England and Wales Court of Appeal (Civil Division) Decisions

IN THE COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE HIGH COURT OF JUSTICE
CHANCERY DIVISION
PATENTS COURT
Mr Justice Birss
[2015] EWHC 1395 (Pat)

		A3 2015 2162 Royal Courts of Justice Strand, London, WC2A 2LL
		21/03/2017

Lord Justice Floyd:

1. Synthon B.V. ("Synthon") appeals from the decision of Birss J dated 21 May 2015 dismissing its action to revoke European patent (UK) 2 361 924 ("the patent") in the name of Teva Pharmaceutical Industries Ltd ("Teva"). A further patent, European patent (UK) 2 177 528, was in issue before the judge, but is not an issue before us. The judge held all claims of the patent to be valid notwithstanding attacks based on lack of novelty, obviousness and insufficiency, and dismissed the claim for revocation. On this appeal, Synthon only relies on obviousness. By a separate appeal, Teva challenges the judge's decision that certain claims of the patent, claim 20 and claims dependent on it, were invalid for added matter.
2. The patent relates to glatiramer acetate, which is a mixture of synthetic polypeptides made from the four amino acids, alanine, glutamic acid, lysine and tyrosine. The patent is particularly concerned with levels of free bromine and metal ion impurities which may be present in the manufacturing process and in the final product.
3. Teva markets a low molecular weight fraction of glatiramer acetate under the trademark Copaxone. Copaxone is used as a treatment for relapsing remitting multiple sclerosis and is responsible for $4.2 billion worth of annual sales worldwide, representing 21% of the Teva group's total revenues. It is not suggested, however, that the commercial success of the product has any bearing on the issues we have to decide.

Technical background

4. The judge summarised the common general knowledge relevant to the case starting at paragraph 29 of his judgment. What follows borrows heavily, and with gratitude, from that passage.
5. Polypeptides are made up from individual amino acid monomers joined together by peptide bonds. Each amino acid has a carboxylic acid group and an amino group. The peptide bonds are formed by reaction between the carboxylic acid group on one amino acid and the amino group on another. In this way a polypeptide is built up by a sequence of peptide bonds.
6. Each of the amino acids also has a side chain. Some side chains are inert in the conditions used to make the polypeptide, whilst others are able to react. It was common practice, when amino acids have reactive side chains, to mask (i.e. protect) those side chains during the polymerisation reaction. Lysine, glutamic acid and tyrosine all had potentially reactive side chains.
7. One method for making polypeptides involves a random polymerisation reaction in which the amino acids are attached to the chain in an unpredictable sequence. This random polymerisation reaction leads to a mixture of molecules with different sequences of amino acids. One method for random polymerisation is by the use of an activated amino acid in so-called N-carboxy anhydride (NCA) form. The NCAs of the four amino acids involved in glatiramer acetate can then be reacted together to form random polypeptides.
8. The masking or protecting of side chains is undertaken by using a protecting group. A well-known protecting group of glutamic acid is a benzyl group which forms a γ-benzyl ester. A well-known protecting group for lysine is a trifluoroacetyl (TFA) group.
9. One of the ways of removing the γ-benzyl protecting group from the glutamic acid side chain is acidolysis. Acidolysis uses a reagent such as hydrobromic acid (HBr) in glacial acetic acid (AcOH) i.e. HBr/AcOH. Following reaction with HBr/AcOH, the deprotected product may be precipitated out as a hydrobromide salt by the addition of a suitable solvent.
10. HBr is a strong acid and HBr/AcOH is a highly corrosive agent. Due to its highly corrosive nature, the reagent is supplied and stored in glass and plastic bottles or containers, and reactions involving it may be carried out in non-metallic vessels.
11. The basic synthetic scheme for making glatiramer acetate was set out in a 1971 paper by Teitelbaum et al, in US Patent 3 849 550 and in Example 4 of international patent application WO 95/31990 ("Lemmon"). Example 4 of Lemmon is the prior art relied on by Synthon for its obviousness case. The process involves the following steps:

i) Polymerisation of the NCAs of tyrosine, alanine, γ-benzyl glutamate and N-TFA lysine.

ii) Deprotection of the γ-benzyl group used to protect the carboxylic acid group in glutamic acid. The product produced by this step is sometimes referred to as TFA-glatiramer acetate because the TFA protecting group is still present. This is the important step for the purposes of this case.

iii) A second deprotection step in which the TFA protecting group is removed by treatment with piperidine in water.

iv) Purification.

12. The amino acid composition of polypeptides can be characterised by a number of different analytical techniques. These include amino acid analysis, size exclusion chromatography (SEC), nuclear magnetic resonance spectroscopy (NMR) and matrix assisted laser desorption/ionisation time of flight mass spectrometry (MALDI-TOF).
13. Important issues in the case surrounded the possible presence of free bromine (Br₂) in the HBr/AcOH used in the deprotection step. Free bromine imparts a deep red-brown colour to a solution. The judge's findings were that the skilled person: (a) would not be able to quantify the free bromine level based on colour; (b) would not assume that there was much free bromine in a solution simply because it was coloured; (c) would know that the colour might not be due in whole or in part to free bromine but would infer that this was the most likely explanation if prompted to think about it; (d) would think that the free bromine level in a solution which was light yellow to colourless was low, if he or she thought about it, although they would not know what the level actually was; (e) would think that the free bromine content was likely to be lower in less coloured material.
14. Importantly, the judge found that there was no reliable evidence from which he could infer what the bromine level in freshly prepared HBr/AcOH kept away from light and oxygen would actually be.
15. A major dispute about the common general knowledge was whether it included knowledge that there was a potential problem caused by free bromine in HBr/AcOH (when used in a deprotection reaction) undergoing an undesired side-reaction with the phenol group in the tyrosine residues. The judge's conclusion was that the skilled person did not, as a matter of his or her common general knowledge, have a concern about a risk of such a side reaction. Thus skilled persons told to perform that reaction would neither set out to use a solution free of bromine nor would they take steps to remove free bromine by using a scavenger. They would not think about free bromine at all. The fact that, given a prompt to do so, they would conclude that colour in HBr/AcOH is most likely caused by some free bromine made no difference. Firstly they would not be prompted to consider it. HBr/AcOH would be regarded as a robust reagent. Secondly, even if they thought there might be some free bromine in the material, they would not have any reason to expect it to create an appreciable risk of a side reaction with tyrosine in the relevant conditions.

The patent

16. The specification of the patent explains that there are two aspects to the invention. One aspect relates to the level of free bromine, and the other aspect relates to the presence of metal impurities. At paragraphs [0090] to [0093] the specification explains that during the development of the production process for glatiramer acetate it was found that some of the tyrosine residues were brominated and that free bromine present in the HBr/AcOH caused the bromination of the tyrosine residues. The patent describes how this reaction can be prevented by adding a bromine scavenger, to mop up the free bromine.
17. At paragraphs [0094] to [0099] the patent explains that a red colour was detected after syringes containing solutions of Copaxone were kept at room temperature for 12 to 24 hours and that the source of this colour was unknown. It was determined that production of HBr in metal apparatus led to trace metallic ion impurities in the HBr. When HBr was later mixed with protected glatiramer acetate, the metallic ion impurities in the HBr were chelated by TFA-glatiramer acetate and glatiramer acetate itself. These entities contributed to the red colouration.
18. The patent continues:

"As a result, another measure taken to ensure purity, e.g. in the GA product, was the use of a non-metal reactor for the production of 33% HBr/acetic acid solution. The reactor used for the production of HBr/acetic acid solution was glass lined in order to prevent the formation of impurities which could later affect the purity of, e.g., the GA. In order to prevent contact of HBr solution with metal, parts of the piping were Teflon-lined. Similarly, other types of non-reactive, acid resistant non-metal apparatus can be used to prevent the formation of trace metal ions in the HBr/acetic acid solution. The use of a non-metal apparatus for the production of HBr/acetic acid solution was successful in eliminating the red colour from the GA. When the non-metal apparatus was used for the production of HBr/acetic acid solution, the result was that the solution was free of metal ions and the red GA was not formed."

19. The patents also make use of the American Public Health Association (APHA) colour index for "yellowness". Paragraph [0115] describes how an APHA measurement is made. Example 5 describes judging the colour of HBr/AcOH solution by visual comparison of the solutions with standard solutions from the APHA colour index. A table containing the APHA values for five batches of HBr/acetic acid that were manufactured using non-metal apparatus is provided (at paragraph [0118]). The APHA values for the five batches are between <300 and 700 APHA. The specification states (at paragraph [0119]) that the colour of the five batches indicated that the batches were free of bromine and free of metal ion impurities.

The claims

20. Teva made the judge's task more difficult by relying on the independent validity of a large number of claims. For present purposes the relevant claims are as follows.
21. Claim 1 of the patent is to a process for making TFA-glatiramer acetate in which the glutamic acid deprotection step involves using HBr/AcOH with a defined level of free bromine of less than 0.1%.
22. Claim 3 of the patent is not dependent on any other claim. It is for a process for obtaining a mixture of TFA-glatiramer acetate where the solution of HBr/AcOH used for the deprotection step comprises less than 500 ppm of metal ion impurities. There is no limitation in claim 3 to the level of free bromine. Claim 3 was not said by Teva to be independently valid.
23. Claims 8, 11 and 12 are all dependent on claim 7 (not itself said to be independently valid) which is limited to less than 0.05% free bromine. The claims themselves introduce successive limitations to the amount of metal ion impurities of 100 ppm, 10 ppm and "free of metal ion impurities" respectively.
24. Claims 13 and 16 of the patent depend on any of claims 1 to 12 and import limits relating to the colour of the HBr/AcOH in the process.
25. Claims 22 to 26 of the patent (for all of which Teva asserted independent validity) depend ultimately on either of claims 19 or 20 (via claim 21). Teva relied on claims 22-26 insofar as they depended on claim 20. The broadest claim said to have independent validity in this set is claim 22, in effect a claim to a mixture of glatiramer acetate with a desired molecular weight and less than 100 ppm of metal ion impurities. The narrower claims 23-26 consist of a cascade of metal ion levels in the glatiramer acetate down to "free of metal ion impurities" in claim 26.
26. Claims 27 and 28 depend on any of claims 19 to 26 and are limited by reference to the APHA colour of the mixture.
27. Finally, claim 29 of the patent is directed to a process of making a pharmaceutical composition containing glatiramer acetate which has a step of measuring the percentage of brominated tyrosine of each batch of product and including the batch in the pharmaceutical composition only if its percentage of brominated tyrosine is lower than 0.3%.
28. A distinction which it is necessary to keep in mind is between claims directed at levels of free bromine or metal ion impurities in the HBr/AcOH solution on the one hand and claims directed to the levels of free bromine or metal ion impurities in the final resulting product, namely the mixture of glatiramer acetate. Claims 1, 3, 7, 8, 11-13 and 16 are in the former category, whereas claims 20 and 22-28 are in the latter category.

The judgment of Birss J

29. The judge dealt with novelty over Example 4 of Lemmon at paragraphs [96] to [103] of his judgment. Claim 1 of the patent had one difference as compared with Lemmon, namely that the level of free bromine in the HBr/AcOH solution had to be below 0.1%.
30. Given the absence of any teaching in Lemmon about free bromine, the judge concluded that Lemmon did not anticipate claim 1. The judge then turned to the novelty of claims 22 to 26 of the patent which involved a limit on metal ions in the overall mixture. Again, there was no express teaching in Lemmon to produce the HBr/AcOH solution in a glass-lined vessel or use glass-lined apparatus. The judge concluded that it was more likely than not that a skilled person would only ever use a non-metallic vessel to produce the HBr/AcOH solution. However this did not dispose of the issue of novelty because there were other potential sources of metal ions in the overall process. The judge concluded that it was not inevitable or necessary to avoid potential sources of metal altogether. It followed that glatiramer acetate produced following Lemmon would not necessarily have metal ion impurities at the level required to satisfy claim 22.
31. On the issue of inventive step the evidence of the parties' independent expert witnesses was of importance. Synthon called Prof Alethea Tabor, Professor of Chemical Biology at University College, London and an expert on polypeptide synthesis. The judge was critical of her written evidence but attributed this to over-reliance on Synthon's legal team. By contrast, in the witness box, she had struck the judge as candid, answering questions without equivocation. Teva called Prof Ben Davis, who is Professor of Chemistry at Oxford University. His research centred on biomolecular functions with an emphasis on proteins and carbohydrates. Many of his publications were in the field of proteins and peptide science. The judge found that Prof Davis had the occasional tendency to get carried away, but overall this tendency did not undermine his reliability as an expert witness.
32. The judge started by considering the obviousness of the claims limited by reference to the level of free bromine. Prof Tabor had advanced two main points in her report as to why it was obvious to eliminate free bromine in the HBr/AcOH solution. The first point was based on the proposition that it was known that the free bromine might react with tyrosine. The second point was that it was obvious to include a scavenger such as phenol to take care of benzyl cations which could arise in the deprotection step and that such a scavenger would, as it happens, remove free bromine as well.
33. The judge rejected both these points. The first argument was defeated by the judge's factual findings as to the common general knowledge, which I have set out above. The skilled person simply would not be alerted to a risk of bromination of tyrosine as a result of using HBr/AcOH solution. That was so despite the higher temperature and longer time proposed in Example 4 of Lemmon. The second argument depended on there being a risk of the presence of benzyl cations. However the judge found that that there was no such risk, given the mechanism of the acidolysis reaction, a position which was common ground by the end of the evidence. Neither of these two obviousness arguments is now advanced on this appeal.
34. Synthon advanced a third approach at trial. This approach was based on the suggestion that the skilled person would routinely perform an amino acid analysis to characterise their product having followed Lemmon. That analysis would reveal the presence of some brominated tyrosine. The skilled person would readily identify the HBr/AcOH solution as the source of the bromine atoms. It would be obvious to remove the bromine, and the use of a bromine scavenger such as phenol an obvious expedient.
35. The judge dealt with this argument at paragraphs [116] to [126] of his judgment. He accepted that if the skilled person discovered a significant level of bromotyrosine in the products of Example 4 of Lemmon, then the following steps in Synthon's argument would be made out as a matter of obvious chemical reasoning. He identified the issue to be whether the process of reasoning would start at all. He found as a fact that none of the available methods of analysis would lead a skilled person to discover the presence of bromotyrosine if he was not actively looking for it, and even then not all of the available analytical techniques would discover its presence. The third argument is no longer pursued on this appeal.
36. A further, fourth, way of putting the obviousness attack was based on Agrevo/Triazoles (Case T 939/92). Synthon argued that all the invention amounted to was the idea of using a pure reagent in a known reaction, and this, in law, could not support a patent. The judge held this argument not to be open to Synthon on the pleadings. Further, he was unpersuaded by it in any event: the patent made a relevant technical advance. This fourth argument is no longer pursued on this appeal.
37. The judge then turned to the other aspect of the invention, metal ion impurities. Synthon's case was that a skilled person observing the redness in samples of polypeptide made according to Example 4 of Lemmon would consider metal ion impurities to be a possible cause. The presence of metal ion impurities could be readily established by standard analytical techniques. Once detected, it would be entirely obvious to use non-metallic apparatus to produce the HBr/AcOH solution, since it was well known that that solution was highly corrosive to metals, and the interaction between the reagent and metal would be an obvious and likely source of metal ion impurities. Taking the step of using non-metallic apparatus would in fact eliminate the source of metal ions identified in the patent and from the patent one could take it that this would solve the problem.
38. The judge held that this argument failed on the evidence of both experts. It was common ground between the experts that the standard practice of a skilled person faced with a problem of discolouration would be first to identify the cause of the colouration and then, having identified the cause, employ a solution based on that knowledge. It was true that metal ions would be one possible cause of the discolouration but it would be one among a number of others, such as a peptide degradation product, a bromine atom containing species (which could generate free bromine) and an impurity which leached from the syringes in which the discolouration was observed. Each of the possible causes would require different resolution. The judge expressed his conclusions in this way:

"130. The skilled person would set out to identify the likely cause of the colour. They might well measure the metal ion level as part of that investigation. However to take Synthon's approach and jump to the conclusion that metal ions should be eliminated is not the activity of the uninventive skilled person. I am not satisfied the skilled person would just try things without having identified the cause of the colour. I am not satisfied that even a substantial research project over a number of months would identify that metal ions are the/or a likely cause. If having failed to pin down the cause a skilled person then changed tack and started just trying out different expedients, alighting upon elimination of metal ions may or may not be one of the expedients to test. However either way I reject the idea that that sort of exercise would be indicative of obviousness. It is not.

131. Finally, just as in the novelty section, I am not satisfied that using non-metallic vessels would necessarily eliminate the colour (because there may be metal ions from other sources) nor am I satisfied that it would have been obvious to go as far as to eliminate all metal equipment altogether unless the skilled person was motivated to do so by a finding that metal ions were indeed likely to be the cause of the problem. It follows that the step of using non-metallic vessels would not inevitably solve the problem. If it did not then assuming the skilled person had not identified the cause of the problem when doing what Synthon contend for, they are no further forward. As before whether this has an impact on sufficiency is a matter I will address below."

39. The argument dealt with there by the judge is again not pursued in this court. As will be seen, Synthon now say that the problem of redness is one which would not be observed in the first place.
40. The judge also rejected the obviousness attack on claim 29, but did not deal with it separately.
41. The judge then turned to the issue of added matter, which arises on Teva's appeal. Synthon's case was that the application for the 924 patent disclosed TFA-polypeptides containing less than 1000 ppm metal ion impurities and processes involving a solution of HBr/AcOH also containing less than 1000 ppm metal ion impurities. However, what was not disclosed was glatiramer acetate which itself contained less than 1000 ppm metal ion impurities. Claim 20 of the patent therefore involved a new disclosure, namely a glatiramer acetate mixture containing less than 1000 ppm metal ion impurities.
42. Teva argued that the general disclosure of the application was of the need to avoid metal ion impurities. Coupled with the disclosure of the cascading lower levels of metal ion impurities in the HBr/AcOH solution, there was sufficient to amount to a disclosure of low metal ion impurities in the final product. Teva also pointed to some third-party observations submitted to the European Patent Office in the course of prosecuting the patent which relied on this added matter argument, and to the fact that the EPO granted the patent despite the observations.
43. The judge was not persuaded by these arguments. It did not follow that the metal ion level in the glatiramer acetate would always be the same as the level in TFA-glatiramer acetate (or in the HBr/AcOH solution). Levels at one stage did not necessarily follow from the other. Claim 20 disclosed added matter and claims 21 to 26, insofar as they depended on claim 20 added matter too.

The arguments on Synthon's appeal

44. Mr Andrew Lykiardopoulos QC, who appeared on the appeal with Ms Anna Edwards-Stuart, recognised the difficulties which he faced when appealing from a judge's evaluative judgment on an issue of obviousness. He submitted, however, that the judge had fallen into correctable error on both aspects of the invention.
45. On the metal ion impurity aspect of the invention, Mr Lykiardopoulos emphasised that the teaching of the patent was that HBr/AcOH should be kept away from metal during its production by the use of glass-lined reactor vessels and Teflon-lined pipework. That was, in essence, all that was claimed about metal ion impurities in claims 3, 8, 11 and 12. However, it was obvious to keep metal away from HBr/AcOH during its production. It was not necessary to be prompted to do so by observing red colouration in syringes of finished product. It was simply good manufacturing practice, given the common general knowledge that HBr/AcOH was highly corrosive to metal. The evidence showed that it was supplied and stored in glass or plastic bottles. The point was made good by the specification sheets issued by commercial suppliers of the solution.
46. The patent taught expressly that, provided non-metal apparatus was used, the HBr/AcOH solution was free of metal ions. The metal ion impurity levels in claims 3, 8, 11, and 12 were all obvious on this basis.
47. The judge, submitted Mr Lykiardopoulos, had approached this issue from the standpoint of claims 22 to 26. His conclusions (at paragraphs 93, 101 and 131) that adopting glass reactors for HBr/AcOH would not prevent metal ions coming in at other stages did not apply to the manufacture of HBr/AcOH, or its use in the deprotection reaction. These conclusions were all in the context of the overall process of making the finished glatiramer acetate product, where metal ions could come in from other sources. The judge had failed to direct his mind to whether it was obvious to use a solution of HBr/AcOH which, during its manufacture, storage and putting into use had not come into contact with metal.
48. Mr Lykiardopoulos accepted that the case had not been put in this way before the judge, and that for that reason the judge may have considered that he only needed to look at claim 22. However, he submitted that these earlier claims were clearly invalid. He did not challenge the judge's finding that claims 22 was not invalid for obviousness. He did not need to because claim 22 was, as he submitted, invalid for added matter.
49. On the free bromine aspect of the invention, a preliminary procedural issue was how low a level of free bromine in the HBr/AcOH solution had to be shown to be obvious. Claim 1 put the level at less than 0.1% whereas claim 7 (on which claim 8 depended) claimed less than 0.05%. Teva had not contended at trial that Claim 7 was independently valid. Mr Lykiardopoulos submitted that it was not open to Teva to contend that there was inventive significance in the 0.05% limitation if it was found that the 0.1% limitation was obvious.
50. Mr Lykiardopoulos advanced two obviousness arguments on this aspect. The first argument was that prior art, freshly made, clear HBr/AcOH would have a free bromine level of less than 0.1% and therefore fall within claim 1. It was obvious to use freshly made HBr/AcOH. He submitted that the judge had been wrong to say in paragraph 56 of his judgment that there was no reliable evidence as to the level of free bromine in freshly prepared HBr/AcOH kept away from light and oxygen. He ought to have accepted evidence from Prof Tabor that it was less than 0.1%.
51. The second argument was that the judge had lost sight of the fact that the context of the invention was the manufacture of a pharmaceutical, where purity was an important consideration. He had been led astray by evidence as to what would be done in a university research laboratory, where less stringent requirements for purity would apply. In the pharmaceutical context, impurities such as free bromine, when noticed because of the discolouration, would be removed.
52. Finally, Mr Lykiardopoulos argued that claim 29, the claim which involves a bromotyrosine measurement step was obvious as well. The claim was wide enough to cover the levels of brominated tyrosine which the patent shows would result from the operation of the known processes. There could be no invention in simply performing a measurement on the prior art process. The step of requiring a measurement did not solve any technical problem.

Discussion – Synthon's appeal

Metal ion impurities – claim 3

53. Claim 3 merits separate consideration because it is limited only by reference to metal ion impurities in the HBr/AcOH solution. Prof Davis, in his second report, recognised that, on the laboratory scale, the skilled person would know that HBr/AcOH was a corrosive substance that should not be allowed into prolonged contact with metal, for example in storage or production. However, he had gone on to say that the skilled person would not have the same level of concern in situations where contact time between the HBr/AcOH and the metal was low, for example during transfer by syringe or, on a larger scale, through a pipe. On an industrial scale, he understood that apparatus made of Hastelloy (a special alloy) could be used in the handling of the reagent, including the piping.
54. Mr Lykiardopoulos submitted that this evidence recognised that measures would have to be taken to prevent contact between the reagent and metal in the industrial context, and that this extended to the pipework. A relaxed attitude to a short contact time between the reagent and metal in the laboratory did not translate to the industrial context. Although Hastelloy was one way of dealing with this, it was also obvious to use glass and Teflon. If this was done, metal ions in the solution would be avoided, as the patent taught.
55. An oddity about this case is that Teva did not identify claim 3 as an independently valid claim. It is an oddity because claims 1 and 3 are directed to the free bromine and the metal ion aspects respectively, and there is no reason why a finding of obviousness of claim 1 should impact on the validity of claim 3. However I did not understand Mr Lykiardopoulos to be submitting that Teva had conceded claim 3 to be invalid. Rather, he was submitting that the judge ought to have made a finding to that effect on the state of the evidence. He wished this court to make a specific finding for reasons connected with the continuing opposition proceedings in the EPO.
56. It is important to bear in mind that the obviousness attack on claim 3, as it is now advanced, is no longer based on the suggestion that the skilled person would have discovered a problem associated with metal ions and be seeking to remove them. Armed with knowledge that metal ions are a bad thing for the finished product, the skilled person would know how to take steps to avoid them. The argument, as it is now put, is based simply on a concern to keep HBr/AcOH away from metal surfaces at all times.
57. Keeping HBr/AcOH away from metal surfaces, and creating a solution of HBr/AcOH with less than 500 ppm metal ion impurities (as claim 3 requires), are not the same thing. Synthon bore the burden of proving that, if steps were taken to keep the HBr/AcOH away from metal surfaces, this level of purity would be achieved. They sought to do so without adducing evidence, and by pointing to the teaching of the patent specification.
58. However the evidence at trial went further than the teaching of the patent. Thus Prof Davis gave his evidence about the ways in which the HBr/AcOH could pick up metal ions, even on what he described as "a larger scale". Moreover in paragraph 1.51 of his second report Prof Davis explained that metal ions could come from a whole variety of sources, not just the reactor vessels and pipes. They might be present in "the other reagents used in the synthesis of the GA, for example, the amino acid NCAs, dioxane, piperidine and even in the process water." Although he does not mention the acetic acid specifically, his statement is entirely general.
59. On this evidence, therefore, I would not regard it as having been established at trial that HBr/AcOH would have less than 500 ppm metal ion impurities provided only that it was kept in glass vessels. It follows that I do not consider that it was established that claim 3 was obvious.

Free bromine claims

60. The evidence relating to the amount of free bromine in colourless HBr/AcOH was as follows. Prof Tabor's report stated that one could tell whether a solution had free bromine or only trace amounts by whether it was colourless. With increasing concentrations the solution would darken. One would expect the solution to be colourless if it was newly sourced or had been stored away from light and oxygen and then, for only a short period of time.
61. In Prof Davis' second report he did not accept (a) that there was any direct correlation between colour and free bromine concentration, (b) that colourless HBr/AcOH would contain only trace amounts of available free bromine, (c) that "newly sourced" solution from suppliers would be colourless, referring to the specification sheets, which showed it to "colourless to yellow to light orange to brown".
62. Prof Davis' cross-examination elicited the fact that dissolving hydrogen bromide into acetic acid could result in a clear solution if there were no interactions going on. He also accepted that there was a qualitative relationship of increasing colour as one added bromine, albeit that one could not quantify the amount of bromine from the colour.
63. In cross-examination Prof Tabor accepted that it was not possible to assess from the colour of a solution how much bromine was present. She was asked to explain what she meant by the "trace amounts" which would be present in a colourless solution. Her answer on the first day of the trial was as follows:

"I think all I meant there was an amount that you cannot detect by the naked eye. As I say, I am not concerned in this paragraph to quantitate in any respect the level of contamination of the HBr in acetic acid by the colour because I simply believe it is impossible to do that."

64. She was asked the question again on the second day of the trial. She replied:

"A. I think all I meant there was that it was too low to be detected by the naked eye. Probably -- that is a good question. I did actually look into this. It was talking certainly less than one part in 1000, I think -- probably much less than that. It is not that easy to find absolute values for these things.

Q. Indeed. There is actually nothing in the literature that actually helps you on the amount of bromine that leads to colour, is there?

A. No.

Q. You did look for it?

A. I did look for it, yes.

Q. And you could not find anything?

A. I could not find any absolute correlation between the absolute amount of bromine in a solution and the colour as evinced by, for example, the APHA scale, which is referred to later in the patents – – and I did look."

65. Prof Tabor also explained her reference to "newly sourced". By newly sourced she meant generated in your own laboratory. She was asked whether she would still expect some colour even in newly made solution, and said that she could not put her hand on her heart and say absolutely definitely that it would be colourless at the point of generation.
66. I do not think this evidence compels the conclusion that freshly prepared HBr/AcOH is necessarily free of bromine. The most that could be said is that it might be colourless. Mr Lykiardopoulos placed his principal reliance on Prof Tabor's answer that the amount of free bromine to cause colour was much less than one part in 1000. He submits that this evidence is supported by statements in the patent. Thus the batches tested in Example 5 were plainly coloured to some degree, but were deemed free of bromine. He accepts that this latter argument, based on the colours in Example 5, was not put to the judge.
67. Stripped to its essentials, this aspect of the appeal simply asks whether the judge was entitled to treat as unreliable Prof Tabor's evidence as to the level of bromine which would cause discolouration. In my judgment he was plainly so entitled, and this court would not be justified in making a finding that the level was less than 0.1%. Firstly, both experts were agreed that colour did not enable one to quantify the amount of free bromine. The relationship between colour and increasing bromine content was a purely qualitative one. Secondly, Prof Tabor's evidence was that calculating a level of free bromine which would cause colour was a very complicated thing to do. Thirdly, although Prof Tabor had looked into the question, she was unable to find anything in the literature which would produce a figure. Fourthly, there was evidence before the judge that other interactions were at least possible, such as production of Br₃- _, and the extent to which those interactions were influencing colour was difficult to predict.
68. I do not think that Example 5 in the patent can be used to bolster Prof Tabor's evidence. If colour is an unreliable guide to absolute quantities of free bromine, then the fact that low quantities of bromine were thought to be present in particular, slightly coloured batches does not establish what the levels of bromine were. Once the judge had accepted the lack of any quantitative connection between colour and bromine content, Example 5 cannot be used to advance Synthon's case on the facts.
69. Against that background the judge was plainly entitled to reject as unreliable Prof Tabor's evidence on this issue. I would reject this ground of appeal.
70. The second way in which the obviousness case is put assumes that the skilled person obtains coloured, not colourless HBr/AcOH. The evidence here went as follows. Prof Tabor said in paragraph 85 of her report that the skilled person would be aware of the general need to keep impurities to a minimum. This was particularly so where the process involved the manufacture of a pharmaceutical product for use in humans. Free bromine would be an impurity, and when carrying out any chemical reaction the skilled person "is looking to use as pure reagents and therefore produce as pure a product as possible. A skilled addressee seeking to reproduce Lemmon would therefore, in my opinion, have sought to use a solution of HBr/AcOH which was free of impurities". Prof Davis responded that this view required qualification. The skilled addressee also needed to take into account the cost and availability of reagents. Although it is generally possible to obtain ultra-pure reagents, for example of analytical grade, these tended to be much more expensive than less pure reagents. A standard reagent, such as HBr/AcOH, was generally not ultra-pure because there was no need for extreme levels of purity. He did not specifically respond to Prof Tabor's point concerning pharmaceutical processes, but he did not expressly accept it or qualify his evidence by reference to it.
71. Mr Lykiardopoulos submits that the skilled person who sees the colouration in the HBr/AcOH solution will, in the context of pharmaceutical preparation, take steps to remove the impurity which causes it. On the judge's findings he would know that the impurity was free bromine. It could easily be removed using a scavenger, such as phenol. The judge had made all his findings about the skilled person's attitude to the presence of free bromine ignoring the context in which the question would arise.
72. Whilst admiring the elegant simplicity of this argument I am unable to accept it. Firstly, beyond the exchange of views in the evidence, it was not an argument which was developed or advanced below. Secondly, Prof Tabor accepted in cross-examination that her view in paragraph 85 of her report ("free of impurities") was subject to qualification. It did not mean completely free. This is important, given that the context of the statement was the pharmaceutical one. Thirdly, Prof Davis had challenged Prof Tabor's assertion on the basis that it was an excessive generalisation. His view was that one could not justify the use of ultra-pure reagents in every case. Fourthly, I find it difficult to accept that Birss J, a judge with enormous experience of cases involving pharmaceutical processes, was not alive to the fact that the relevant context was a pharmaceutical process and that the attitude to impurities in that context might be more rigorous than in a research laboratory. Fifthly, given the way in which the argument is now advanced, the skilled person would have no reason to suppose that free bromine would take part in any side reaction, and therefore present a significant problem.
73. In the circumstances it is not necessary to resolve the question of whether Teva had accepted that there was no independent invention in a bromine level of 0.05%. Synthon have failed to establish that the broader claims limited to 0.1% free bromine are invalid.

Test for brominated tyrosine – claim 29

74. Very little time was taken up in argument over this claim. On the case as it is now argued, the skilled person would have no reason to suppose that bromotyrosine was present at any stage in the process, and therefore have no reason to develop or include a test for it in a manufacturing process. I cannot see, therefore, how inclusion of that step in the manufacturing process, coupled with a step of only including batches below a certain level, can be regarded as obvious.
75. Mr Lykiardopoulos pointed to data in the patent which showed that prior art methods could result in levels of bromotyrosine below 0.3%, and that only one reading was above 0.3%. Thus all the patent was doing was imposing a measuring step on the prior art process. No batches would need to be removed. There was no invention in performing an unnecessary measurement on the prior art.
76. I am not persuaded by this argument. Bromotyrosine was an impurity which no one had any reason to suspect was present. It was a useful advance to impose a control step and exclude batches where the level is 0.3% or above. The data in the patent shows that such levels could be achieved in the prior art.
77. I would therefore dismiss Synthon's appeal.

The arguments on Teva's appeal

78. Mr Waugh submitted that the judge had been wrong to find added matter in claim 20 as compared with the application for the patent. The application for the patent discloses the following:

i) HBr/AcOH solution with less than 1000 ppm metal ion impurities;

ii) Trifluoroacetyl polypeptides with less than 1000 ppm metal ion impurities, and glatiramer acetate as a paradigm example of the polypeptides of the application;

iii) that trifluoroacetyl glatiramer acetate (TFA-GA) is the immediate precursor to glatiramer acetate in the synthetic process disclosed in the application;

iv) the need to reduce or eliminate metal ion impurities because of the ability of TFA-GA to chelate metal ions, thereby producing an unwanted red colour in glatiramer acetate.

79. Mr Waugh accepts that glatiramer acetate obtained from TFA-GA which itself has less than 1000 ppm metal ions is not expressly disclosed by the application. He submits, however, that because the application clearly disclosed the need to avoid metal ion impurities and discloses HBr/AcOH and TFA-GA with less than 1000 ppm, it would also implicitly disclose to the skilled person that he should have correspondingly low levels of metal ions in his final GA product.

Discussion – Teva's appeal

80. The principles applicable to the objection of added matter were not in dispute. The matter in question must be disclosed directly and unambiguously, although this does not exclude an implicit disclosure: see European Central Bank v Document Security Systems Inc [2007] EWHC 600 (Pat) at [100] (Kitchin J) approved in this court in Vector Corporation v Glatt Air Techniques Ltd [2007] EWCA Civ 805 at [7] (Jacob LJ with whom Wall and Smith LJJ agreed). The same passage emphasises that the test is not an obviousness test.
81. I have no hesitation in concluding that the judge was right in relation to the issue of added matter. The application discloses TFA-polypeptides containing less than 1000 ppm metal ion impurities, but it is not inevitable that glatiramer acetate made from this material will have a corresponding quantity of metal ions. A general teaching to minimise the levels of impurity is not equivalent to a disclosure of glatiramer acetate with less than 1000 ppm of impurity.
82. As the judge pointed out, it was part of Teva's case that downstream steps from the TFA-polypeptides could themselves introduce metal ion impurity. Whether the skilled person would undertake steps to exclude metal ions in these steps is a question of obviousness. There is no express or implied teaching that this must be done. Moreover, the claim has transposed a concentration in parts per million of one compound, the TFA-polypeptide, into a concentration in parts per million of another, the finished glatiramer acetate. The appropriate "corresponding" limit for the finished product is not disclosed in the application, and is most unlikely to be 1000 ppm.
83. As for the proceedings in the EPO, we were told that the Opposition Division has now held that there is added matter, so the EPO's earlier finding in the course of prosecution is of no weight.

Conclusion

84. For the reasons I have given I would dismiss both appeals.

Lord Justice Briggs

85. I agree.

Lord Justice Kitchin

86. I also agree.