Les Laboratoires Servier & Anor v Apotex Inc & Ors

This is an appeal from one of the last first instance decisions of the late Lord Justice Pumfrey. He held that Servier’s EP (UK) 1 296 947 was invalid for lack of novelty and obviousness, but that if the patent had been valid, Apotex’s product would have infringed. He gave permission to appeal applying the Pozzoli test ([2007] FSR 37 at [10]). But he also refused to continue an interim injunction pending appeal on the grounds that he considered there was no real prospect of success. An application to this court for continuation of the injunction was refused on the same basis, [2007] EWCA Civ 783.

Undaunted, and doubtless because much money is at stake, Servier persisted in the appeal. Following Mr Purvis QC’s opening of the appeal, we decided that it was unnecessary to hear Mr Watson QC for Apotex. These are my reasons for dismissing the appeal.

The priority date of the patent is 6^th July 2000. It is for a particular crystalline form of the tert-butylamine salt of perindopil, a process for making it and for pharmaceutical compositions containing it. The patent, using its own nomenclature, calls this particular crystalline form of salt, the α form. It claims that the α form “especially exhibits valuable characteristics of filtration, drying and ease of formulation.” It does not say with what other crystalline form the comparison is being made or why this form is “especially valuable.” When asked about this, Mr Purvis could provide no answer.

The first and basic patent for perindopril and its tert-butylamine salt (without any indication or specification of crystalline form) was EP 0 049 658. It had a priority date 2^nd October 1980. Servier’s pharmaceutical compositions containing the salt have been on the market since the late 1980’s (the first, French, marketing authorisation was June 1988). The basic patent was effectively extended by a supplementary protection certificate which expired on 21^st June 2003. The market for the product is vast; UK sales alone are about £70m per annum at Servier’s patent protected price. It is not surprising that Servier have sought to exploit the patent system as far as possible to protect that vast income stream.

Servier sought and obtained additional protection for perindopril and the tert-butylamine by way of EP 0 380 341 (“341”) filed on 16^th September 1988, expiring on 16^th September 2008. This is for “the industrial synthesis of perindopril” and covers a process for making it and its tert-butylamine salt. There has been no investigation of the validity of this patent, and it does not matter for present purposes.

Finally Servier sought yet further protection for the tert-butylamine salt of perindopril by applying for three patents on 6^th July 2000, covering the only three crystalline forms (called by Servier α, β and γ) which, to date, have ever been found. We were shown the patent for the β form. It claims that that form too “especially exhibits valuable characteristics for formulation” without saying what these are supposed to be. The Judge describes the fact that there were simultaneous applications for the other two forms “curious.” That was perhaps a kind way of saying that Servier were simply trying to extend their monopoly in the salt.

The Judge held this about the three crystalline forms at [10]:

Professor Motherwell [Servier’s expert] accepted that it seems that form α is the stable form of perindopril and, absent special circumstances, it follows that any process for producing perindopril from ethyl acetate will produce this form in the absence of impurities. The β form, identified in the patent to which I have referred, is not stable and is created by special crystallisation techniques. Professor Motherwell said that eventually it reverts to the α form. The γ form is not produced by recrystallisation from ethyl acetate and is accordingly irrelevant.

So you need special conditions for making the β and even if you do, it will revert to α.

The upshot of all this is that were the patent valid, Servier’s monopoly in practice would last until 2020. But, as the Judge held and we confirm, it is invalid. And very plainly so. It is the sort of patent which can give the patent system a bad name. I am not sure that much could have been done about this at the examination stage. There are other sorts of case where the Patent Office examination is seen to be too lenient. But this is not one of them. For simply comparing the cited prior art (‘341) with the patent would not reveal lack of novelty and probably not obviousness. You need the technical input of experts both in the kind of chemistry involved and in powder X-ray diffraction and some experimental evidence in order to see just how specious the application for the patent was. The only solution to this type of undesirable patent is a rapid and efficient method for obtaining its revocation. Then it can be got rid of before it does too much harm to the public interest.

It is right to observe that nothing Servier did was unlawful. It is the court’s job to see that try-ons such as the present patent get nowhere. The only sanction (apart, perhaps, from competition law which thus far has had nothing or virtually nothing to say about unmeritorious patents) may, under the English litigation system, lie in an award of costs on the higher (indemnity) scale if the patent is defended unreasonably.

I turn back to the patent. As I have said the only citation against it is ‘341, itself acknowledged in the patent. ‘341 discloses in detail a process said and claimed to be “for the industrial synthesis” of the salt. I have to say I have never seen a patent claim which contains a limitation “for the industrial synthesis”. I do not know what it adds. It is not as though it is suggested that the chemistry is any different on a laboratory or any other scale. The reacting or crystallising molecules do not know about the size of the pot they are in.

After the description of the process of making the acid, ‘341 goes on to describe the process for making the salt (Stage 3D). It says:

“Place in reactor approximately 140 litres of ethyl acetate and 10kg [of the acid]. Add gradually approximately 2.20 kg of ter-butylamine, heat to reflux until all has dissolved; filter. Cool, filter off and dry. Yield 95%”

It does not say anything about the nature of the crystallisation which takes place during cooling, or indeed explicitly that the product is crystalline at all. But no one disputes that you will get crystals – there is no suggestion that the salt has an amorphous form.

The patent in suit is a little more specific about the crystallisation process said to produce the α form. In its widest form the process claim (original claim 2) merely said that a solution of the salt in ethyl acetate “is heated at reflux and is then cooled gradually until crystallisation is complete.” That differs from what was said in ‘341 only in the qualification “gradually”.

By a proposed amendment the rate of cooling was more specific, saying “cooled to a temperature of from 55 to 65^o at a rate from 5 to 10^oC/hour then to ambient temperature until crystallisation is complete.” This was the subject of original claim 5 and is based on a passage in the body of the specification saying that such conditions were “advantageous” but without saying why. It is not suggested that the more general original teaching is wrong – so although the claim has this limitation, it is not one which is meaningful technically – just a limitation in scope. It is not surprising that the judge held the proposed amendment did not save the claim, amounting as it does to merely technically arbitrary limitations.

The Judge’s finding of obviousness was based a short piece of cross-examination of Servier’s witness, Professor Motherwell. To understand it one needs to know that Servier had used the specified cooling conditions since 2000. It went like this:

The Judge said at [40]:

Professor Motherwell accepted that Servier’s cooling regime, at least so far as that is described in the evidence, is an obvious implementation of 341

Mr Purvis criticised that. He submitted it was based on a compendious question about both the pre-and post-2000 regimes, neither of which were in the public domain. He suggested the answer might be tainted with hindsight and the teaching of the patent.

I reject all of these criticisms. The Professor was unequivocally agreeing that each of the regimes was obvious. The Judge was not only entitled to accept that evidence and hold the process obvious, it would have been perverse to do otherwise.

One might have thought that would be an end of the case. After all if the cooling regime was obvious, its product, the α form would be too. But Mr Purvis had a further valiant attempt to save the claim. It went with his attempt to save claim 1. His argument was indeed desperate: he submitted that even if you used the cooling conditions specified in the patent you do not always get the α form. Although generally you would, sometimes you would not.

The premise is wrong (see below), but even if it were not, the process claim would be invalid. The patent teaches that this cooling process will produce what the patentee calls the α form. If the process does not do so reliably then the patent will be insufficient – it will not, in the language of Art. 83 of the European Patent Convention “disclose the invention in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art.” I do not think it would be a satisfactory answer to the sufficiency objection to say: “well it works most times.” And even if it were, if it were more likely than not that the α form would be produced, the claim would be obvious. It would be a monstrous thing if the law were otherwise. People are entitled to implement old recipes in obvious ways.

I turn to the novelty of claim 1. One must start with the interpretation of the claim as it would be understood by a person skilled in the art. It is to the:

“α crystalline form of [the tert-butylamine salt of perindopil] characterised by the following powder X-ray diffraction diagram measured [details specified]”

The “diagram” (more accurately it is a table) sets out a range of angle 2θ values (from “7.680^o” to 29.213^o”) 25 results for the inter-planar distance, intensity and relative intensity. The table is the same as that set out in the body of the specification.

On the evidence the skilled reader would understand the table to be obtained by the use of computer software followed by the use of some degree of subjective assessment to try to deal with overlapping peaks, lost peaks, peaks that have been smoothed out by averaging routines within the computer program or which have been obscured by noise suppression techniques. The reader would realise that there had been a single measurement rather than a series of repeats followed by averaging (the latter being the normal, more careful, way scientists measure things) and that the angles have been specified with a spurious degree of accuracy coming from the computer program.

The Judge rightly said that the “essential problem with this claim lies in the area of precision.” Its purpose is to provide parameters by which one can say “this is the α form.” Those parameters have an inherent degree of imprecision. The skilled reader would recognise this.

The reader would also recognise that the individual angles in the table were selected from a continuous graph of the kind shown by the Judge at [12]. Those chosen would be the angles at which there were peaks which the patentee had picked out as being significant. Finally the reader would know that different crystalline forms would normally give a quite different set of peaks even on a relatively crude basis.

Given that background, I see no reason to suppose that the reader would understand the figures in the table to be of absolute precision. The reader would take a set of very similar, if not exactly similar, peaks to be those of the crystalline form claimed by the patentee. No exact match is called for.

I turn to the experiments conducted for the purpose of this case. Apotex relied on four. Three were repeated at the request of Servier. Because experience has shown that experiments conducted for the purpose of litigation are apt to be biased, under English procedure, a party who wishes to rely upon an experiment which he has carried out, must give notice to the other side giving details and he must allow the other side to see a repetition of the experiment if it so wishes. Here Apotex relied on two experiments. Experiment 1 was on a laboratory scale and Experiment 2 on scale of ‘341 (which was called “pilot”). These were repeated with Servier representatives present. So there were four in all relating to ‘341.

The protocol for the experiments was created by an independent expert who had been provided with ‘341 but not the patent in suit. The protocol was accepted as fair.

Servier accept that three of these experiments produced results falling within the table. They accept that even though one of the peaks in the table, at 8.144⁰ is not present in any experiment. They need to say that because the PXRD table for Apotex’s own product also has no peak at that angle – so their own product would not be covered by the patent. To my mind the “missing peak” and the fact that all the experts attached no significance to it, shows that the individual peaks of the table should not have too much significance attached to them – it is the overall set which matters, both in position and relative intensity.

Servier say, however, that the remaining experiment (one of the pilot scale experiments) produced peaks which are different – so different as to take the product out of the claim. It follows, they say, that it is not an inevitable result of carrying out the prior art. And since the test for novelty is one of inevitable result (see Synthon’s v SKB [[2006] 1 All ER 685 at [23]) the test is failed.

To succeed on this point Servier need claim 1 to be construed narrowly. For the reasons I have given I do not so construe it. And the evidence was overwhelming that all four experiments produced what was essentially the same pattern.

First is the fact that that there are only three known crystalline forms. The patterns for β and γ are very different. So either the experiment produced a wholly new and hitherto undiscovered crystalline form which just happened to have a XRPD pattern very close to that of the α form or it produced the α form. The overwhelming probability is the latter.

Mr Purvis submitted that somehow this involved construing the claim with hindsight. I agree it is not permissible to do that. But I do not begin to see why taking into account the post-claim knowledge that only three forms have been discovered, when trying to decide whether a particular crystalline form is the claimed product, involves hindsight construction. You know that the claimed product is to have the properties shown in the table, give or take a bit. You have a product whose properties are close to those specified. Knowing that that there are only three possibilities and that the other two are very different leads you to the conclusion that the one you have is that of the claim – it cannot be anything else.

Mr Purvis also relied on a Micawber (Charles Dickens’ Mr Micawber, was always hoping that “something would turn up”) point here. There was some evidence from Servier’s expert, Dr Tarling (whose evidence in some respects the Judge described as not “altogether satisfactory”) that one can never be sure that all crystalline forms have been identified. A new one might turn up, even years later. But it was not suggested that this was likely in the case of this compound, or that it was a common occurrence. Still less was it shown that there was a realistic probability that a new form would turn up with an XRPD pattern so close to that of a known form as that between that of the claim and the experiment concerned. The overwhelming probability is that the product of the experiment is that of the claim. That is enough to satisfy the “inevitable” test. It is what a scientist would go on. The law, like science, is not concerned with the philosophical meaning of “inevitable” and whether anything can ever be said to be truly so.

Secondly there was the unchallenged evidence of Professor Cima. He provided an expert report in which he said of the experiment concerned:

“A comparison of the results obtained for Experiments 2 and 4 of the Notice with the 25 peaks claimed in the Patent shows that most of the peaks, including the major peak at 9 degrees 2θ, occur at identical angles. Only 3 of the peaks show a noticeable shift from the peaks claimed in the Patent and the other Experiments. This is the kind of result that might be expected from a slight distortion of the crystal structure”

And:

“As a crystallographer, I would not say that they were a different form but that they are a distorted form of that produced in Experiments 1, 2 and 4 of the Repeats.”

The Professor then noted that the PXRD for β and γ were very different and concluded:

“Given all of the above, in my opinion, the results of the Notice and Repeats all show the alpha crystalline form (both distorted and undistorted)”

Now some – particularly perhaps those unused to the probing powers of skilful cross-examination - have a bit of a cynical view as to party-provided expert evidence. The notion is that such an expert will say what his paying client will want him to say. There is no denying that sometimes indeed happens, though far from always. When it does, cross-examination will generally expose the witness as no more than a hired gun.

But even those of a cynical view ought not to take such an attitude as regards expert evidence which the other side is not prepared to challenge. Here Professor Cima was called for cross-examination. The later decision not to challenge his evidence was Servier’s. That means only one thing – it was not thought his opinion could be shown to be wrong or unsafe. Setting aside the Micawberism of Servier’s position his evidence was simply overwhelming.

Accordingly I think the judge was clearly right to hold that claim 1 lacked novelty. If that were so, Mr Purvis accepted that the process claim fell too. As I have said in my opinion that claim was bad in any event.

So I would dismiss the appeal. It was, as the Respondents said in their skeleton argument, “without merit.”

I agree that the appeal should be dismissed. I agree both with the reasons expressed by Jacob LJ and with the Lord Chief Justice's additional observations.

I agree that this appeal should be dismissed for the reasons given by Jacob LJ. The judge reached a firm decision based on a conclusion of fact. That conclusion itself turned on evidence given by experts in the analysis of powder X-ray diffraction profiles. That evidence gave the case a spurious veneer of technical complexity. It none the less led the judge to a firm and simple conclusion. The product in respect of which the patent was claimed was the natural, and indeed the inevitable, product of the process that was the subject of patent 341. It followed that the patent was bad for lack of novelty and obviousness. There was no realistic prospect of upsetting the critical finding of fact. I do not consider that this case satisfied the test in Pozzoli. Permission to appeal should not have been given.