components will give rise to a series of peaks, displayed by the data system.

HPLC can be analytical or preparative. In the former, the eluent is discarded. In the latter, the eluent is directed to a series of collection of vessels, allowing different fractions to be collected from which the compounds can then be recovered.

A non chiral stationary phase cannot distinguish between enantiomers. Accordingly, where the objective is the resolution of a racemate into its component enantiomers, a chiral stationary phase (“CSP”) is packed into the column.

There was no dispute that both analytical and preparative HPLC were matters of common general knowledge at the priority date. However there was a dispute between the parties in relation to chiral HPLC and, in particular, as to the use of chiral HPLC for preparative purposes. This is important because preparative chiral HPLC would be required to produce the milligram quantities of enantiomer necessary for the tests described in the Patent.

By June 1988, there were five classes of commercially available CSPs which were classified according to the mechanism by which the stationary phase distinguished the enantiomers.

Pirkle columns were invented by Professor Pirkle in the late 1970s. By the early 1980s they were widely used and commercially available. They worked on the basis of electrostatic interactions between the solute enantiomer and the CSP.

This class of CSP works on the basis of differential complexation of the enantiomers with the CSP. A number of these CSPs became available commercially around the mid 1980s and were manufactured by companies called Daicel and Macherey Nagel.

These CSPs are based on immobilised proteins. By June 1988, two types were commercially available. One was based on bovine serum albumin and sold under the trade name Resolvosil by Macherey Nagel from about 1983. The other was a particular acid glycoprotein known as AGP and sold under the brand name EnantioPac by a company called LKB. No preparative columns of this type were available.

This was a large class and by June 1988 comprised three main types of CSP: polymethacrylate, microcrystalline cellulose triacetate and cyclodextrin. They all work on the basis that the enantiomeric molecule of interest is attracted into a cavity or “bucket” – a process called inclusion complexation. Different buckets can accommodate different sized molecules. The most relevant for present purposes are the cyclodextrin CSPs. These were marketed by a company called Advanced Separation Technologies (“Astec”) from about 1984 under the brand names Cyclobond I, II, and III, which corresponded to beta, gamma and alpha cyclodextrin respectively. Cyclobond III had the smallest cavity and was only suitable for resolution of lower molecular weight molecules. Cyclobond II had the largest cavity and was rarely used for chiral separation of pharmaceutical compounds because they were usually too small to achieve a tight fit in the cavity. By June 1988, an acetylated derivatised version of Cyclobond I (acetylated beta cyclodextrin) had become available as part of the range and a marketing brochure suggests that all were being advertised for analytical and preparative purposes.

This class of CSP was developed by Dr Okamoto from about 1984. Its main mechanism of operation is the formation of a solute-CSP complex, but inclusion complexation also plays an important role. Such CSPs were available commercially from the mid 1980s under the brand name Chiralcel from Daicel. For most of the late 1980s, the range consisted of five products, Chiralcel OA, OB, OC, OE and OK. These were all available on a preparative scale, except for OE.

There was a substantial issue at trial as to whether or not a further product, Chiralcel OD was available at the priority date. This has a particular relevance to the issue of obviousness because it is the one CSP which has been shown to permit the resolution of citalopram on a preparative scale. I am satisfied on the evidence that it was. It was advertised in April 1987 in the Journal of Chromatography and is referred to in a number of scientific papers published in 1987. Preparative and analytical Chiralcel OD columns are also specifically referred to in a Daicel promotional brochure entitled “Chiralpak Chiralcel, HPLC column for Optical Resolution, New Chiral HPLC Column” dated July 1987.

The parties were in agreement that a literature search to see what columns were available would turn up four important review articles written by significant figures in the field of chiral separation. The first was written by Dr Däppen and co-workers and published in 1986 (Journal of Chromatography 373 (1986) 1-20) (“the Däppen review”). This identifies the principal classes of CSP to which I have referred and identifies their manufacturers. There is of course no reference to Chiralcel OD in the light of its date. It suggests that preparative separations are possible with ligand-exchange, polysaccharide and cyclodextrin phases, but are impossible with protein phases.

The second is the review written by Professor Pirkle and Dr Pochapsky and published in 1987 to which I have referred in paragraph [104] above (“the Pirkle and Pochapsky review”). The authors wrote:

“The use of chiral stationary phases (CSPs) for the direct analytical and preparative chromatographic separation of enantiomers has attained increasing prominence in the past decade. The reasons for this popularity are clear: The advantages inherent in any chromatographic separation (rapid analysis and separation of complex mixtures, reproducibility, flexibility) are extended to a hitherto demanding preparative and analytical problem, namely, how does one go about separating enantiomers?”

They also provided a table of racemates separable on CSPs and scattered through the table are some limited examples of preparative separations.

The third was written by Dr Armstrong and published in 1987 (Analytical Chemistry, Vol. 59, No.2, Jan 15, 1987 (“the Armstrong review”). Dr Armstrong observed that the resolution of enantiomers traditionally had been considered one of the more difficult problems in separation science. He suggested the 1980s was proving to be a major turning point because of the tremendous number of new and improved chiral stationary phases and additives which had been introduced, accompanied by a corresponding increase in publications. Further, widespread commercialisation of chiral stationary phases had occurred and a number of extensive theoretical and mechanistic studies involving chiral recognition were beginning worldwide.

He identified the various commercially available chiral stationary phases and their manufacturers, including acetylated beta-cyclodextrin. Finally, he offered this conclusion:

“As the analytical separation of enantiomers becomes more common, the demand for analogous preparative scale separations will grow. Research on preparative scale chiral columns is currently underway in several laboratories. Some preparative and semi-preparative columns are now available. What is needed is a process that can efficiently isolate kilogram and greater quantities of optically pure compounds. Scaling sensitive analytical separations up to preparative size generally results in a significant loss of resolution. Consequently, large selectivity factors are often needed for effective large-scale separations. The number of racemates which have large αs (on currently available CSPs) is much smaller than those with small αs. Thus, many future preparative separations may use specifically designed CSPs that have large selectivity factors, good reproducibility, and good regeneration properties for a given separation. Another possible approach is to use multidimensional separation procedures, which can greatly increase selectivity.

Currently, the LC separation of optical isomers is a popular, highly visible area of research. Rapid advances are being made that will have a significant effect on many areas of science and technology. At some not too distant time, the chromatographic separation of enantiomers will be considered routine and perhaps even uninteresting. This in itself will be a tribute to the success of the many researchers who have worked and are working in what was once considered one of the more difficult areas of separations. ”

The fourth was written by Dr Irving Wainer and published in 1987 (Trends in Analytical Chemistry, Vol.6, No.5, 1987, pages 125-134) (“the Wainer review”). This contains a table of commercially available HPLC CSPs, including all of those to which I have referred, and identifying their manufacturer. In respect of Chiralcel OD, it states this “will be available soon”.

The Wainer review recognised that the large number of available CSPs presented a problem: how to choose the right one. It acknowledged that at that time a simple answer to the question did not exist. One approach was to seek to classify CSPs into classes based upon how they worked and this was approach he sought to develop in his paper. Dr Wainer concluded with the observation that with the rapidly increasing number of commercially available CSPs, it was difficult to decide which column to buy but it was hoped the presentation took a step towards the development of a system which would solve the problem.

Dr Pochapsky accepted that most organic chemists would have been aware of the existence of CSPs by 1988. Indeed, he thought that chiral HPLC was becoming the instrument of choice for enantiomeric separation at the analytical level. The technique had been discussed in papers published in specific chromatographic and more general chemical journals and was also the subject of conference presentations. In March 1988, Dr Collicott attended a course entitled Bradford Analytical Course, Chiral Separations in HPLC at Bradford University attended by some 30 delegates, the majority of whom were representatives from pharmaceutical companies. I should note that this course did contain one session on preparative chiral HPLC. From 31 May to 2 June 1988, the first International Symposium on Chiral Separations took place in Paris at which a number of papers describing the chiral HPLC technique were presented to some 500 delegates.

The position in relation to chiral HPLC for preparative purposes was not so clear. Dr Collicott suggested that the technique was a matter of common general knowledge, at least for analytical chemists. He explained that while his work up to 1988 was mainly directed to analytical procedures, the columns he had at Glaxo at the time could have been used, had he wished, to isolate a small sample of a particular enantiomer in order to study its biological activity. He suggested there was no difficulty in scaling up and that there was no conceptual difference between analytical and preparative separations.

Nevertheless, Dr Collicott accepted that preparative scale chiral HPLC was the subject of research in 1988 and its use was far from routine. Preparative separation was not something he was doing as a matter of course at Glaxo. In fact, he could recall doing so on only one occasion, and then was because he was asked to. The procedure was carried out on an analytical column involving only a few micrograms, which was not suggested to be enough to carry out the biological analyses described in the Patent. Indeed, Glaxo itself had no preparative columns until 1990. Moreover, analytical chemists essentially held themselves out as a service to the medicinal chemists, analysing the materials that the medicinal chemists were producing. When asked about the extent of knowledge of medicinal chemists (or chemists generally) of the preparative use of CSPs, he suggested that “a number” of people were aware of it.

I should also refer to the fact that Dr Collicott assisted with the compilation of a literature survey of chiral separations which was published by the Chromatographic Society in 1991. Over the period 1985 to 1989, it listed 648 publications referring to analytical separations, and only 23 referring to preparative separations, only 16 of which were published before 1988. Of these 16 publications, about six of them refer only to separating microgram amounts or having nothing to do with CSPs at all. All 23 papers relating to preparative separations were published in specialist journals with which the average medicinal chemist would not have been familiar.

Dr Pochapsky accepted that chiral HPLC was originally developed primarily as an analytical tool but really did not know whether or to what extent it had come to be used for preparative purposes in the pharmaceutical industry by 1988. He was taken to the passage in the Pirkle and Pochapsky review set out in paragraph [134] of this judgment and explained that he wrote this as an academic but considered it a wise thing for those in the art to try and do. Similarly, he was taken to observations made by Wilson De Camp in his 1989 article (to which I have referred in paragraph [101] of this judgment) that the successful development of CSPs in the 1970s permitted the routine separation of enantiomers on a preparative basis. Dr Pochapsky accepted that the development of these CSPs altered the perspective of those in the field. But in my judgment the evidence as a whole does not support the sweeping observations made in this paper as to the routine use of CSPs for preparative purposes.

Drawing the threads of all the evidence together, I arrive at the following conclusions. The analytical and medicinal chemists in the team would have had a considerable familiarity with the use of chiral HPLC as an analytical tool. The analytical chemist would have been aware that various categories of CSP existed and were commercially available. I do not accept that he would have known of them all but I think he would have known of the Pirkle columns and perhaps one or two others depending upon his personal experience. He would have known that the technique of analytical chiral HPLC was the subject of considerable ongoing research and that many papers had been published, although it has not been shown that any particular papers were common general knowledge. He would also have known that there was considerable uncertainty as to whether any particular racemate could be separated by this technique and, if so, on which particular CSP.

As to the use of chiral HPLC on a preparative scale, the position was different. I accept that preparative CSPs were commercially available and preparative chiral HPLC was something that the bulk of analytical chemists would have read or heard about. However, it was something that the ordinary skilled analytical chemist would have had little or no practical experience of using. It was certainly not used routinely. On the contrary, in so far as it was known about, it was understood to be a fast moving area of research. Further, it was one which was understood to have particular difficulties associated with it because it was perceived that scaling up could result in a significant loss of resolution. Moreover, it has not been shown that the ordinary skilled medicinal chemist would have had any significant familiarity with the technique at all. In my judgment Lundbeck fairly described the use of chiral HPLC as a preparative tool as being very much in its infancy. It was not something that the ordinary skilled analytical chemist would have had in mind as part of his ‘mental toolkit’ as a way to resolve milligram quantities of enantiomers. It was not something he would have turned to as a matter of course.

As to Chiralcel OD, I am satisfied that this was commercially available at the priority date. However, I do not accept that it was generally known as a CSP for analytical, let alone preparative purposes.

I will deal with the 193 and 884 patents in turn.

This patent was invented by Dr Bogeso and discloses a class of compounds, including citalopram, which are described as having an ability to potentiate 5-HT. Citalopram falls within the class and indeed is named at column 6, lines 4-8. Preparation of tablets of citalopram is described at column 8, lines 7-31. There is no dispute that these tablets contain the racemate. A method of making the compounds is also described. It involves the racemic diol (formula II) and closing the ring under acidic dehydration conditions. This proceeds by an S_N1 reaction via a planar carbonium ion.

The difference between 193 and the claims of the Patent is that 193 discloses only the racemate and methods which will only make racemic compounds. Neither the (+) enantiomer, its properties, nor any means of making them are disclosed by 193.

This patent was published eight years after 193, in 1987. It was also invented by Dr Bogeso. It explains that citalopram has shown great promise as a valuable antidepressant drug with few side effects but that the processes described in 193 for its production possess scale up problems. The invention is an improved method for the production of citalopram which involves a series of steps, the final one being the reaction of the diol with 70% sulphuric acid at 80 degrees C for three hours. The patent says that the cyano-group shows surprising resistance to the rather drastic and prolonged treatment with strong acid in the step of ring closure.

The parties agreed that it would be apparent to the skilled addressee in 1988 that the final step of the process is a ring closure by an intramolecular nucleophilic substitution reaction involving a CH₂ OH group acting as the nucleophile. It proceeds by activation by protonation of the tertiary alcohol and the formation of a carbocation. This would be recognised as an S_N1 type mechanism.

Once again the difference between 884 and the claims of the Patent is that 884 discloses only the racemate and methods which will only make racemic compounds. Neither the (+) enantiomer, its properties, nor any means of making them are disclosed by 884.

It is clear that the pleaded prior art patents do not describe any method for obtaining single enantiomers of citalopram. Assuming the skilled person decided to embark upon the task of trying to obtain the individual enantiomers there was no real dispute between the parties that he would initially choose to attempt to resolve the final molecules. Only if this failed might he then consider alternative methods such as using enantiomerically pure precursors. But here his anxiety would be that subsequent reaction steps might result in the production of an enantiomerically impure final product. As I shall explain, scientists at Lundbeck attempted to resolve citalopram by classical resolution techniques using diastereoisomers, by using new resolving agents, by attempting to resolve derivatives and by using techniques of asymmetric synthesis, all over the period 1980 to 1988. In addition, they attempted to resolve citalopram using analytical chiral HPLC from 1983 to 1987. None of these techniques were successful. The claimants accepted that citalopram is not easily resolved by making diastereoisomers. However, they submitted the skilled addressee would not regard failure in relation to this molecule as the end of road. I accept that this was so. But the work Lundbeck undertook illustrates that the road to which the claimants referred was long and uncertain.

The claimants contended that if it proved impossible to separate citalopram into its enantiomers by conventional means then an obvious alternative route, and an attractive one, was the possibility of resolving a chiral intermediate. They based this case on the 884 specification. Professor Davies accepted that this was something the skilled person would think of, but he would have to convince himself that he could find a synthetic route to the final product that did not compromise the stereochemistry.

There was no dispute that a study of the 884 specification would reveal the diol as a chiral intermediate. As I have indicated, the experts also agreed that it would be apparent that the described method of producing citalopram from the diol is a dehydration reaction which proceeds via a planar carbocation (that is to say an S_N1 reaction) and would thus lead to racemisation:

It is apparent that this is a quite different reaction to those depicted in Reaction Schemes I and II of the Patent and set out in paragraphs [31] to [32] of this judgment. The essential ring closure step in those schemes proceeds in this way:

The claimants did not seek to argue that it was obvious to produce the individual enantiomers of citalopram by the use of Mosher’s reagent, as described in Reaction Scheme I of the Patent. Their case was that it was obvious to carry out Reaction Scheme II, to produce diastereomeric salts, separate them, produce a labile ester and then carry out the ring closure reaction which did not threaten the stereo specificity of the diastereoisomers so separated.

I am satisfied that attempting to form salts with the diol would have been considered routine, albeit unpredictable. Professor Davies accepted that it would not have been possible to predict from a failure to resolve citalopram whether or not any intermediate molecule could be resolved. The Patent lists at page four, lines 50 to 52 various acids which can be used, and all were conventional in 1988. Both experts proceeded on the basis that the teaching of the Patent was correct and that the listed acids would work. Dr Bogeso suggested in his evidence that by 1988 Lundbeck had only limited success in crystallising salts of the diol - a matter to which I return in considering the efforts made by Lundbeck to resolve citalopram. But both the claimants and Lundbeck invited me to disregard this evidence in so far as it was inconsistent with the statement in the Patent. On the assumption the skilled person proceeded down this path he would therefore have found that the diol formed salts which crystallised differentially, permitting the enantiomers of the diol to be separated.

This brings me to the crucial question, namely whether it would have been obvious to the skilled person that the ring of the resolved diol could be closed without risking loss of stereochemistry. For this to be so he must have appreciated that the synthesis could proceed by way of an S_N2 reaction with the tertiary alcohol acting as the nucleophile and the hydroxyl group of the primary alcohol, selectively activated, acting as the leaving group. As the claimants accepted, the skilled person would not blindly go ahead with crystallisation experiments without first satisfying himself that an S_N2 reaction was feasible.

Here the claimants turned to Baldwin’s Rules, which were accepted to be common general knowledge. They are named after two papers written by Professor Sir Jack Baldwin, Professor Davies’s predecessor as the Waynflete Professor of Chemistry at Oxford University.

The first paper (“Baldwin I”) was published in 1976 (J.C.S. Chem. Comm, 1976, pages 734-736). It describes rules which Professor Baldwin had found useful, on an empirical basis, to predict the relative facility of various ring forming reactions. The rules are of a stereochemical nature. Scheme II of Baldwin I depicts tetrahedral schemes (indicated by the suffix “Tet”), all being cases where the carbon atom undergoing the ring-closure reaction has a tetrahedral geometry. Professor Baldwin used the prefix “Exo” where the breaking bond is exocyclic to the smallest formed ring (and “Endo” correspondingly) and a numerical prefix to describe the ring size. The paper also includes trigonal (indicated by the suffix “Trig”) and diagonal schemes (indicated by the suffix “Dig”).

Scheme II depicts seven schemes, five of which are said to be favoured and two of which are said to be disfavoured. The favoured schemes include one described as 5-Exo-Tet. The claimants focussed upon this and contended that a consideration of the diol molecule in the 884 patent shows that the required ring-closure is a 5-Exo-Tet reaction and that the Baldwin Rules show that such a reaction is favoured. This, they said, would have been appreciated by the skilled person who would therefore have been encouraged to proceed.

This was supported by Dr Newton. In his first report he said that the synthesis of structures that contain rings is a basic part of synthetic chemistry and that intra molecular cyclization reactions were well known to be an efficient method of forming cyclic heterocycles, referring to Baldwin I. He elaborated upon this in Annex 2, explaining that Baldwin’s Rules provide a general guide as to whether reactions are favoured or not, and that the reaction whereby the five membered ring system of the Patent is formed is of the favoured 5-Exo-Tet type. He illustrated this by means of a figure which appears to be of a saturated system.

In paragraph 55 of his first report, Dr Newton also referred to the undergraduate text book Heterocyclic Chemistry by Joule and Smith, 1972 in which the authors stated at page 435, when discussing the formation of the saturated heterocyclic rings :

“Many methods have been developed for the synthesis of saturated heterocyclic systems, and often these are specific for the heterocycle concerned. One general method, however, is defined by the following formulae:

<center><IMG alt="Diagram 18" hspace=15 src="1040(image18).png" border=0> </center>

Cyclization takes place by nucleophilic displacement of the functional group at one end of the chain by that at the other end.”

Again, it is to be noted that this is a saturated system.

In paragraph 12 of his second report, Dr Newton reiterated that the skilled person would be aware of the general methods via intra molecular nucleophilic substitution of making saturated heterocyclic rings described in Joule and Smith.

Professor Davies disagreed that this is would be the approach of the skilled person. He pointed out that the ring closure of the Patent involves an unsaturated system on an aromatic ring. This is a significant matter. He explained that Baldwin’s Rules relate to systems in which the required transition state geometry is achievable, and this would have been well understood by the average skilled medicinal chemist in 1988. I accept that this is so. As Professor Davies explained, undergraduate students of organic chemistry are taught to think about organic molecules in three dimensions. They work with three dimensional models. Their studies include transition states and transition state geometries. It is a basic principle in such studies that the most energetically favourable approach for a nucleophile is along the line of centres of the carbon and the leaving group. Baldwin’s Rules reflect these basic concepts. If the nucleophile is not permitted to approach in the most favourable way then the reaction will occur less readily. As Baldwin I itself says, favoured ring closures are those in which the length and nature of the linking chain enables the terminal atoms to achieve the required trajectories to form the final ring bond. The disfavoured cases require severe distortion of bond angles and distances to achieve such trajectories; consequently alternative reaction pathways, if available, will dominate and the desired ring closure will be difficult.

If molecules are in solution then the most favourable configuration can readily occur. However, a different set of considerations apply when the reaction involves a cyclisation of two parts of a molecule which are fixed relative to each other. Here the distinction between a saturated and an unsaturated system is readily apparent and becomes very important. In a saturated system no sections of the ring are planar, as illustrated in the diagram of hexane set out in paragraph [12] of this judgment. By contrast, the diol in issue comprises an aromatic (unsaturated) ring. This ring and the two atoms adjacent to it are in a planar configuration. The ring closure results in the production of a phthalan, as illustrated below:

These are matters which would have been well understood by the skilled addressee. I believe he would not therefore have expected unsaturated and saturated systems necessarily to behave in the same way. Joule and Smith is specific in its terminology. It refers to saturated sytems. Similarly, Scheme II of Baldwin I uses diagrams which are conventionally understood to represent saturated systems. I do not accept that the skilled addressee would simply have taken the description of the 5-Exo-Tet type reaction being favoured as applying equally to saturated and unsaturated systems.

Professor Davies produced a series of diagrams showing models of the relevant molecules. They suggest the most favourable configuration for an S_N2 reaction can be achieved in the case of a saturated system but not in the case of the unsaturated diol of the Patent. I consider they fairly represent how the skilled person would have visualised the molecules.

Professor Davies and Dr Newton were both extensively cross examined upon these diagrams. In addition, Dr Newton produced his own diagram in his third report. This illustrated Dr Newton’s opinion, which he elaborated in cross examination, that the desired configuration can be achieved because the relevant bonds are free to rotate. Professor Davies accepted that the bonds can rotate but maintained the skilled person would have thought that the manner of rotation would be such that the bonds would inevitably spend very little time in the configuration necessary for an S_N2 reaction.

In his third report Dr Newton also referred to a second paper by Professor Baldwin which was published in 1977 (J. Org. Chem, Vol. 42, No. 24, 1977) and referred to as “Baldwin II”. This presents the results of further studies on ring closure by nucleophilic attack of oxygen on conjugated double and triple bonds. Dr Newton suggested Baldwin II made clear that Baldwin’s Rules do not relate only to fully saturated systems. I accept this, to a point. Baldwin II involves a study of the closure of five and six membered rings to test the utility of the rules described in Baldwin I. Specifically, Professor Baldwin investigated 5-Endo- Trig, 5-Exo-Trig, 5-Endo-Dig and 6-Endo-Trig closures and provides examples of favoured ring-closure reactions which do not involve fully saturated systems, notably 5-Exo-Trig and 6-Endo-Trig. But, as Professor Davies explained, in each of these cases the figures make it clear that this is so. By contrast, the reactions depicted in Scheme II of Baldwin I are not so drawn. Further, there is obviously a distinction between these and Tet reactions. The geometry of the approach of the nucleophile is very different.

I have little doubt that the average skilled person would never have engaged in a debate about bond rotation at the level it was conducted in the course of the cross examination before me by Professor Davies and Dr Newton. However, I believe the skilled person would not have proceeded down the diol route unless he was satisfied there was a real prospect of an S_N2 reaction working. This was by no means a “one way street”. There were a number of avenues of research open to him, as the work of Lundbeck demonstrates.

In the light of all the evidence I do not believe the skilled person would have been so satisfied from a consideration of Baldwin I or Baldwin II. Nor would he have been so satisfied from a consideration of the configuration of the relevant interacting atoms. I do not accept it would have been obvious to the skilled person that Baldwin’s Rules could be applied to any system, whether saturated or not. Nor do I accept it would have been obvious that closure of the ring of the diol in issue by an S_N2 reaction would have been favoured. In my judgment the skilled person would have appreciated that saturated and unsaturated systems are materially different and would have believed that the favoured configuration for an S_N2 reaction would not be achieved in the ring closure mechanism required to produce the phthalan of the kind found in the citalopram molecule.

Professor Davies also conducted a literature search in an endeavour to find publications before the priority date showing substitution reactions where a ring-closure had been performed by an S_N2 reaction to produce a phthalan of this kind. He annexed the results to his second report. He considered the nearest paper of potential relevance to be one published in 1952 by Entel et al, (Journal of the American Chemical Society, 1974, Vol. 47, pages 441-444). But both Professor Davies and Dr Newton agreed that this was unsuitable for citalopram due to the harsh conditions employed. The paper in fact reports an attempt to improve the yield of the reaction by using process conditions which would direct the reader towards an S_N1 reaction.

Professor Davies’s search revealed a number of other papers describing the formation of a phthalan structure by ring closure. On reviewing them Professor Davies concluded they all involved the use of an acidic S_N1 ring closure. One of the papers was, however, subjected to closer analysis in the course of cross examination. This was the paper by Cannone et al to which I have referred in paragraph [48]. This analysis revealed it also included an S_N2 reaction of the kind with which these proceedings are concerned. But there was no suggestion that this paper was common general knowledge and the evidence did not establish it would have been revealed by a search at the priority date. In my judgment the paucity of references to the use of an S_N2 reaction in the context of the ring closure in issue is further supportive of its non obviousness.

Professor Davies made two further points, each of which has considerable force. First, it was common general knowledge that tertiary alcohols carbon atoms such as that at the stereogenic centre in the diol intermediate were known to be poor nucleophiles compared to primary alcohols. Reaction Scheme II requires the tertiary alcohol to be a nucleophile and engage in an S_N2 reaction.

Secondly, it was also a matter of common general knowledge that factors that stabilise carbocations favour the S_N1 reaction. A carbocation intermediate of an S_N1 reaction is stabilised by delocalisation of the positive charge over a proximal benzene ring. As Dr Newton himself said in his first report, where there are two benzene rings this effect is even more pronounced. Professor Davies explained that the citalopram molecule contains two phenyl rings and these would similarly be expected to exert a considerable stabilising effect and so promote the S_N1 mechanism.

For all these reasons I have reached the conclusion that the reaction schemes described in the Patent would not have been obvious to the skilled person in 1988. Professor Davies thought it is only with hindsight that it is possible to explain the outcome of a reaction which would otherwise have been unexpected. I agree with that opinion. As I shall explain, this conclusion is supported by the work carried out at Lundbeck.

The starting point for the obviousness case is the contention by the claimants that citalopram can in fact be resolved using chiral HPLC. This was said to be evidenced by four post-priority date documents. The first is an article by Rochat et al (Chirality, 1995, 7, 389-395) referred as “Rochat I”. The second is another article by Rochat et al (Therapeutic Drug Monitoring, 1995, June, 17(3), 273-279) referred to as “Rochat II”. The third is an article by Carlsson and Norlander (Chromatographia, 2001, 53, March, No. 5/6, 266-272) referred to as “Carlsson” and the last is Lundbeck’s PCT application WO 03/006449.

Rochat I was said to demonstrate separation using Chiralcel OD, and a “partial separation” on acetylated beta-cyclodextrin. Rochat II and Carlsson were said to demonstrate separation on acetylated beta-cyclodextrin. Lundbeck’s PCT application was the only publication of the four to demonstrate separation on a preparative scale and it did so in relation to the bromo derivative of citalopram on Chiralcel OD. In the end, the claimants not surprisingly accepted that their best case was in relation to Chiralcel OD.

The claimants case on obviousness then proceeded through a number of steps as follows:

Dr Pochapsky was cross-examined with great skill through each of these steps. The result was a powerful case that the invention was obvious. However, not without some hesitation, I have reached the conclusion that it must be rejected. I believe that this is one of those cases where each step seems very simple and logical with the benefit of hindsight. This, it must be remembered, was a fast moving field and by the mid 1990s many techniques were routine which were still very much at an experimental stage in 1988. The words of Lord Diplock in Technograph Printed Circuits v Mills & Rockley (Electronics) [1972] RPC 346 at 362 are particularly apposite:

“The cross-examination of the respondents’ expert followed with customary skill the familiar “step by step” course. I do not find it persuasive. Once an invention has been made it is generally possible to postulate a combination of steps by which the inventor might have arrived at the invention that he claims in his specification if he started from something that was already known. But it is only because the invention has been made and has proved successful that it is possible to postulate from what starting point and by what particular combination of steps the inventor could have arrived at his invention. It may be that taken in isolation none of the steps which it is now possible to postulate, if taken in isolation, appears to call for any inventive ingenuity. It is improbable that this reconstruction a posteriori represents the mental process by which the inventor in fact arrived at his invention, but even if it were, inventive ingenuity lay in perceiving that the final result which it was the object of the inventor to achieve was attainable from the particular starting point and in his selection of the particular combination of steps which would lead to that result. ”

The starting point is the common general knowledge in relation to separations. I have addressed this in paragraphs [114] to [147] of this judgment. A variety of techniques were commonly used, as I have explained. However those techniques did not include preparative chiral HPLC. Chiral HPLC was routinely used as an analytical tool and this was a matter of common general knowledge. But it was not routinely used to separate enantiomers for preparative purposes, that is to say, for the purpose of obtaining sufficient material to carry out biological testing and to make pharmaceutical compositions. Nor was it a technique which was generally known or accepted to be useful by medicinal chemists. This is important because, as I have found, the medicinal chemists would turn to the analytical chemists to carry out analyses as and when necessary. For these reasons I am not satisfied that it would have occurred to the skilled team to consider using preparative chiral HPLC, save as part of a research programme.

Nevertheless, and on the assumption the skilled team did consider progressing down this path, there was a difference of opinion between Dr Collicott and Dr Pochapsky as to whether citalopram would have been recognised as a reasonably promising candidate for resolution, even on an analytical scale. Dr Collicott considered that citalopram has three groups rigidly attached to the chiral centre and these made the molecule a reasonably promising candidate for resolution because they would maximise the difference between the two enantiomers and hence the likelihood they would interact differently with the CSP. Dr Pochapsky disagreed. I found Dr Collicott’s opinion more reasoned than that of Dr Pochapsky. However, both must be seen in context. Drs Clark and Fell, both leaders in this field and organisers of the Bradford conference, wrote in the 1991 in the Overview to the survey of The Chromatographic Society to which I have referred in paragraph [143]:

“Notwithstanding the wide variety of chiral columns available, no satisfactory rationale has yet been developed for selecting a particular packing material for a specific problem. Given the relatively high cost of columns, there is an urgent need to establish a suitable set of guidelines and rules for designing chiral separations in LC. Here it is most likely that an expert systems approach may prove fruitful, when the database of expertise has been sufficiently well developed. ”

Dr Collicott did not agree with this opinion, at least not entirely. He accepted that it was not possible to take a particular compound and say there was an obvious column to resolve it on but maintained there were guidelines that would lead the skilled person towards classes of phase that would narrow the choice down to a sensible number of columns. He said that one had to produce a short list of columns somehow and for that purpose one had to go out and seek guidance and advice from the literature. No such literature was, however, pleaded or identified as being part of the common general knowledge.

The difficulty, even on the analytical scale, was that it was recognised that a very small change (even as small as a single atom) might prevent a racemate from being resolvable on a particular column. Dr Okamoto (who developed the polysaccharide CSPs) wrote in an article published in 1987 (Chemtech, March 1977, pages 176-181) that chiral recognition mechanisms of CSPs, particularly polymeric CSPs, were still obscure and would have to be clarified in order to design more effective CSPs. This was a view with which Dr Collicott agreed, subject to the qualification that it was “not a complete mystery”.

Further, Dr Armstrong said in his 1987 review under the heading “Future directions in chiral separations research” :

“A variety of new, modified, and improved stationary phases will continue to be introduced over the next few years. Some of these will be commercialised. This will greatly increase the number and type of enantiomers amenable to rapid LC determination. Rationales will develop by which one can quickly choose the one or two columns most likely to separate any of a great many enantiomers.”

By inference, Dr Armstrong was there indicating that there was no such rationale at the time, a view to which Dr Collicott did not offer any reasoned dissent. I have set out the conclusions expressed by Dr Armstrong in this review in paragraph [137] above. They emphasise two further points. First, scaling analytical separations up to preparative size generally results in a significant loss of resolution, a matter with which Dr Pochapsky agreed. Secondly, Dr Armstrong considered that chromatographic separation of optical isomers was an area of research, albeit a highly visible one. Dr Wainer evidently agreed. He thought there was no simple answer to the question how to choose the right CSP.

Assuming the skilled person in fact proceeded then Dr Pochapsky and Dr Collicott thought that he would look at review articles including, in particular, the Pirkle and Pochapsky, Wainer, Däppen and Armstrong reviews. These provide some guidance as to which columns were commercially available and which types of column potentially allowed preparative work. Both experts agreed that the Pirkle and the ligand exchange columns would be considered unsuitable but this still left the three classes of polysaccharide, inclusion complex and protein based CSPs.

Both experts thought that protein columns would be less favoured but interestingly this was inconsistent with the Wainer review which points in the direction of a protein type column for a compound such as citalopram.

As for the inclusion complex CSPs, the microcrystalline cellulose triacetate, polymethacrylate and cyclodextrin CSPs would all have been possible ones to try. Narrowing down to the cyclodextrins, both experts considered that beta cyclodextrin was the most promising in the light of the size of the citalopram molecule. But I have no satisfactory evidence this would separate citalopram on a preparative scale. As for the Daicel phases, Dr Collicott thought that he would have been encouraged to try Chiralcel OD on the basis of structural considerations. Dr Pochapsky did not share his view. I was left with the clear impression that Dr Collicott’s process of analysis was not something which would have been obvious to the ordinary skilled analytical chemist, let alone the ordinary skilled medicinal chemist. As I indicated at the outset of this judgment, he approached the problem with knowledge of the solution and his reasoning was based upon the contents of a promotional brochure which was not shown to be common general knowledge or one which would have been provided upon an enquiry to Daicel. So the skilled person would have been left with a programme of testing to see which, if any, column might work on an analytical and then a preparative scale.

To summarise, there was no clear understanding of the chiral recognition mechanism of CSPs and there was no generally understood or accepted rationale as to how to choose a small number of columns most likely to separate the enantiomers of a particular racemate. In the result the skilled person might choose a number of columns, none of which worked. Anyone familiar with CSPs would have known this was a real possibility.

As I have mentioned, scaling up also posed a problem because of the loss of resolution. Moreover, whether any particular separation is successful or not may depend upon a number of factors other than the nature of the CSP, such as impurities, ratios of solvents, temperatures and equilibration times. Dr Pochapsky indicated that he would not have attempted a preparative scale separation unless he had a separation factor of at least two on the analytical scale. Dr Collicott accepted that there was a problem but maintained that its extent really depended on the quality of the stationary phase. I conclude that anyone embarking upon this course in 1988 would have recognised that a preparative scale separation involved yet further problems and uncertainty.

An alternative course suggested by Dr Collicott was to carry out a preparative separation on an analytical scale column. However, Dr Pochapsky gave evidence that it would have taken an enormously long time to obtain the milligram quantities of material necessary for testing using the conditions described in the Rochat I paper. Dr Collicott suggested the time would be rather less than that proposed by Dr Pochapsky. I am simply not satisfied on the evidence that this was a course which would have occurred to the skilled person as a sensible one to follow.

In conclusion, I do not believe it was obvious to resolve citalopram on a preparative scale using chiral HPLC in 1988. It was a rapidly evolving field. The ordinary skilled analytical chemist would have had no practical experience of preparative chiral HPLC and the ordinary skilled medicinal chemist would probably not have heard of it. The team would have been faced with a research programme with an uncertain outcome.

I heard evidence from two witnesses on this issue from Lundbeck. Dr Bøgesø is now the Vice-President for Lundbeck Research DK and is one of the named inventors of the Patent. In 1971, he began work with Lundbeck as a research chemist in the Medicinal Chemistry Department and has been continuously employed by Lundbeck since that time. He was subjected to heavy criticism by the claimants on the basis that he came to court with Lundbeck’s party line well in mind. In support of this submission the claimants relied upon the fact that Dr Bøgesø said in the course of his cross examination that the teaching of the Patent was wrong in that the only optical acid which works in Reaction Scheme II is di(p-toluoyl) tartaric acid. As I have indicated, both sides have disclaimed any reliance upon this evidence in support of their substantive cases. Nevertheless, the claimants submitted there was a straight conflict between the teaching of the Patent and the evidence given by Dr Bøgesø and that in the circumstances I should not place any reliance upon the evidence of Dr Bøgesø on the issue of obviousness. The claimants developed this submission by reference to Dr Bøgesø’s second witness statement which, it was said, was incomplete in that it implied that although Lundbeck anticipated having difficulty crystallising the intermediate diol with optically-active acids, it was in fact able to do so. I accept that there is an inconsistency between the evidence of Dr Bøgesø and the teaching of the Patent. But I do not accept that the evidence Dr Bøgesø gave in his witness statement was misleading. It says, as he elaborated in cross examination, that in fact the diol is not easy to crystallise with optically active or other acids. Moreover, Dr Bøgesø dealt with the matter quite openly in cross-examination. I therefore feel able to rely upon his evidence.

The other witness who gave evidence on behalf of Lundbeck on this issue was Dr Gundertofte. As the claimants accepted, he was a straightforward and honest witness about whom no criticism could be made.

Dr Bøgesø began to consider the possibility of separating the enantiomers of citalopram in about 1980. He was interested in this from an academic and personal point of view and because he thought it would be useful for Lundbeck to have data about the individual enantiomers.

Dr Bøgesø began with the classical method for attempting to resolve racemic amines by reacting citalopram with a chiral organic acid in an attempt to make diastereomeric salts which might then be separated on the basis of their different physical and chemical properties. Citalopram had shown a low propensity to form any salts. The oxalate and hydrobromide salts were the only ones Dr Bøgesø had managed to make, and neither was chiral. Nevertheless, Dr Bøgesø attempted to make diastereomeric salts using the (+) and (-) forms of tartaric acid, dibenzoyl tartaric acid, ditoluoyl tartaric acid, mandelic acid and (+) camphorsulfonic acid. He found that only the latter produced crystals with citalopram. But he was unable to find conditions in which these crystals selectively precipitated.

In all these experiments he used a variety of techniques in order to try to promote crystallisation. These included using a range of solvents, modifying the molar ratio of the optical acid, leaving solutions and oils of the reaction mixtures standing, either in the fridge or the freezer or at ambient temperature for long periods, scratching the glass reaction vessels, using anti-solvents and evaporating solvents off on a rotary evaporator.

In December 1983, Dr Bøgesø attended a course called “Strategies for Optical Resolution” run by Professor Collet, a leading expert in the resolution of enantiomers. This brought to his attention a new resolving agent, bis naphthyl phosphoric acid (“BNPPA”) which does not contain a chiral carbon but does exhibit planar chirality. Dr Bøgesø synthesised some BNPPA himself but was unsuccessful in his attempts to use it to resolve citalopram.

Dr Bøgesø also continuously reviewed the literature to try and find successful resolutions which had some similarities with citalopram and which he might adapt. He found a book entitled Optical Resolution Procedures for Chemical Compounds - Volume 1, Amines and related compounds, edited by Paul Newman from Manhattan College. This was a collection of hundreds of examples of successful resolutions taken from scientific papers. He also regularly exchanged ideas with other chemists in the department. He used a number of unusual chiral acids as a result, including N-acetyl-L-leucine. Again these attempts were unsuccessful.

By about 1984, it was becoming clear that direct resolution of citalopram was proving extremely difficult. Members of Dr Bøgesø’s group discussed the possibility of resolving the diol as one of a number of alternative strategies. It was quickly dismissed because they did not believe that it would work. The principal reason for this conclusion was that they believed that even if they were able successfully to resolve the diol, the subsequent step of conversion to the final citalopram molecule in acid conditions would result in full or partial racemisation. They were also conscious of further difficulties. The process they were actually using to make citalopram did not involve isolation of the diol in a pure form. Further, they did not believe that the diol would prove any easier to resolve by way of crystallisation than citalopram. They observed that the structure of the diol, with its hydroxyl groups, potentially offered a method of resolving the enantiomers by creating an ester on one of the groups by using an optically pure chiral acid chloride and then resolving the diastereoisomers so produced. However, they were concerned as to whether they could create an ester on only one of the hydroxyl groups. Had esterification of both occurred it would have resulted in a mixture of compounds. A further potential problem was the stability of any such ester because it had to be sufficiently stable to separate the diastereiosomers by chromatography or crystallisation but be sufficiently labile that it could be removed under conditions that would not induce racemisation.

Lundbeck therefore rejected the option of trying to proceed by way of the diol intermediate and they looked at yet other ways to resolve citalopram.

The next route Dr Bøgesø followed was to try to produce derivatives of citalopram, to resolve them and then convert them back into the single enantiomers of citalopram.

He thought the functional group onto which such substitutions could most easily be made was the amine group at the end of the side chain. He could use established amine chemistry to make the substitutions, attempt to resolve the substituted compounds and then convert the group back to the correct amine once resolution had been achieved.

The first derivative selected was desmethyl citalopram, a known metabolite of citalopram, in which one of the methyl groups on the amine has been removed. A number of attempts were made to resolve this molecule but Dr Bøgesø was not able to form crystals.

Another derivative tried was an N-benzyl derivative in which one of the methyl groups on the amine group was replaced by a benzyl group. Once again Dr Bøgesø could not form crystals. Many standard techniques of the kind to which I have already referred for inducing crystallisation were tried, but all were unsuccessful.

At some point between August 1985 and September 1986, Dr Bøgesø concluded in a note that it was very unlikely that citalopram or N-derivatives could be resolved via salts with optically active acids. He formed the view that some sort of asymmetric synthesis of the enantiomers had to be considered. He also thought about chiral phase HPLC, although work on this was already being undertaken by Mr Gundertofte, as I shall explain.

Dr Bøgesø theorised that previous attempts might have been unsuccessful because the amine group on the citalopram molecule was located too far away from the chiral carbon. He thought that if the two chiral centres were closer together each might have more influence on the other, thereby enhancing any difference in physico-chemical properties of the various diastereoisomers.

The first to be tried was a 5-bromo-carboxylic acid derivative. Although initiated by Dr Bøgesø, the work was carried out by Dr Perregaard. Despite considerable efforts made to utilise the bromo-carboxylic acid derivative over several months, it was found that the carboxylic acid group, when situated on the central carbon, tended to be sterically hindered by the proximity of the large phenyl groups around it. They were therefore unable to get many reagents to react with it to produce the closely associated chiral carbons they needed, and the strategy was ultimately unsuccessful.

Yet another route tried by Dr Perregaard was to react the main part of the citalopram molecule with epichlorohydrin to produce a side chain with a second chiral carbon as part of an epoxide ring. Ultimately he was unable to get the reaction to take place on the intended carbon atom, probably due to steric hindrance of that position in the molecule.

From 1983, Lundbeck tried using analytical scale chiral HPLC. Dr Gundertofte was responsible for this work. He tried a variety of columns including triacetate cellulose, a Pirkle-type, immobilised protein and beta-cyclodextrin columns. None were successful, despite many runs using different combinations of solvents, temperature, pH and pressure in an attempt to optimise conditions.

Towards the end of 1987, Dr Bøgesø and Dr Perregaard revisited the possibility of the racemic diol intermediate. They remained concerned about the low likelihood of success but, after a number of attempts, Dr Perregaard was able to make an ester with Mosher’s Acid chloride, an agent normally used as an analytical tool. With the benefit of this work under their belts the team began work to prepare diastereomeric salts of the diol with toluoyl tartaric acid followed by the same basic ring-closure mechanism observed with the ester. They found they had again produced the pure enantiomers of citalopram.

In my judgment this work is supportive of the conclusions I have reached on the basis of the expert evidence. This is something of an unusual case in that researchers at Lundbeck were equipped with everything said to render the reaction schemes described in the Patent obvious. They knew of citalopram and its activity and were free to work with it. They were interested in identifying and testing its enantiomers as early as 1980. Dr Bøgesø embarked upon a lengthy programme of research. It certainly was not obvious to him that the enantiomers of citalopram could be resolved by working with the diol in the way described in Patent. Later in the programme, Dr Gundertofte attempted to resolve citalopram using analytical scale chiral HPLC. A variety of columns were used including immobilised protein columns and beta-cyclodextrin. None were successful. Of course, Chiralcel OD did not become available until shortly before the priority date. But again, it was not something Lundbeck tried. In my judgment this is one of those cases where failed attempts to solve the problem addressed by the Patent are further evidence of non obviousness.

This issue raises two separate questions. The first is whether or not escitalopram has a superior efficacy which was unexpected. The second is whether or not this is relevant in law in the circumstances of this case.

Professor Montgomery carried out a review of escitalopram for the Current Medicine Group, the second edition of which he re-wrote in 2005. He concluded that in vitro and in vivo studies had shown escitalopram to be a more potent SSRI than citalopram and that the (-) enantiomer is practically devoid of activity. This, of course, is described in the Patent. However, he also explained in his review and his evidence to me that a further surprising finding was that studies with some animal models had shown an earlier response to escitalopram compared to citalopram. Further, the role of the (-) enantiomer had been investigated in a series of preclinical studies, which had shown that that the (-) enantiomer reduces or delays the effect of the (+) enantiomer.

Data supporting these conclusions came from experiments carried out on rats by Mork et al (Neuropharmacology, 2003; 45:167-173). Surprisingly, when 2mg/kg escitalopram was compared to 4mg/kg citalopram, escitalopram elicited an increase in brain serotonin levels that was about twice that with citalopram. Further, when escitalopram was administered together with twice or four times as much (-) citalopram, the (-) enantiomer significantly inhibited the (+) enantiomer induced increase in serotonin level in a dose-dependent manner. Professor Montgomery considered that these ratios of (+) to (-) enantiomers are clinically relevant because, in humans, the two enantiomers are metabolised at different rates, the (-) enantiomer being metabolised more slowly that the (+) enantiomer. The results showed, in his opinion, that escitalopram alone is more effective at increasing extra-cellular serotonin levels in the brain than an equivalent dose of citalopram, and that the (-) enantiomer counteracts the activity of the (+) enantiomer in vivo. Thus escitalopram is not only more potent than citalopram, but also more efficacious. I did not understand there to be any serious challenge to this analysis of the work by Mork, and I accept it.

Professor Montgomery also considered the binding mechanism of escitalopram. He noted that there were at least two binding sites for citalopram on the serotonin transporter: a primary high affinity binding site that mediates the inhibition of serotonin reuptake and a secondary low affinity allosteric site. He explained that work published in 2005 had shown that escitalopram increased the stabilisation of the binding to the serotonin transporter via the allosteric mechanism compared to other SSRIs and venlafaxine, and this went a long way to explaining the obvious clinical superiority of escitalopram as an antidepressant and anxiolytic. Once again, I accept this analysis of the binding action of escitalopram. I turn now to consider the clinical data said to show the superior efficacy of escitalopram.

The story begins with four studies carried out by Burke et al (J. Clin. Psychiatry 2002; 63:331-336), Wade et al (Int. Clin. Psychopharmacol. 2002; 17:95-102), Lepola et al (Int. Clin. Psychopharmacol 2003; 18:211-217) and Forest Laboratories (unpublished). Patients in all studies suffered from moderate to severe MDD with a severity of 22 or more on the MADRS scale. The primary measure of efficacy in each study was the mean change of total score on the MADRS scale from baseline to last assessment. Patients were assessed over a period of weeks. I am satisfied on the evidence that a difference of two points on the MADRS scale is of clinical significance.

The Burke study compared citalopram 40mg, escitalopram 10mg and 20mg and placebo. The Lepola study compared citalopram 20-40mg, escitalopram 10-20mg and placebo. The Wade study compared 10mg escitalopram with placebo but did not include citalopram as a comparison. The study by Forest Laboratories compared citalopram and escitalopram but was not successful in that neither showed a significant result over placebo.

The results from these papers were then pooled and the subject of further analysis in a series of later papers. These do not produce an entirely consistent picture. Nevertheless, I believe that overall they do demonstrate a statistically significant better response by patients to escitalopram over citalopram, as Professor Montgomery maintained. It is sufficient to refer to two.

Lepola et al (Int. Clin. Psychopharmacol. 2004; 19:149-155) demonstrated severely depressed patients showed a 2.67 point improvement over citalopram on the MADRS scale. The authors also found evidence of earlier onset of action.

Llorca et al (Int. J. Clin. Pract. March 2005; 59, 3, 268-275) focused on severely depressed patients. The improvement of escitalopram over citalopram was 3.5 points compared to citalopram over placebo of 1.3 points.

These studies were not specifically designed to test the efficacy of escitalopram over citalopram. Professor Montgomery therefore drew attention to another study by Moore et al (Int. Clin. Psychopharmacol. 2005; 20, 131-137). This was a multi centre study of 280 patients with MDD. It did not establish any significant difference between the efficacy of citalopram and escitalopram at either the one or four week time points. However, it did show a significant difference in efficacy using the MADRS scale at eight weeks. The study also established that the responder rates were 76.1% for escitalopram and 63.3% for citalopram, representing a significantly important and clinically relevant advantage of 14.8%.

Finally, I would refer to a study by Lamb and Anderson (Pharmacopsychiatry 2006; 39:180-184) This demonstrates that as the severity of the patients’ MDD symptoms increases, the efficacy of citalopram levels off, while the efficacy of escitalopram continues to increase.

Professor Montgomery maintained on the basis of his clinical experience and consideration of many studies, including those to which I have referred, that escitalopram has superior efficacy compared with citalopram, and this was unexpected.

Professor Reid was altogether more dubious about the advantages of escitalopram. He pointed out that the NHS has expressed the view that the evidence to support the claim that escitalopram has improved efficacy over citalopram in the treatment of depression is not compelling and that routine use of escitalopram ahead of citalopram for the treatment of depression in primary care is not justified. He considered that the reported superior benefits of escitalopram are marginal at best, have not been consistently demonstrated and have been observed only in relatively small scale clinical trials.

I have reached the conclusion that Professor Montgomery’s evidence on these points is to be preferred. Unlike Professor Reid, he has had extensive clinical experience of both escitalopram and citalopram. Further, the studies to which I have referred establish a general picture which is consistent with the clinical experience of Professor Montgomery. As to the NHS, I do not know which papers it has considered or the criteria it has applied. I believe the weight of the evidence is that escitalopram has been shown to have significant unexpected benefits.

As I have found, escitalopram does have unexpected benefits which could not have been predicted from a simple consideration that it was likely that one enantiomer was primarily responsible for the action of citalopram.

But there is one other material matter I must take into consideration. There is no suggestion of such surprising benefits in the Patent itself, as Professor Montgomery accepted. It reports that the inventors have found that the (-) enantiomer is essentially ineffective and that the (+) enantiomer is about twice as effective as the racemate – in fact it needs slightly more than half a unit dose of the (+) racemate to produce the same effect as a unit dose of the racemate. This is what the skilled person would expect of a racemate where the activity lies in one of the two enantiomers. The data in the Patent does not indicate the greater efficacy found in practice.

There is no doubt that a surprising technical benefit can be regarded as an indication of inventive step. But whether it does so or not depends upon all the circumstances. For example, if it is found that the claimed invention is obvious for one purpose then it is not saved because it is found to have added benefits: Hallen v Brabantia [1991] RPC 195 at 216.

Likewise, I do not believe it is permissible to take into account surprising technical benefits which are not described or foreshadowed in the specification. In Richardson-Vicks Inc.’s Patent [1995] RPC 568, the alleged invention related to mixtures of a non narcotic analgesic with one or more of an antihistamine, a decongestant, an expectorant and a cough suppressant, to be sold for the treatment of coughs and flu. The patentee argued that the combination had an unexpected advantage resulting from synergy between its components. Jacob J said at 581:

“Formally I think the experiments were irrelevant. Curiously Mr Thorley, for the purpose of his argument before me, accepted this. Whether or not there was synergy demonstrated by experiments conducted after the date of the patent cannot help show obviousness or non-obviousness. Nor can the amended claim be better if only the components of the amended claim (as opposed to the unamended claim) can be shown to demonstrate synergy. The patent does not draw any such distinction and it would be quite wrong for later acquired knowledge to be used to justify the amended claim.”

In Glaxo Group Ltd’s Patent [2004] EWHC 477 (Ch); [2004] RPC 43, Pumfrey J took much the same line. The patent related to an inhaler that administered simultaneous doses of two drugs, salmeterol and fluticasone propionate. The specification taught that they should be administered together but gave no data to support any assertion that they were particularly compatible or complementary in their activity. The patentee sought to rely upon evidence to show synergy between the elements of the combination but ultimately abandoned any attempt to rely upon synergy per se. The judge explained at [113]:

“It is sometimes thought that a patent may be saved from a finding of obviousness if a combination otherwise obvious has some unexpected advantage, and, in particular, an advantage caused by an unpredictable co-operation between the elements of the combination. I do not consider that such an approach is in general justified. There is a limited class of cases in which the patentee has identified an advantageous feature possessed by some members only of a class otherwise old or obvious, has described the advantageous effect in his specification and has limited his claim to the members of the class possessing this advantageous feature. Such a claim may be justified on the basis of what is called selection. Unexpected bonus effects not described in the specification cannot form the basis for a valid claim of this kind. ”

After referring to the passage from Richardson-Vicks cited above, he continued at [114]:

“If a synergistic effect is to be relied on, it must be possessed by everything covered by the claim, and it must be described in the specification. No effect is described in the present specification that is not the natural prediction from the properties of the two components of the combination.”

Both of these cases were concerned with synergy. But it seems to me that the logic behind them is not limited to such cases. A patentee cannot seek to bolster the inventive nature of his monopoly by relying on a discovery which he had not made at the time of the patent. That is the position here. At the date of the Patent, Lundbeck had not found that escitalopram was more efficacious or was effective in treating more patients than citalopram. Those discoveries were not made until some time later. They are nowhere hinted at in the specification and could not have been predicted from what is described. In these circumstances I do not believe that it is legitimate for Lundbeck to rely upon them in support of the alleged invention.

Lundbeck also relied upon commercial success to support inventiveness. It contended:

Commercial success can support a case of inventiveness. As Tomlin J said in Parkes v Cocker (1929) 46 RPC 241 at 248:

“The truth is that, when once it had been found …. that the problem had waited solution for many years, and that the device is in fact novel and superior to what had gone before, and has been widely used, and used in preference to alternative devices, it is, I think, practically impossible to say there is not present that scintilla of invention necessary to support the patent.”

The essential elements of the argument are that there has been a “long felt want”, that is to say a known problem in need of a solution, that the invention has achieved striking commercial success and that the monopoly is limited to embodiments which have the qualities giving rise to the success. Then, so the argument goes, the invention cannot be obvious or it would have been made before.

Although these are the essential elements of the argument they must be approached with some caution. The commercial success may be attributable to factors other than the invention, for example effective advertising, better workmanship or more attractive presentation. Or it may be that those in the field were not aware of the prior art said to render the invention obvious or were unable to exploit any developments of it. Careful consideration must be given to all the circumstances, many of which were identified by Laddie J in Haberman v Jackel [1999] FSR 683.

Evidence was given on this issue by Mr Kjerulf-Jensen, the Senior Product Director responsible for escitalopram at Lundbeck. Escitalopram received its first marketing approvals in Sweden and Switzerland in December 2001 and was first launched in March 2002 in Switzerland and then in May 2002 in Sweden. By the end of 2003, escitalopram had been launched in 52 countries and by the end of 2004 in 70 countries.

Between the first launch of citalopram in 1989 and the launch of escitalopram in 2002, the market for antidepressants, and particularly SSRIs, had radically expanded. In 1989, SSRIs were a new class of antidepressants and citalopram was one of the first to market. In 2002, however, there were already a number of SSRIs on the market, including fluoxetine, citalopram, paroxetine and sertraline. There were also other successful antidepressants such as venlafaxine. Of these, generic versions of fluoxetine were available in the major markets worldwide, and generic versions of sertraline, paroxetine and citalopram were available in all major markets except the US.

Despite the presence of these other drugs on the market, sales of escitalopram grew rapidly. Four years after its launch, by the third quarter 2006, escitalopram was the largest selling branded antidepressant by volume and the second largest by value in the Northern American market, the European market, and worldwide. It accounts for 60% of Lundbeck’s turnover and in 2006 had worldwide sales of in excess of 500 million Euros per quarter.

Escitalopram has evidently been a substantial commercial success and it is tempting to say that any product which has enjoyed success on this scale cannot have been obvious. However, and as I have explained, commercial success does not of itself support the case for inventiveness. In evaluating its relevance in the present case I think the following considerations are of importance.

First, I do not believe it indicates that third parties had not appreciated there were benefits to be gained from separating the (-) enantiomer from the (+) enantiomer of citalopram as Lundbeck contended (although I do accept that the superior therapeutic qualities which escitalopram has subsequently been shown to possess were unexpected). It must be remembered that citalopram was protected by patent rights from the date of its first publication in the 1970s to a date after the priority date of the Patent. So third parties were not in a position to exploit improvements to the molecule. Indeed, Lundbeck was asked for further details of its case of commercial success and made clear that it did not allege that it was normal or common practice for third parties to concentrate attention on improvements that might be made to antidepressants covered by patent protection. Moreover, and as I have explained in paragraph [103] of this judgment, the proprietors of the other antidepressants on the market (fluoxetine, duloxetine, sertraline, paroxetine and venlafaxine) did take steps to resolve and test the enantiomers of their respective products. Indeed, Lundbeck did so too through the efforts of Dr Bøgesø and Mr Gundertofte.

Second, Lundbeck itself maintains that the success of escitalopram is due its superior therapeutic qualities, particularly its greater efficacy and that more patients respond to treatment. But none of this is disclosed in the Patent. For the reasons I have given I do not believe it is permissible for Lundbeck to rely in support of inventiveness upon the technical merit of discoveries about the surprising efficacy of escitalopram which it had not made at the date of the Patent. Likewise it seems to me that commercial success which is attributable to these discoveries cannot be supportive of inventiveness either.

Finally, I must deal with advertising and promotion. It is clear on the evidence that escitalopram has been supported by extensive advertising and promotion. Worldwide marketing spend for the third quarter of 2006 was about 150 million Euros. By contrast branded duloxetine achieved sales of approximately 255 million Euros supported by a marketing spend in the region of 132 million Euros and venlafaxine achieved sales of approximately 700 million Euros supported by a marketing spend in the region of 101 million Euros.

Mr Kjerulf-Jensen explained that because of the prominence of escitalopram and the fact that Lundbeck is a comparatively small pharmaceutical company having only few products, Lundbeck’s sales representatives normally focus upon escitalopram almost exclusively during their visits to practitioners which makes their marketing costs relatively high compared to companies that present a more extensive portfolio of products, and can therefore spread their marketing costs. I accept that is so but it also means that escitalopram has been marketed more intensively. Further, and importantly, representatives have stressed the technical benefits of the product, including the claim that it has an earlier onset of action.

This advertising has obviously played a considerable part in the commercial success of the escitalopram but precisely how much a part is difficult to assess. I have very little evidence to assist me one way or the other. Lundbeck accepts that the US marketing spend is higher than for duloxetine and venlafaxine but says that in Europe it is about the same. Sales of branded venlafaxine in Europe are, however, very much larger than those of escitalopram when assessed by value but about the same when assessed by volume. Overall I have been left with the impression that it is not easy to distinguish between escitalopram and venlafaxine, but both have been more successful than duloxetine. I think it also important to keep in mind that sales of escitalopram have been achieved in the face of the generic citalopram. Doing the best I can, I conclude that advertising has undoubtedly been important but the spend cannot be said to be out of line with other branded pharmaceuticals in this sector. However I believe the advertising has been particularly effective because Lundbeck has been able to point to the clinical benefits of the product.

To summarise, I accept that there was at the date of the Patent a long felt want for improved antidepressants and that escitalopram has been commercially successful. However, that success is attributable to discoveries which Lundbeck had not made at that time. Further, the molecule was protected by patent rights and so it does not indicate that third parties had not appreciated the benefits of separating the enantiomers. In all the circumstances this is not a case in which commercial success assists in determining the question of obviousness.

The claimants contended the Patent is insufficient because citalopram was an obvious target for resolution. If, contrary to their primary submission, any technical contribution has been made by the invention then it lies in finding a way to carry out that resolution. However, claims 1 and 3 claim the (+) enantiomer however obtained. They are therefore too broad. They claim an obvious goal.

The allegation is founded on the principles explained by the House of Lords in Biogen v Medeva [1977] RPC 1. The claim of the patent in issue in that case was for an artificially constructed molecule of DNA carrying a genetic code which, when introduced into a suitable host cell, would cause that cell to make antigens of the hepatitis B virus (“HBV”). In 1978, the priority date, it was known that the infective agent responsible for causing hepatitis B was a small particle called the Dane particle consisting of a circular molecule of DNA in a protein core surrounded by a protein surface. It appeared to have at least two antigens, one at its core (HBcAg) and one at its surface (HBsAg). One way of obtaining these antigens was to purify them from Dane particles taken from infected blood. Another promising way to was to use recombinant DNA technology.

Shortly before the priority date it had been found by Dr Villa-Komaroff that it was possible to express the DNA for the production of eukaryotic proteins (which those of HBV necessarily were) in prokaryotic bacterial host cells. But those working in the HBV field faced a difficulty which Dr Villa-Komaroff did not. Eukaryotic DNA had been found to contain introns – sequences of DNA that do not code for anything. Eukaryotic cells have a mechanism for stripping out the introns as part of the process by which the DNA is transcribed into mRNA. However it was assumed that prokaryotic cells such as bacteria did not have this mechanism, which meant that they might not be able to transcribe the eukaryotic genomic DNA.

One way of addressing this problem was to use artificial cDNA made from mRNA from which the introns had been removed, and this is what Dr Villa-Komaroff had done. But no source of mRNA for HBcAg and HBsAg was available by the priority date. Another way was to sequence the HBV genome, discover where the relevant genes were and see if they did in fact contain introns. However this had not been achieved at the priority date.

It was against this background that Professor Murray made the discovery which led to the patent. He purified genomic DNA taken from Dane particles and cut it into the largest possible fragments so as to give himself the best chance of not cutting the relevant genes, and then he inserted these fragments by established techniques into bacterial host cells. The cells produced HBV antigens. This was surprising because, as he put it, genes from eukaryotic organisms would not normally be expressed in bacteria – because of the presence of introns.

Some six months after the priority date, the DNA for the Dane particle was sequenced. It was found that the genes which coded for HBsAg and HBsAg did not have introns. Once the sequence became known it was accepted no one would follow the path taken by Professor Murray. Instead, enzymes could be chosen to digest the sites closest to the HBsAg and HBsAg genes and then the fragments of genomic DNA could be inserted into bacterial cells by known techniques and expressed to produce the HBV antigens. Biogen accepted that at this point the invention was obvious.

Lord Hoffmann fist considered the question of inventive step. The trial judge had identified it as the idea or decision to express a polypeptide displaying HBV antigen specificity in a suitable host. This, Lord Hoffmann thought, was far too wide. The idea of making HBV antigens by recombinant technology was shared by others. The problem which required invention was a way of doing it. A proper statement of the inventive concept needed to include some express or implied reference to the problem which required invention to overcome. So a more accurate way of stating the inventive concept was to say it was the idea of trying to express unsequenced eukaryotic DNA in a prokaryotic host. This, Lord Hoffmann was prepared to assume, was not obvious.

Lord Hoffmann then turned to the key issue before the House, namely whether the claimed invention was supported by the priority document (referred to as Biogen I) in accordance with section 5(2)(a) of the Patents Act 1977. Biogen I described the work of Professor Murray.

Lord Hoffmann affirmed that for matter to be capable of supporting an invention within the meaning of section 5(2)(a) it must contain an enabling disclosure, and that this was also a requirement of sufficiency under section 72(1)(c). He then proceeded to elaborate what the concept of enabling disclosure means at pages 48-49:

In fact the Board in Genentech I/Polypeptide expression was doing no more than apply a principle of patent law which has long been established in the United Kingdom, namely, that the specification must enable the invention to be performed to the full extent of the monopoly claimed. If the invention discloses a principle capable of general application, the claims may be in correspondingly general terms. The patentee need not show that he has proved its application in every individual instance. On the other hand, if the claims include a number of discrete methods or products, the patentee must enable the invention to be performed in respect of each of them.

Thus if the patentee has hit upon a new product which has a beneficial effect but cannot demonstrate that there is a common principle by which that effect will be shared by other products of the same class, he will be entitled to a patent for that product but not for the class, even though some may subsequently turn out to have the same beneficial effect: see May & Baker Ltd. v. Boots Pure Drug Co. Ltd. (1950) 67 R.P.C. 23, 50. On the other hand, if he has disclosed a beneficial property which is common to the class, he will be entitled to a patent for all products of that class (assuming them to be new) even though he has not himself made more than one or two of them.

Since Genentech I/Polypeptide expression the E.P.O. has several times reasserted the well established principles for what amounts to sufficiency of disclosure. In particular, in Exxon/Fuel Oils (T 409/91) [1994] O.J. E.P.O. 653, paragraph 3.3, the Technical Board of Appeal said of the provision in the European Patent Convention equivalent to section 14(5)(c) of the Act:

"Furthermore, Article 84 EPC also requires that the claims must be supported by the description, in other words, it is the definition of the invention in the claims that needs support. In the Board's judgment, this requirement reflects the general legal principle that the extent of the patent monopoly, as defined by the claims, should correspond to the technical contribution to the art in order for it to be supported, or justified."

Lord Hoffmann then turned to consider whether Biogen I contained an enabling disclosure which supported the claims of the patent. He accepted that the teaching of Biogen I enabled the skilled man to produce both HBcAg and HBsAg in any cells and, in that sense, to produce products across the scope of the claim. However that was not the end of the matter. At pages 50-51 he said:

But the fact that the skilled man following the teaching of Biogen 1 would have been able to make HBcAg and HBsAg in bacterial cells, or indeed in any cells, does not conclude the matter. I think that in concentrating upon the question of whether Professor Murray's invention could, so to speak, deliver the goods across the full width of the patent or priority document, the courts and the E.P.O. allowed their attention to be diverted from what seems to me in this particular case the critical issue. It is not whether the claimed invention could deliver the goods, but whether the claims cover other ways in which they might be delivered: ways which owe nothing to the teaching of the patent or any principle which it disclosed.

It will be remembered that in Genentech I/Polypeptide expression the Technical Board spoke of the need for the patent to give protection against other ways of achieving the same effect "in a manner which could not have been envisaged without the invention". This shows that there is more than one way in which the breadth of a claim may exceed the technical contribution to the art embodied in the invention. The patent may claim results which it does not enable, such as making a wide class of products when it enables only one of those products and discloses no principle which would enable others to be made. Or it may claim every way of achieving a result when it enables only one way and it is possible to envisage other ways of achieving that result which make no use of the invention.

One example of an excessive claim of the latter kind is the famous case of O'Reilly v. Morse (1854) 56 U.S. (15 How.) 62 in the Supreme Court of the United States. Samuel Morse was the first person to discover a practical method of electric telegraphy and took out a patent in which he claimed any use of electricity for "making or printing intelligible characters, signs, or letter, at any distances". The Supreme Court rejected the claim as too broad. Professor Chisum, in his book on Patents (vol. 1, § 1.03[2] summarises the decision as follows:

"Before Morse's invention, the scientific community saw the possibility of achieving communication by the .galvanic' current but did not know any means of achieving that result. Morse discovered one means and attempted to claim all others."

A similar English case is British United Shoe Machinery Co. Ltd. v. Simon Collier Ltd. (1908) 26 R.P.C. 21. The patentee invented a piece of machinery for automatically trimming the soles of boots and shoes by means of a cam. One of the claims was in general terms for automatic means of trimming soles. Parker J. said, at pages 49-50:

"[T]he problem was simply how to do automatically what could already be done by the skill of the workman. On the other hand, the principle which the inventor applies for the solution of the problem is the capacity of a cam to vary the relative positions of two parts of a machine while the machine is running. Assuming this principle to be new, it might be possible for the inventor, having shown one method of applying it to the solution of the problem, to protect himself during the life of his patent from any other method of applying it for the same purpose, but I do not think that the novelty of the principle applied would enable him to make a valid claim for all means of solving the problem whether the same or a different principle were applied to its solution."

Applying these principles to the facts of the case before him, Lord Hoffmann concluded that the technical contribution made by Professor Murray did not justify a claim to a monopoly of any recombinant method of making the antigens. Its excessive breadth was not due to an inability to produce all the claimed results, but due to the fact that the same results could be achieved by different means. Professor Murray had established no new principle which all his successors had to follow if they were to produce the same results. It was not enough that Professor Murray had showed by his invention that the expression of Dane particle DNA in a host cell could be done. As Lord Hoffmann explained at page 52:

“…care was needed not to stifle further research and healthy competition by allowing the first person who has found a way of achieving an obviously desirable goal to monopolise every other way of doing so. ”

Finally, Lord Hoffmann confirmed that the reasoning by which he had come to the conclusion that the patent was not entitled to the claimed priority also led to the conclusion it was insufficient.

In this case Lundbeck submitted that the technical contribution of the Patent was the discovery and realisation of a new and non obvious compound, the (+) enantiomer. It was not an obvious goal. No one had produced or tested it before the priority date. Further, Lundbeck had shown a way of making it and so the Patent was sufficient.

I accept that if a patentee describes a new and non obvious compound which has a beneficial effect and describes a way by which it can be made then he is entitled to a patent for the compound. This is made clear in the passage from the speech of Lord Hoffmann which I have cited in paragraph [260] above. In such a case the technical contribution lies in the provision of the new and useful compound. Others might find different ways of producing it. But this does not render the original patent insufficient because in each case they are making use of the technical contribution – the knowledge they are making the new and useful compound.

In my judgment this is not such a case. For the reasons I have set out at length earlier in this judgment I am satisfied that medicinal chemists working in the field of SSRIs at the priority date considered it obviously desirable to separate out and test the enantiomers of active racemates. True it is that in the case of citalopram no one knew the activity would lie in the (+) enantiomer. However it was entirely obvious that the activity might lie primarily in one enantiomer rather than the other. Further, once the enantiomers had been separated the tests which the inventors carried out to determine where the activity lay were routine and straightforward, as were the steps necessary to formulate the (+) enantiomer into a pharmaceutical composition. The inventive step taken by the inventors of the Patent was not deciding to separate the enantiomers of citalopram but finding a way it could be done. The technical contribution they made was the discovery that the diol intermediate could be resolved and then the enantiomers of the diol converted into the enantiomers of citalopram whilst preserving their stereochemistry.

Claims 1 and 3 of the Patent cover all ways of making the (+) enantiomer of citalopram. For example, they cover resolving citalopram on a preparative chiral HPLC column. Does this method of resolution owe anything to the teaching of the Patent or any principle it discloses? In my judgment it does not. By June 1988 the preparation of the individual enantiomers of citalopram was an obviously desirably goal and their testing was trivial. There is no teaching in the Patent as to how that goal is to be achieved other than by the use of the diol intermediate. Just as in Biogen, it not enough to say that the inventors showed that resolution could be done and that they found the activity lay in the (+) enantiomer. The first person to find a way of achieving an obviously desirable goal is not permitted to monopolise every other way of doing so. Claims 1 and 3 are too broad. They extend beyond any technical contribution made by Lundbeck.

Claims 1 and 3 of the Patent are invalid for insufficiency. The other allegations of invalidity fail. Claim 6 is valid. I will hear argument as to the appropriate form of order if it cannot be agreed.