Les Laboratoires Servier & Anor v Apotex Inc & Ors

I am now called upon to consider the enforcement of the cross undertaking in damages. It was not at the principal hearing suggested that there were any grounds for not enforcing the undertaking: the question was simply one of quantification. (That position altered).

The principles of law sufficient to enable me to quantify compensation in this case may be shortly stated:-

Although that summary is probably a sufficient statement of the principles that I must apply in determining this case I should clarify my approach to the relevant principles in three respects.

First, I am following the obiter guidance contained in the opinion of Lord Diplock in Hoffman la Roche because, on the evidence and argument presented at trial, it is sufficient to enable me to determine the issues that arise. For my own part, I think it should be recognised that the award is of equitable compensation (not of damages strictly so called) and that there may be occasion to examine whether such equitable compensation should be fettered by rigid adherence to common law rules; and further, that if common law rules are to be applied, whether those relating to contract are more appropriate than those relating to tort or some other breach of duty (in which connection it will be noted that the judgment of Brett LJ upon which Lord Diplock founded his view referred to “[a] contract with or duty to the opposite party”). The difference between the two sets of common law rules would be important, for example, in the context of aggravated or exemplary damages for a blatant or cynical interference with a defendant’s right to enter a pharmaceutical market with a generic drug by means of a second generation patent that is a “try on” (to adopt the language of Jacob LJ).

Second, proceeding on the footing that the enquiry on the cross undertaking is to be conducted according to the principles applicable to contractual damages, I should record that the form of the case before me has meant that I have not been called upon to consider whether it is appropriate to depart from the conventional contractual basis of assessment and instead to apply the exceptional “restitutionary” basis of assessment in contract considered in Attorney General v Blake [2001] 1 AC 268. Where what is found to be a wrongful extension of patent protection results in a benefit to the patent holder which exceeds and outstrips the loss which is occasioned to the generic company whose market entry is delayed then it seems to me that “restitutionary damages” might be called for (notwithstanding the rejection by the Court of Appeal in Smithkline Beecham v Apotex [2006] EWCA Civ 658 of the argument for a general restitutionary claim based on unjust enrichment and enforceable by parties and non-parties alike). In the present case I know only (because it was accepted in argument) that the profits made by Servier from the sale of perindopril during the period of the wrongfully granted injunction “significantly exceed” the maximum sum which it can be called upon to pay on the present enquiry.

Third, whilst it is for Apotex to establish its loss by adducing the relevant evidence, I do not think I should be over eager in my scrutiny of that evidence or too ready to subject Apotex’ methodology to minute criticism. That is so for two reasons, quite apart from an acceptance of the proposition that the very nature of the exercise renders precision impossible. (a) Whilst, in order to obtain interlocutory relief, Servier will not have had to persuade Mann J that it was easy to calculate Apotex’ loss in the event of the injunction being wrongly granted, it will have had to persuade him that that task was easier than the calculation of its own loss in the event that the injunction was withheld. The passages I have cited from its skeleton argument and evidence show that it did so. Having obtained the injunction on that footing it does not now lie in Servier’s mouth to say that the task is one of extreme complexity and that the court should adopt a cautious approach. Having emphasised at the interlocutory stage the relative ease of the process, it should not at the final stage emphasise the difficulty. (b) In the analogous context of the assessment of damages for patent infringement, in General Tyre [1976] RPC 197 at 212 Lord Wilberforce said:-

“There are two essential principles in valuing the claim: first, that the plaintiffs have the burden of proving their loss: second, that the defendants being wrongdoers, damages should be liberally assessed but that the object is to compensate the plaintiffs and not to punish the defendants.”

With that summary of the principles I turn to consider my findings of fact. I shall deal first with the way in which the pharmaceuticals market operates, before turning to my detailed findings as to what occurred in the present case; I will then turn to the process of assessment based on those general and specific findings.

Each of the witnesses who gave evidence was entirely honest and straight forward although displaying, in differing degrees, circumspection in addressing the factual matters which presented the greatest difficulties for their respective companies. The central issue for me has been its reliability, for in key areas the evidence has not concerned recollection of events which did in fact happen, but an attempted recollection by witnesses of what their intentions were or would have been in relation to decisions they did not have to take in circumstances which did not in fact occur and in assumed ignorance (but actual knowledge) of what did in fact transpire. Evidence of that sort may be given with perfect integrity, but be innocently self serving, and must therefore be subjected to scrutiny with the aid of whatever objective tools may be to hand, such as inferences properly to be drawn from the actual transactions which did take place, and commercial realism. But Mr Purvis QC warned me (and I thought the point well made) that undertaking this exercise does not entitle me to construct out of the available materials a convenient hypothetical world which simply substitutes my estimate of commercial advantage from the way the parties themselves perceived their commercial interests to lie. I have endeavoured to heed that warning.

Before turning to the detailed matters which must be determined in order to arrive at a result in this case, I will set out my findings of fact as to the way the market operates in general. There are two features, both of which are in play in the instant case.

First, the “at risk” period. Where a drug patent has been registered but its validity is under challenge any company which brings onto the market a competing generic drug does so “at risk”. The risk is enormous. The “protected” branded product is generally sold not simply at a “premium” price but at a hugely profitable price. Coversyl was on the market at about £11 per unit, whereas the “floor” price for the generic product (which is obviously still profitable for those who manufacture and sell it) is currently £1.50 per unit. The whole point of the generic product is to provide a cheaper alternative. A generic pharmaceutical company which launches its generic product in the “at risk” period in order to make a margin of x% on each unit sale may (if the patent is upheld and its product found to be infringing) therefore ultimately find that it is liable to pay damages in respect of every unit it sold at 2x% or 5x% or 7x% or more. Thus if a unit of generic perindopril sells for £1.50 and yields Apotex 50 pence profit, but Coversyl sells for £11 per unit, in seeking to make its profit of 50p per unit Apotex is having to run the risk of having to pay Servier damages of £10 per unit. Entering a market “at risk” thus requires (a) a high degree of confidence in the accuracy of the “judgment call” on the validity of the patent, (b) a company capitalised at a sufficient level to secure that any misjudgement on that validity question can be survived and will not lead to the destruction of the company (which may have a range of other profitable generic products not “at risk”); and (c) experience both of the market into which the competing products are being sold and the strengths and weaknesses of the brand leader with whom the fight will have to be conducted or a deal struck.

The second feature is the market dynamic. The market ultimately moves from one absolute state (the monopoly of the patent holder) to another (an entirely open market in an unprotected product). But the move from one such state to another is not a smooth transition properly represented by a straight line or a simple curve. There are transitional stages which themselves are characterised by periods of rapid price adjustment (“transition periods”) interspersed with periods of relative price stability (“plateau periods”). The transition periods represent the market response to an actual or rumoured new entrant (whose only ability to gain market share from existing participants will be through price advantage, but who will have no interest in driving prices immediately to rock bottom whilst there remains some advantage in sharing in the profit margins established by the earlier and fewer participants). The reason for the plateau period is that if the number of participants in the market is relatively stable, then gradually market shares and unit prices emerge with which each participant is comfortable, and which yield a satisfactory return. The move from monopoly to open market will take three or four years. The number and individual length of the intervening “plateau periods” will depend on the number and timing of new entrants. The steepness of the price fall in the transition periods will depend on the degree of aggression of the new entrant and the extent to which there is scope for cutting prices to obtain market share.

It is now necessary to interrelate these two features. First, it is in my judgment an inference properly to be drawn from the nature of the “at risk” market and the market dynamic that where there is a limited number of participants in an “at risk” market it is not in the interests of either the patent holder or the new entrant to drive down prices hard. So far as the patent holder is concerned that is because (as Servier explained in its argument in support of interim relief) downward price movements are irreversible if the NHS reimbursement price has been adjusted. (The pharmaceutical company sells its drugs to the pharmacist who then provides those drugs to the patient upon prescription by the doctor, but instead of the patient paying the pharmacist for the drug, the pharmacist is “reimbursed” by the NHS. The reimbursement is not at the price which the pharmacist has actually paid the pharmaceutical company but at some other (invariably higher) price fixed by the NHS; if the NHS becomes aware that it is paying £11 per unit for Perindopril but the pharmacist is only paying £1.50 per unit because of a number of competing suppliers it will (eventually) adjust its reimbursement price. The procedure for doing so categorises drugs. “Category C” is where there is only a branded product on the market: the Category C listing persists for some while after a single generic competitor enters the market, so that the NHS will reimburse pharmacists at the premium branded price even though a generic is present on the market. This is very satisfactory to suppliers and purchasers. But once there are several generics on the market the drug technically moves to another category, probably “Category M”, which warrants an immediate refixing of the reimbursement price determined by a volume weighted average of selling prices of the various suppliers). It is not in the interests of anyone in the pharmaceutical industry to bring about blatant circumstances in which the NHS re-pricing machinery might be invoked. That is as true of the new entrant generic drug supplier as it is of the patent holder. It is in everyone’s interests to keep the price as high as is compatible with their obtaining or retaining market share and generating the maximum profit.

The second aspect of the interrelationship between “at risk” period and the market dynamic in relation to which I make findings is the position of the patent holder in the face of challenge. There are a number of weapons the patent holder can deploy. If he thinks his right to the patent is not impregnable he can himself become a manufacturer to a generic drug supplier. His own original product (perhaps differently coloured or differently packaged) is then placed on the market in the name of some other company. This is called an “authorised generic” (as opposed to a “true generic” which is manufactured and supplied by a competing company). The entry of such agreements enables the patent holder, whilst continuing to support its premium brand, to make profitable additional sales (and thereby to avoid losing sales to competing “true generics”) and at the same time have some influence over the volumes in which and price at which generics are marketed. Furthermore, the patent holder can by means of confidential arrangements with its customers, adjust the true price at which its branded products are sold. By this means it can retain market share apparently at the “headline” price, and the apparent maintenance of the headline price will be reflected in published market data which will in turn influence the NHS reimbursement price. If the challenge to the patent is then seen off and the generic company is forced to leave the market, then the rebates in place during the “at risk” period can be withdrawn by the patent holder once the risk is eliminated. The patent holder thereby restores his margin and has to a substantial degree avoided any irreversible price decline.

Having described the general operation of the market I can now describe where the battle lines were drawn on the enquiry.

17A. I now turn to my detailed findings of what actually happened. At the request of Servier and without objection by Apotex some of Servier’s evidence was delivered on a confidential basis and parts of the hearing were conducted in private. I indicated that I was content to adopt this course for the purpose of getting on with the hearing, but that when I delivered judgement I would need to give a coherent public explanation of my reasons. In so doing, whilst it will be necessary for me to refer to specific commercial arrangements, the identity of the parties is not material. I will refer to the relevant true generic companies (i.e. those who manufactured and sold their own perindopril formulations) as “TG1”etc; to authorised generic companies (i.e. those whose sold perindopril manufactured by Servier) as “AG1”, “AG2” etc; and to those generic companies who sold perindopril manufactured elsewhere as “G1”, “G2” etc.

The EPO challenge to the 947 patent was well-known and Apotex was not the only generic company considering marketing generic perindopril in the UK: nor was Servier simply standing idly by. It is simplest to look at the intending market participants in turn and to adopt a thematic approach, even though this involves overlapping chronologies.

In November 2005 TG1 (an East European company) obtained UK marketing authorisations for 2 mg and 4 mg dosages of perindopril (though not the 8 mg dosage, which on my analysis of the actual sales figures would have constituted approximately 20% of the market). But it did not use those authorisations to launch “at risk” in the UK. It did however launch “at risk” in Hungary. On the 31 May 2006 (when the outcome of the EPO challenge was unknown) TG1 put Servier on notice that it might launch generic perindopril in the UK at any time after the expiration of 14 days. It did not in fact do so. On 28 June 2006 (the outcome of the EPO challenge being still unknown) TG1 then put Servier on notice that it intended to launch perindopril in the course of July. It did not in fact do so. On 28 July 2006 Servier commenced proceedings against TG1 for infringement of the 947 patent. TG1 opposed the injunction sought and filed evidence in those proceedings in which its witness professed that TG1 was “ready to launch” and had already begun taking orders (“pre-selling”). That evidence (filed in Servier’s proceedings against TG1) was simply produced as part of Servier's disclosure in the enquiry. No Civil Evidence Act notice was served in respect of it, and the witness himself was not called or made available for cross-examination. In the circumstances I attach relatively little weight to the material so far as it relates to TG1’s intentions.

The arrangements that TG1 had in place were (a) to distribute its perindopril in the UK through G1 and (b) to supply its perindopril to G2 (a German company which had also obtained marketing authorisations in the UK, though again only for 2 mg and 4 mg tablets). G1 was a small Irish company (Mr Watson QC described it as “a minnow”) offering a range of generic drugs through “on-line” direct sales to pharmacies: it apparently had no experience of launching products “at risk” and no history of selling to wholesalers. As to G2, under its arrangements with TG1 there would have been a four-month time lag between G2 requesting the supply and TG1 making the supply. According TG1's witness it was well-known that being the first or second generic product to enter the market was critical to capturing market share. Because of Apotex’ actual entry TG1 could hope for no better than second: and G2 would have been at a greater disadvantage. In August 2006 TG1 agreed with Servier not to launch its product in the UK in return for a cross undertaking in damages from Servier. On 1 September 2006 TG1 then made a defendant's application for summary judgement in respect of the claim by Servier, seeking revocation of patent 947. TG1 failed to achieve certainty by this means, and the action was ordered to proceed to trial. The action was then compromised on the 27 October 2006. The essence of the compromise was that TG1 withdrew its challenge to patent 947 and agreed not to sell in the UK in return for being permitted to market its generic perindopril in certain designated Central and Eastern European countries. In cross-examination Mr Langourieux of Servier acknowledged that TG1 was aware that it had “a lot of risk” in relation to a UK launch of perindopril and that they were more interested in countries where they did not have this risk.

Servier itself was also putting arrangements in place. I find that the context in which it was doing so was the pending decision of the EPO on the substantial objections to the validity of the 947 patent (not the anticipated immediate commencement of the sales of generic product by Apotex, which came as a surprise). The arrangements were to supply perindopril manufactured by Servier to two authorised generics.

In December 2005 Servier had entered into a licence agreement with AG1 relating to the supply of 2 mg and 4 mg doses of perindopril (so again excluding the 8 mg dosage). The date of the agreement shows that this was not a considered response to the particular threat posed by Apotex but rather an attempt to regulate the open market that would develop in the event that patent 947 should be declared invalid by the EPO: this much was acknowledged by Mr Langourieux in cross-examination. The agreement granted AG1 a licence for the commercialisation of a generic version of perindopril and bound AG1 to obtain its supplies from Servier. Under the pricing arrangements in the agreement perindopril was not supplied at a fixed price, but at a price which guaranteed AG1 a margin on the actual selling price, but subject to a floor price per unit. The agreement imposed no specific obligations upon Servier to supply perindopril: that was to be the subject of a separate supply contract. No supply contract effective as at August 2006 was in evidence: the evidence suggested that none was in place until February 2007 (and it then took effect only from a date ascertained by reference to the revocation of the 947 patent or the final determination of the proceedings between Servier and Apotex). Although there was no supply contract in place arrangements were under way to make limited supplies to AG1 at the time when the outcome of the EPO proceedings became known, whereupon those provisional arrangements were immediately terminated. Servier did not at the time it terminated the provisional arrangements know of any sales by Apotex. The termination of the arrangements was caused by the success in seeing off the challenge to patent 947.

In June 2006 (at a time when the judgement of the EPO was awaited and Apotex had yet to obtain its marketing authorisations) Servier entered into an agreement with AG2 for the supply of 2 mg and 4 mg dosages of perindopril (and eventual supplies of the 8 mg formulation) of in excess of 200,000 units per month. The arrangement was thus directed at Servier's participation in the generic market in the event that the EPO invalidated the 947 patent. Under the terms of the agreement AG2 agreed not to challenge Servier’s patent in the United Kingdom and not to import or sell generic perindopril. In return Servier agreed to provide stated quantities of perindopril by 1^st August 2006 at a guaranteed margin (subject to a floor price): but crucially Servier had the option to pay liquidated damages instead of actually effecting supply, and if it exercised that option then AG2 had “no other right or remedy (including any right of termination) in respect of a failure by Servier to supply…”. Because the agreement bound AG2 not to sell perindopril manufactured other than by Servier, but did not bind Servier to supply perindopril to AG2, it gave Servier the right to exclude AG2 from the market. Servier obtained this right by agreeing to pay AG2 £5 million, and further to pay £500,000 per month for each month of non-supply (irrespective of the amount of perindopril that AG2 would have ordered in that month). Servier was initially preparing to supply AG2 in anticipation of the revocation of patent 947 on the 27th of July 2006. But then in August 2006 Servier exercised the option not to supply, and continued to do so throughout the period for which the injunction against Apotex was in force. Servier thus paid AG2 approximately £10 million to keep it out of the market.

The evidence suggests that prior to August 2006 agreements were also entered into by Servier with G3 and G4: but Servier has refused to produce copies of those agreements (and an application to procure their production failed). Mr Langourieux would say only that they complied with competition law, that the G3 agreement included terms relating to the UK market and that G4 covered markets where Servier had a valid patent. In these circumstances I will assume that they were substantially in the same terms as that relating to AG2: if their terms reflected upon Servier’s case in some way more favourably than the terms of the agreement with AG2 I am sure I would have been shown the agreements. There was unchallenged evidence that G3 had been paid £10 million by Servier not to enter the perindopril market, and that G4 had been paid $20 million.

So much for the agreements relating to authorised generics and others, all of which were preparing the way for the possible invalidity of the 947 patent (and none of which was directed to responding to a market challenge by Apotex). But Servier also entered into agreements in direct response to Apotex’ entering the market. These were the brand equalisation or rebate agreements. On 2 August 2006 Servier entered into a rebate agreement with P to supply Coversyl at a headline price of £10.95 but an effective price after rebate of £7. On 4 August 2006 Servier entered into a rebate agreements with Q and with R to supply Coversyl at a headline price of £10.95 but an effective price after rebate of £6.25. On an unknown date in August 2006 Servier entered into an agreement with S to supply Coversyl at an effective price after rebate of £7.50. It was a feature of each of these agreements that it did not have immediate effect: it was for Servier to bring the discount into play when it chose, and likewise for Servier to terminate the arrangement. None of the agreements was actually brought into play.

As to Apotex itself I have recounted how Mr Darroch (managing director of Apotex UK Limited) personally collected the marketing authorisations. Apotex immediately began a telephone sales campaign on 28 July and commenced delivery of perindopril on 30 July 2006. Servier obtained interim relief as from 3 August 2006 but effectively on terms that orders already taken by Apotex could be fulfilled. In the 6 days between 28 July 2006 and 3 August 2006 Apotex sold 529,751 units at an average price of £7.79 yielding total net revenue of £4,128,788. It is the evidence of Mr Darroch (who has 30 years experience of assessing the market) that his company's sales amounted to a 60% market share. This is not accepted by Servier. They make two principal points. First, when a generic drug first enters the market there is a rush to purchase. The fact that a drug has in its first week gained a 60% market share does not mean that over the month it will have achieved a 60% penetration. Second, when a generic drug first enters the market there is a tendency to “stock fill” in order to build up inventories: the order placed is not for one month’s supply, but perhaps six or eight weeks’. Mr Darroch accepted that both of these points were theoretical possibilities, as did Apotex’ expert Mr Bezant. Servier could have adduced and put to Mr Darroch actual evidence to establish both points. It could have analysed its actual sales figures for July, August and September 2006 and made an accurate calculation of the size of the market (in which it was the only other participant), and computed Apotex’ actual market share. It could have produced the sales data for the customers supplied by Apotex but previously and subsequently supplied by Servier so that a comparison could have been made between the orders actually placed by such customers with Apotex and those ordinarily placed with Servier in a typical month. But it did not volunteer to do so and successfully resisted an application that it should. There is thus no evidence before me demonstrating that Mr Darroch’s assumptions are actually wrong in this case, though there is a theoretical possibility that they might be. In these circumstances I shall accept as a starting point Mr Darroch’s estimate of an initial 60% market share but in my overall assessment bear in mind that this is probably a generous estimate of the maintainable market share.

As to what the management of Servier itself was doing or preparing to do there is little documentary evidence: in particular there are no internal management reports or board papers which set out an analysis of the position, advance options for consideration, and record courses of action decided upon. The only document I found afforded an insight was an e-mail dated the 10 August 2006 by Patrick Lenoir (Mr Langourieux’s deputy). This of course was after the outcome of the EPO proceedings and of the injunction proceedings against Apotex was known. He wrote:-

“… even and if the EPO decision may further prevent other generic companies to market their perindopril, I think that we have to keep in our warehouse a certain volume of perindopril enabling us to react quickly if needed….. therefore I would suggest to keep the bulk tablets available, and pack them later if unfortunately needed, knowing that in the future, we made directly put on the market some perindopril with both [AG1] and [AG2] livery (once again, the later the better)…”

The evidence of Servier’s intentions in August 2006 is therefore to be found in the evidence of Mr Langourieux as to what was going on in his mind. This evidence was not entirely consistent. At some points he said in his evidence that in the event that Servier had failed to restrain Apotex by injunction then

“… we would have supplied the product once the generic came on the market. We would have supplied [AG1] and [AG2] and “short line” wholesalers. We would have supplied all customers willing to buy our generic…”

When the injunction was lifted Apotex recommenced sale of perindopril, now in direct and simultaneous competition with AG1 and AG2 (who were being supplied by Servier) and in circumstances in which the entry of other generics (who would not be “at risk” in the same way) was to be anticipated. It was the evidence of Mr Darroch that following the Apotex relaunch in July 2007 upon the revocation of the 947 patent customers were nervous that there would be other entrants to the market at any time, offering lower prices; and on that ground they sought significant discounts. One major customer (Boots) sought such a significant discount that Apotex was unwilling to accept it. It was under these conditions that Apotex achieved sales of 515,996 units between 9 and 31 July 2007 at an average price of £7.20, but falling in August 2007 to 201,726 units at £6.59, in September 2007 to 276,643 units at £6.25, in October 2007 to 285,900 units at £4.49, in November 2007 to 320,889 units at £4.28, in December 2007 to 122,347 units at £4.43, and in January 2008 to 164,544 units at £4.70.

It is Apotex’ case that when it relaunched its product in July 2007 it achieved only a 40% market share. In February 2008 there was then another significant change in the market because of anticipated new entrants. Neolab Ltd had obtained marketing authorisations for perindopril in its three formulations in October 2007 and launched its product in March 2008. Somex obtained marketing authorisations for the same three formulations on 21^st February 2008. In February 2008 the presence of market rumour had reduced the average unit price to £3.52 and by March 2008 to £2.03. Sundry other companies then sought marketing authorisations for generic perindopril and the market was from March 2008 undoubtedly an open market.

The three identifiable market phases were therefore (a) an “at risk” period which effectively ended in July 2007 (b) a transition stage from July 2007 until February 2008 as a few generics launched their products as soon as it became pretty clear that they did not infringe the 947 patent and (c) an open market from February 2008. (I have not overlooked the outstanding appeal, only resolved in March 2008: but the appeal court had let its provisional views be known in refusing to continue the injunction, so the “risk” was of an entirely different order). Apotex was kept out of the “at risk” market and forced to participate in the transition phase as a new entrant (along with all the others) rather than as an established market participant. I must estimate the financial consequences.

I reject the argument that the events which followed Pumfrey J.’s judgment and the refusal of the Court of Appeal to extend injunctive relief in July 2007 would have occurred in August 2006 so that all Apotex has lost it is 11 months’ of open market sales. The scenarios are entirely different.

First, in August 2006 any entrant to the market was “at risk”: the EPO had dismissed the challenge to the 947 patent, and the English court had not ruled on the issue. In July 2007 any entrant to the market knew that the 947 patent had been revoked (subject to appeal). The fact that a company chose to enter what had been established by judgment to be an “open market” is no guide to what it would have done in an entirely “at risk” market. Second, in August 2006 Servier must be taken to have held the view that its patent would eventually be proved good, and its conduct must have been influenced by that consideration. In July 2007 Servier knew that the 947 patent had been declared invalid and it had been refused injunctive relief by the Court of Appeal, and its conduct must then have been influenced by the fact that it was competing in an open market. What Servier actually did in and after July 2007 is not a sound guide to what it would have done in August 2006. Third, in August 2006 only Apotex was in fact selling to real customers: it had secured an actual head start in the race. Its commercial object would be to retain those customers, and any new subsequent entrant would have had to persuade Apotex’ customers to change generic supplier. In July 2007 the starting gate rose on all competitors at the same time. Apotex’ commercial object was to obtain or regain customers, at the same time as all other entrants were seeking to invite potential customers simply to choose between rivals. The outcome in July 2007 will not provide a guide to what would have happened in August 2006. Fourth, in August 2006 only Apotex had marketing authorisation and was in a position to sell perindopril in all three dosages: all other generic companies were either able to offer only a limited range or were dependent on eventual supply of 8 mg dosages by Servier. In July 2007 Apotex’ advantage in being the only alternative supplier of 8 mg dosages to Servier had evaporated.

The task in hand is therefore to reconstruct the hypothetical market in August 2006 not simply to identify the actual market in February 2008.

I begin my reconstruction of the hypothetical market with the observation that it is startling how different the picture presented on this enquiry has been from that which was presented to the judge when interim relief was sought. To get its injunction Servier's witness Mr Falcand (who did not attend the enquiry for cross-examination) stated:-

“I would expect Apotex to gain a major part of the UK market without delay. Apotex is well-positioned to achieve this aim as it is a large, well established generic manufacturer with sufficient capacity to supply the UK market”.

I begin with the consideration of whether anything may be regarded as certain. I regard it as certain that Apotex would have continued to supply the market and would have made a profit from so doing throughout the period for which the injunction was in force. The chance of profit was therefore 100%.

I now consider whether anything can be regarded as so speculative as not to warrant any weighting in the assessment of the lost chance. I consider the chance that TG1 would have entered the market in or shortly after August 2006 to be entirely speculative and therefore to be discounted. The nature of the evidence adduced to establish directly the intention of TG1 means that I can accord it little weight. Much more telling are the inferences that may properly be drawn from what is objectively known of TG1’s position assessed in the light of my general findings as to the market and its dynamics. Although it had marketing authorisations and although it threatened to enter the market it never in fact did so, even at a time when the 947 patent was under attack. It chose to make its threat using a very small distributor with no established track record of “at risk” launches. TG1 itself had no presence on the UK market. It understood the nature of the risk it would run by launching in the UK and had much greater interests in Eastern Europe (where it in fact did launch “at risk”) that would be jeopardised by an unsuccessful UK launch. It was using the threat to enter the UK as a bargaining counter to establish its position in Eastern Europe (in which objective it succeeded). That is why TG1 provoked Servier into the commencement of proceedings, and why in those proceedings it sought a summary determination, and why (when certainty could not immediately be achieved) it settled with Servier and withdrew its threat to enter the UK. It was unwilling to enter a market “at risk”. Until the Court of Appeal confirmed the revocation of the 947 patent TG1 would have been “at risk”, but not well placed to enter the “at risk” market. On the other hand, Servier was always willing to buy off potential entrants (as demonstrated by the deals it did with AG1 and AG2). TG1’s market weakness and its unwillingness to carry into effect its threat and Servier's willingness to buy it off mean that the prospect of TG1’s actual entry as a market participant is so speculative as to be ignored.

There are thus only two competing hypotheses. The first (“Scenario 1”) is that in the event that Apotex had not been restrained by injunction, and in the absence of any independent participants, Servier would have competed “head to head” with Apotex in an endeavour to preserve the position pending the establishment of the validity of the 947 patent and the restoration of Servier's monopoly. The second (“Scenario 2”) is that if Apotex had not been restrained by injunction Servier would have regarded the cause as essentially lost and would have operated on the footing that there was in practice an open market; it would accordingly not only have competed head-to-head with Apotex but would also have supplied AG1 and AG2, seeking to gain what advantage it could from the generic market.

If Apotex had not been restrained by injunction one or other of those scenarios would have occurred, and it is certain that either would have resulted in profit to Apotex. The question to be decided is: what is the likelihood of each occurring?

Although Mr. Langourieux was insistent for most of his evidence that it was Scenario 2 that Servier would certainly have implemented, in my judgment Scenario 1 was twice as likely to have occurred as Scenario 2.

First I do not consider that I can regard Mr Langourieux's evidence as reliable even though it was honestly given. In Allied Maples v Simmons & Simmons Stuart Smith LJ warned against placing reliance on the evidence of a claimant as to what he would have done in hypothetical circumstances: and I consider similar caution must be exercised in relation to the evidence of the defendant (see paragraph 11 above). One must measure what the witness now says he honestly believes he would then have done against such objective benchmarks as are available. One useful benchmark is the presumption (and that is all it is) that competent managers will act rationally and in the apparent commercial interest of the enterprise they are managing. Mr Purvis QC warned me against putting myself in the position of a notional Board of Directors without any of the overall knowledge of the market or business strategy which the actual board had, and making findings as to what decision would have been in the best overall interests of Servier. He also warned of the dangers of making any assessment that might be coloured by hindsight. I have sought to heed both points. But it is plain to me that Servier was a very canny operator: and equally plain to me that Mr Langourieux could offer no rational explanation as to why Servier should (at one and the same time) vigorously pursue its claim to a monopoly against Apotex in the apparent bona fide belief that the 947 patent was valid and seek to flood the “at risk” market with generic perindopril that would have the inevitable effect of destroying the very market that the patent exploited.

It is necessary to think both of the position in the interim (pending trial of Servier’s claim against Apotex) and after the trial (when Servier anticipated the establishment of the 947 patent). Servier’s preparations with AG1 and AG2 had been intended as a response to the revocation of the 947 patent by the EPO. When the EPO ruled in favour of the patent Servier immediately put those preparations on hold. Apotex then entered the market “at risk” earlier than Servier expected. Servier’s immediate response was to prepare “brand equalisation” deals in a price range of £7.50-£6.25, but not in fact to implement them. In my judgment, if no injunction had been granted Servier would first have competed “head to head” with Apotex using those and other brand equalisation deals (which maintained the headline price). Apotex came to market at £7.79 per unit and took 60% in the first 5 days (though that is a generous estimate of maintainable share). That was actually above the discounted price at which Servier was prepared to sell. So by using the brand equalisation deals Servier could have retained about half the market pending the ultimate establishment of its patent and achieved price stability. It could then prove a simple damages claim against Apotex when the 947 patent was ultimately established. It had no interest in driving prices down in the interim. Only one generic needed to be removed from the market if the patent was established. The premium price could easily be restored. Such a duopoly left TG1 exposed as the third participant in an “at risk” market and enhanced the prospects of Servier reaching a settlement with TG1.

To have supplied AG1 and AG2 from the outset would on its face have been commercial suicide. It would have engineered the very open market which leads to the irreversible downward price spiral. It would have encouraged TG1 to enter what was an established generic market taking control further from Servier and making settlement more difficult. It would have rendered damages claims against Apotex (and TG1) far more complicated (how many of Apotex’ sales were taken from Servier and how many from other participants? to what extent had Servier caused its own loss by supplying authorised generics and driving the price down?). It would have invited a re-categorisation of the drug for NHS re-imbursement purposes. It would have made re-establishing the monopoly if the 947 patent was held valid that much more difficult.

This analysis is supported by the expert evidence I received. Mr Bezant (the expert called by Apotex) estimated that if the duopoly was maintained Servier was likely to have earned about £25 million during the injunction period (assuming a 40% market share, which I think pessimistic) and could expect to have earned £66 million p.a if its monopoly had been restored. But if it promoted an effective open market by supplying AG1 and AG2 it would have earned only £17 million during the injunction period and (because of the irreversible downward spiral) could expect to have earned £36 million p.a. if its monopoly had been restored. This evidence was not seriously challenged.

The thrust of Mr Purvis QC’s argument was that this is all seen to be the case with the benefit of hindsight, and that matters may not have seemed that way at the time. He says that some allowance must be made for the possibility that Servier would not have regarded the prospect of a duopoly as realistic for two reasons. First, that if Apotex was not restrained by injunction it would itself have used its marketing authorisations (it obtained 15) to supply other generics. Second, that Servier could not be sure TG1 would not launch immediately if Apotex was not injuncted. That being so Servier would have armed itself for an inevitable open market and would have sought to take maximum advantage from the early generic market (even if at some risk to any restored monopoly). Depending upon market share it may indeed have been advantageous to sell heavily through AG1 and AG2 particularly if Servier had come to doubt the strength of its patent position as the case developed. He also submitted that discounting the branded product must be considered not only in the context of the UK market but also in the light of its knock-on effects in other markets (though Mr Langourieux’ evidence on this left much unexplained, in particular how confidential rebates that maintain the headline price could have any effect overseas). I cannot dismiss this possibility as so speculative as to be ignored. But I do regard it as improbable. To estimate the probability at 33% is I feel generous to Servier. But I am prepared to be generous on that account having dismissed as entirely speculative the chance of TG1’s entry. There is no point in attempting spurious precision.

How then to estimate the financial consequences of the two scenarios? In this I was assisted by the expert reports of Mr Bezant of LECG (for Apotex) and of Mr Andrews of KPMG (for Servier). Of the two I found Mr Bezant to be the more reliable, his report more coherent and his responses in cross-examination more cogent. By contrast, I consider that Mr Andrews made errors of principle in his efforts to reduce the “lost profits” (for example, an argument – which I need not elaborate – as to “double counting” of profits which ignored the fact that the period of profit earning is finite, and that time lost can never be recovered: and an argument about applying a discount rate to the lost profit calculation). But I do not dismiss Mr Andrew’s report (or his contribution to Mr Purvis QC’s searching cross-examination of Mr Bezant). Some of the assumptions made by the experts were founded upon material said to be confidential. Much of this material concerned transaction prices. Whilst parties may wish to keep the terms on which they conduct transactions private, pharmaceutical companies stand in this regard in exactly the same position as other litigants. I shall therefore include such information as is necessary to provide a public explanation of my reasons whilst avoiding any unnecessary exposure of business affairs.

I will begin with a consideration of Scenario 1. This is the duopoly in which after initial sparring the two market participants reach an accommodation as to market share and price. The first estimate that must be made is as to the unit price. It is known that Apotex commenced sales at an average unit price of £7.79 in July 2006 and resumed sales at an average unit price of £7.20 in July 2007 (though in an open market this was rapidly eroded to an average price of about £4.50 in the period October 2007 – January 2008). Mr Bezant initially assumed a constant unit price throughout the plateau period of £7.79. But I regard this as too high given that there is sound evidence that Servier would have been prepared to enter brand equalisation deals in the range £7.50 - £6.25. He proffered an alternative calculation founded on a constant unit price of £6.50, which I regard as a fairer assumption.

It is then necessary to estimate the volume of sales. This involves consideration of market share and total market size. Mr Bezant initially assumed that the 60% market penetration achieved in the first week of August 2006 sales would be maintained throughout the plateau period. I held above that that was a generous assessment, but that Servier had not used the means at their disposal to establish another figure. Servier contended that the true market share was only 30-40% (based on theoretical stock-filling). In closing Mr Bezant proffered an alternative calculation based on a 50% market share, and I will take that as a fairer estimate.

It is then necessary to estimate the size of the market. Is this constant, month by month, or are there significant variations? Mr Bezant assumed a constant market of 800,000 units per month. On the basis of the only independent material available to me (data collected by IMS Health, referred to in Mr Bezant’s report) and in the absence of any evidence-based analysis from Servier as to the total volumes actually sold, I consider this about right. The IMS data suggests that for each quarter that the injunction was in force the total market was slightly less than 2,500,000 units.

It is then necessary to consider what impact these assumptions have upon the actual events that occurred in July 2007. The plateau period would have been followed by a transition period (just as Servier’s monopoly was followed by a transition). But it makes no sense to proceed on the footing that the commencing price for the transition was £7.20 per unit (the price actually achieved) if one has assumed a preceding plateau price of £6.50 per unit. The transition period must have a different gradient. Once again Mr Bezant proffered a revised calculation to address this problem. Having done so he estimated the lost profits assuming a unit price of £6.50 and a market share of 50% (allowing for the different transition period gradient) at £23,400,000.

That is a calculated figure based on a number of assumptions. I must stand back and test it against such material from the real world as has a bearing. I know only from the evidence of Mr Langourieux that during the injunction period when Servier had a monopoly its sales amounted to about £74 million, effected at premium prices. A price reduction to £6.50 per unit amounts to roughly a 40% discount to the premium price. That would bring down total sales to approximately £45 million: if Apotex took 50% it would receive £22.5 million gross. This suggests that Mr Bezant’s estimate is broadly right, but slightly high. I shall therefore round down the Scenario 1 estimate to £22.5 million. If this scenario has a probability of 67% then the compensation payable for the loss of the chance is £15,075,000.

I turn to consider Scenario 2. In this scenario Servier engineers a controlled generic market which lasts until February 2008 (when Neolab and Somex enter). Apotex is competing against Servier, AG1 and AG2, (which is what actually happened in July 2007) but does so with the benefit of a head start (arising from its July 2006 launch). Mr Bezant approaches this on the footing that the actual pattern of sales made in the period following discharge of the injunction is used as the pattern for the first few months of the injunction period: and these actual figures are then followed by a period of assumed figures until the advent of the truly open market (anticipated in February 2008 and realised in March 2008 with the entry of Neolab). This is agreed. Mr Bezant then goes on to say that sales actually achieved in July 2007 and thereafter have to be adjusted to take account of that headstart and the undoubted advantage conferred by being the first generic on the market. This is not agreed. In his initial calculations Mr Bezant applied to actual sales from July 2007 a factor of 1.5 (based on Mr Darroch’s analysis that in July 2006 Apotex had a 60% share but in July only a 40% share of the market). After his cross-examination I am satisfied that this is an excessive uplift, both as a factor and in its application to the whole period.

Mr Purvis QC submitted that the cross-examination, coupled with a re-calculation of the July 2006 sales figure (to allow for stock filling) and a detailed analysis of Apotex’ September 2007 sales figures, established that there was no rational basis to any uplift at all. I reject that submission. As I have held above, stock-filling was a theoretical possibility but Servier led no evidence to establish that it had in fact occurred to the degree suggested: and the evidence clearly establishes that there was a genuine advantage to be gained from being the first generic on the market. In July 2006 Apotex was definitely the first generic on the market. AG2 could enter rapidly thereafter, and so too could AG1 (subject to agreeing a supply contract). Servier did not disclose its agreements with G3 and G4: and I will therefore make no assumption that they could have entered the market rapidly. This was not the market position which Apotex actually enjoyed in July 2007. If the July 2007 sales figures are taken as the raw data then there must be some uplift to reflect the loss of that “first entrant” advantage. This advantage would diminish with time (as more competitors entered the market). I therefore intend to apply to the actual July 2007 sales figures an uplift of 10% for the first 4 months.

In assuming a market in July 2006 in which Apotex is the first generic entrant, followed by AG2 and AG1, and in which Servier competes with brand equalisation deals, the July 2007 figures are only a starting point. They provide data from July 2007 until February 2008 (when the “Neolab effect” takes hold and the market becomes truly open). But this 7 months data has to be used to estimate the “but for” market for the 18 months from August 2006 (injunction) until February 2008. Both accountants approached this problem in the same way. One must assume a sales volume and a “plateau price” for the 11 month period.

Mr Bezant ultimately assumed a sales volume of 302,271 units and he assumed a plateau price of £4.90 (revised down from £5.25). Mr Andrews assumed a sales volume of 226,509 units and a plateau price of £3.53 (subsequently revised to £4, which he assessed to be the trend price). In the course of the trial Mr Bezant proffered a recalculation of the “plateau price” of £4.47 (being the arithmetic average of the prices charged from October 2007 until January 2008).

In my judgement the assumed sales volume should be 226,509 (itself an overprecise figure). It represents a 28% share of a total market of 800,000 units in which there are 3 generic and one branded participant. This seems to me to be reasonable.

In my judgement the assumed plateau price should be £4.47. £3.53 is the price obtained in February 2008 when rumours about the entry of Neolab affected the market. The assumed price must recognise the falling trend in prices (until September it was above £7: until November it was above £6). These were the initial phase of the transition period. But November 2007 to February 2008 shows a consolidation around £4.50 (reflecting the market participants reaching an accommodation on price and market share) and I think this steady state would probably have continued until the next market disturbance (the anticipated entry of Neolab).

Adjusting Mr Bezant’s calculations to take account of the reduced uplift I find that Apotex’ lost profits on Scenario 2 are £7,900,000. If this scenario has a chance of occurrence of 33% it is represented by damages of £2,607,000.

On my calculations Apotex is entitled to receive £17,682,000.

Recognising the imprecision inherent in the exercise but the precision of some of the assumptions used I have then stood back and compared that sum with the £74 million Servier earned during the period of the injunction to which it was found ultimately not entitled, to the £11 million which it paid to AG2 to keep it out of the market, and to the $20 million paid to G4; and I have asked myself whether in the round this sum overcompensates Apotex for the loss that it has suffered, reminding myself that the jurisdiction is compensatory not punitive. The range of figures presented for my consideration went from £400,000 (Servier) to £27 million (Apotex). I am satisfied that my figure is broadly right, though I would propose to round it down to £17.5 million to underline the fact that one can only do broad justice where there are so many significant variables. £17.5 million is accordingly the figure which I award as compensation on the enquiry.

Whilst this judgment was in preparation Servier made an application to amend its pleadings. On 2 July 2008 the Federal Court in Canada upheld (in proceedings commenced in November 2006 and to which Servier is not a party) the validity of a Canadian patent for perindopril (different from the 947 patent) held not by Servier but by an associated company. The perindopril which Apotex would have sold if no injunction had been granted would have been manufactured in Canada and would have been an infringement of the associated company’s patent. Servier sought permission to plead this fact and to raise the following issues:-

I am guided on the question of amendment by the oft-cited passage from the judgment of Peter Gibson LJ in Cobbold v L B of Greenwich. Servier say that the days spent on the enquiry so far and the lay and expert evidence filed may indeed prove to be fruitless, but that is simply something that can be compensated in costs. The Canadian judgment could not have been pleaded earlier: and since it establishes (pending appeal) that Apotex must pay somebody (albeit not Servier) damages (or render an account of profits) it would be unjust to award Apotex compensation on a footing which ignores the wrong committed in Canada. A Court of equity should decline to hear someone who has committed a wrong in Canada and is asking to be compensated for something that would not have been legitimate commerce. My attention was drawn to the decision of Scott J in Columbia Pictures v Robinson [1986] 3 WLR 542 in which the judge declined to enforce a cross-undertaking in damages in relation to a business that sold counterfeit tapes. They further submit that if I decide to proceed and to award Apotex compensation it will not be possible later to remedy what Sevier say is an injustice because it is not a party to the Canadian proceedings and because the relief available in Canada to the patent holder compensates it for sales actually made by Apotex, not for sales assumed to have been made for the purpose of assessing compensation for breach of the English cross-undertaking.

I decline to permit the amendment, and will decide the case as it was pleaded, evidenced and argued before me. Proper weight must be given to the whole of Peter Gibson LJ’s observation as to the circumstances in which amendment is allowed. Not only must it occasion no prejudice save such as can be compensated in costs but also the public interest in the administration of justice must not be harmed. Late amendment (particularly after close of evidence and the conclusion of closing speeches) can occasion prejudice which cannot always be compensated in costs and it significantly harms the administration of justice by allocating Court resources to (and allowing parties to incur expense in relation to) determining points which (if the party seeking amendment is right) the amendment renders irrelevant. Further, late amendment has the effect of keeping from a party the case he is expected to fight (and, of equal importance, to settle); it may impact not only upon the technical issues for decision but upon the whole conduct of a case. I consider the observations of Lord Griffiths in Ketteman v Hansel Properties [1987] 1 AC 189 at 220 to retain their full force, notwithstanding the advent of the CPR and notwithstanding that this litigation is part of the on-going international battle between two experienced contenders.

The Canadian court proceedings predate the commencement of the enquiry. If the underlying legal point is good it has been there since the beginning of the enquiry. The Canadian judgment has not brought it into existence. The Canadian proceedings are not at an end. The decision is not final. The application to amend is effectively an application to postpone the enquiry until final determination of the Canadian proceedings (including any Canadian enquiry). That application has been open to be made since the commencement of the enquiry. From the outset Apotex Inc has been a defendant to the proceedings and a claimant under the cross undertaking in damages. From the commencement of the inquiry it has been plain that under the price transfer arrangements existing between Apotex Inc and its UK associated companies Apotex Inc claims 90% of the profit. In the exchange of expert evidence there was specific reference to Apotex’ supply chain (and its internal transfer costs), but no suggestion was made that this raised any relevant issue. Furthermore Apotex’ computation of its production and supply costs (broken down into variable, fixed, semi-variable and stepped, and analysing the costs of the Canadian defendants) was accepted without challenge: there was no mention of some notional licence fee prospectively payable to a non-party in the event of a particular outcome to the Canadian case. The Canadian litigation featured in the evidence and in the argument before me, but no-one suggested that it might render the whole enquiry otiose. The first time it has been suggested that I might exercise a discretion not to proceed with the enquiry at all is after the conclusion of evidence and argument on the enquiry itself. This is simply too late: I do not think the point can be elaborated.

For these reasons I have delivered judgment in favour of Apotex in the sum of £17.5 million.

Mr Justice Norris………………………………………………………9 October2008