Merck Sharp & Dohme Ltd v Ono Pharmaceutical Co Ltd & Ors

It now falls to me to decide what to do about the costs of this action. The main judgment is [2015] EWHC 2073 (Pat). In this judgment I will refer to the two sides as Merck and BMS. The conclusion was that the '878 patent is valid and would be infringed by Merck. There were two actions and it is agreed that BMS's costs in one action, which was the infringement action, should be paid by Merck and the amendment application costs should also be paid by Merck. The dispute is about what to do about the costs of the revocation action. The costs of these proceedings as a whole are extremely high, but that is no surprise given the difficulty and commercial significance of this case. Each side has spent a total of about £5 million litigating this dispute.

The dispute about costs arises because when this case began there were two patents. One is the '878 patent, which I have dealt with, but the other was patent EP (UK) 2 161 336. This patent had a slightly later priority date in 2005. It was filed in 2006 and therefore expires three years after the '878 patent. There was a question whether Merck's product infringed that patent. In the end what happened was that there was an amendment to that patent. As a result of that amendment, the parties agreed terms bringing the action in relation to the '336 patent to an end. They were embodied in a consent order which was approved by Nugee J on 27th April 2015.

That order contains terms as to costs as follows:

Having decided that the '878 patent is valid, in broad terms there is no dispute that the costs incurred after the relevant date when the '336 patent fell away should be paid by Merck to BMS. It is accepted that the Merck stopped incurring costs in relation to the '336 patent on 17th February 2015. The issue is what to do about the costs incurred before that date. There is no dispute by Merck that they should pay BMS's costs relating to '878. The question is what to do about general costs; that is to say, costs which cannot sensibly be attributed to one or other of '336 and '878.

Mr. Hinchliffe submits that the same problem came up before me a year ago in Vringo v ZTE [2014] EWHC 4475 (Pat) in which three patents were in issue. The court had given judgment on validity and infringement in relation to one patent and it was not disputed that costs relating to that should be awarded to the patentee, but there was a further patent in which earlier on claims for validity and infringement had been withdrawn, and then there was a third patent which had not yet been decided. What I decided to do in that case was to deal with four sets of costs separately. Three sets of costs were specific to each of the three patents, but the fourth set were the general costs, i.e. the costs not readily attributable to a particular patent. The general costs were dealt with by apportioning them in different periods. The apportionments were based on evidence before the court at that hearing, in proportion to the expenditure relating to the patents in issue during the relevant periods. That seemed to me in that case to be the appropriate thing to do.

Mr. Hinchliffe, who appears for Merck, submits that I should follow the same approach as was taken in Vringo and make an order, the details of which I will not set out in full, which has the effect of apportioning general costs incurred prior to the date of February 2015 in such a way that Merck pay a proportion of those general costs based on the ratio of their costs relating to '336 and their costs relating to '878 and BMS pay the equivalent calculation of the general costs based on the proportion of BMS's costs of '878 and '336. BMS submit that I should not make this order for essentially two reasons.

First, BMS submits that the principle to be applied, which was applied in Vringo v ZTE, is that the court needs to decide who was the overall winner, and then apply the principles normally applied to costs in patent cases (Monsanto v Cargill). In other words once an overall winner has been identified the court should make such discounts or other orders if there are suitably circumscribed issues as to which costs can fairly be dealt with in a different way from the normal way. The normal way being that the successful party's costs are paid by the unsuccessful party. Ms. Edwards-Stuart, who appears for BMS, submits that in this case there is no reason to make a discount or other order. The purpose and effect of the order proposed by Ms. Edwards-Stuart is that general costs incurred prior to 17th February 2015 would be paid by Merck to BMS. The only costs which would not be paid by Merck to BMS would be costs incurred by BMS which can be attributed to the '336 patent. That would be because the consent order made by Nugee J has already dealt with those costs.

Second, Ms. Edwards-Stuart also submits I should make this order for a further reason: in practice the apportionment proposed by Merck would unfairly produce a windfall to Merck as a result of the way in which costs were incurred in the period prior to 17th February. That is due to the timing and extent of disclosure and to the fact that experiments were conducted which related to the '336 patent. It was Merck's belief that it was those experiments which led BMS to amend the '336 patent. I am in no position to say whether that is right or not. Nevertheless the experiments may explain why substantial costs were relating to '336 before that patent fell away.

The difficulty raised in the evidence of Ms. Noor for Merck is about disclosure. The costs relating to disclosure alone are about £1 million. Ms. Noor is very concerned about the difficulty in being able to tell whether much of the disclosure can be attributed to one or other patent. Her evidence is that this is impossible in many cases. So there is a concern that a very expensive disclosure exercise, which Ms. Noor submits was unnecessary in some respects, will end up falling to be paid for by Merck when that would not be fair. That is one of the main reasons for the apportionment order from Merck’s point of view.

On the other hand, Mr. Watts, who gives evidence for BMS, says that in fact the disclosure can be apportioned, at least to a significant extent. Ms. Edwards-Stuart submits that I cannot resolve that conflict now and therefore the court should not make an order which is on the basis that Ms. Noor's position is right.

This is a difficult problem. I have no doubt of the sincerity of either Ms. Noor or Mr. Watts; in fact I go further than that, I rather think the reality is that it will at least be extremely difficult to trawl through every disclosure document individually in order to allocate it to one patent or another. It will be possible to allocate an awful lot of them, but there is a high chance it would be an incredibly disproportionate task to have to do it for everyone. If the court can find a way of making a fair order, even if it involves significant element of approximation, and what could be called rough justice, that would seem to me to be an appropriate thing to do.

Another aspect of the problem is that the consent order already made provides for an assessment of Merck’s costs on 336. Given that BMS clearly won the trial on 878, each side’s costs will now be the subject of a detailed assessment. If the trial had been about both '336 and '878 and the outcome had been the same – in other words that '336 should be amended in the way it was and 878 decided in the way it has been - then it would have been right to regard the conclusion on 336 as a win for Merck, and the conclusion on 878 as a win for BMS. Nevertheless even in those circumstances it is not inevitable that the Patents Court would have made an order which committed both sides to detailed costs assessments of their own costs. Clearly Merck is not the overall successful party. Either the overall winner in the litigation is BMS or else there is no overall winner. If it is right to regard BMS as the overall winner then the 336 patent would have been seen as a suitably circumscribed issue on which Merck had won and a substantial discount would be made as a result, based on a percentage. This illustrates one of the important reasons why the Patents Court routinely makes orders based on a percentages. They are inevitably approximate but in a case with a properly identifiable winner they do a significant measure of justice because they avoid the problems, which the court is now faced with, of two detailed assessments having to go on in parallel.

The right way to resolve this problem is to start with the same question which I asked myself in Vringo, and to consider whether I can identify an overall winner in this case. That is what the general provisions on costs in CPR rule 44 require. The general rule is that the successful party's costs will be paid by the unsuccessful party, but the court may make another order. I refer to paragraphs 6-10 in Vringo in which I applied that principle and came to the conclusion that it was not sensible to regard a victory on one patent in that case as meaning that one side could be regarded as the overall commercial winner of the entire dispute. That is the basis on which I made the apportionment order in Vringo.

Mr. Hinchliffe submits that the same logic applies here: the fact that the '336 patent was amended in the way that it was is of very significant commercial significance to Merck, particularly bearing in mind the fact that the patent has three more years to run. Ms. Edwards-Stuart submits that while that is no doubt true, on the other hand, looking at the position today, it is clear that the commercial winner of the dispute as a whole, particularly having regard to all the costs that have been incurred, is BMS and Ono because the '878 patent has been found to be invalid and infringed. The product which is in question as far as the alleged infringement is concerned is the Keytruda product.

I remind myself of the words of Lord Bingham MR in the Roache v News Group Newspapers in which he said as follows:

I also remind myself of the judgment of Floyd J, as he then, was in Omnipharm v Merial [2012] EWHC 172 (Pat).

The problem here, as in Vringo, is that there are two distinct patents in the case and the outcomes in relation to them were different. Nevertheless, in my judgment, when thinking about the litigation as a whole and thinking about the issue as a tool for identifying the fair way to deal with the costs, it is not right to say there is no winner. BMS should be identified as the overall winner. In that sense this is a different situation from the situation in Vringo. It would be necessary and appropriate, if all the matters were before me, to make an apportionment and a reduction relating to the costs of '336 but I do not have to do that because the consent order has already provided that the costs relating to '336 are to be paid by BMS.

With that in mind, I turn to the question of whether I should attempt to make any apportionment of the general costs in this action. I will not. I am not attracted to the idea that those costs are sufficiency circumscribable so as to be able to be apportioned in a fair way bearing in mind the overall outcome in this case. Accordingly, I will not do it.

A further reason why I will not do it is that it risks undermining the consent order because it risks attributing to one party costs of '336 which the consent order has already provided for. I appreciate that apportioning the general costs is supposed to deal only with unattributable costs, but the trouble with “general costs” is that there is an elasticity about the concept. They could be costs which it is impossible to attribute or costs which it is just disproportionate to attribute. There will be a difference between those two things. The consent order will require the parties to work out for any given item (whether costs of Merck or the costs of BMS) whether those costs are to be attributed to the '336 patent anyway. That is already a difficult task. To go any further would make things worse.

For all those reasons, I will make an order which has the effect that costs which are not attributed to '336 and which were incurred before 17th February will be paid by Merck. The reason is because ultimately by Merck lost the proceedings overall and it does not make sense to try and apportion them in any other way. I see that what this may mean is that it is in Merck's interests to go to the trouble of attributing costs to the '336 patent if they fairly can. If they can and the costs judge decides that that is right, then they will recover those costs in accordance with the consent order. That is what I am going to do.