Friday, November 23, 2018

UKSC decision in Warner-Lambert v Actavis – Infringement of Second Medical Use Claims

Warner-Lambert Company LLC v Generics (UK) Ltd (t/a Mylan) & Anor [2018] UKSC 56 (UKSC) Lords Mance, Sumption, Reed, Hodge, Briggs var’g [2016] EWCA Civ 1006 Floyd LJ: Patten, Kitchin LLJ var’g [2015] EWHC 2548 (Pat) Arnold J
            EP(UK) 0934061 / pregabalin / LYRICA

The decision of the UKSC in Warner-Lambert v Actavis is important in UK law both for its discussion of the concept of “plausibility” and in respect of infringement in the context of a second medical use patent. This post provides an overview and discusses the issue of infringement, which raises a truly difficult problem, of how to balance the need for an incentive to develop new indications of a known drug, against the need to allow unfettered availability of the same drug for prior indication. I will argue that Warner-Lambert v Actavis does not provide a solution; and it is unlikely to have much impact in Canadian law, except, perhaps, as a cautionary tale about how the well-intentioned restriction on patentability of methods of medical treatment has gone badly awry.

Pregabalin is indicated and marketed by Warner-Lambert for the treatment of epilepsy, general anxiety disorder ("GAD") and neuropathic pain [Pat 2]. Warner-Lambert’s patent EP0641330 claimed pregabalin as such and for use in treating epilepsy and GAD. That patent expired in 2013. The key claims of Warner-Lambert’s 061 patent were Swiss-form claims to pregabalin for the treatment of pain, inflammatory pain and neuropathic pain (Claims 1-3) [5]. Claim 3 was the most important, as neuropathic pain is the most common type of pain for which pregabalin is indicated [7], [13]. Mylan and Actavis, seeking to launch a generic pregabalin product, brought separate actions for revocation of the 061 patent, which were consolidated [Pat 3]. After both revocation actions had been started, Actavis decided to launch with a skinny label; that is, with marketing authorization and corresponding product information targeting only the non-patented uses. Warner-Lambert brought an action against Actavis in respect of that skinny label launch [4]. The market at the time was something like 1/3 for unpatented uses, and 2/3 for patented uses [Pat 399-407,425-42]. (The exact numbers are not important, except to establish that both patented and unpatented uses were significant.) Note that Warner-Lambert is a company in the Pfizer group [4], and Warner-Lambert markets pregabalin, under the brand name Lyrica, but Pfizer holds the marketing authorization [Pat 2]. Nothing turns on this, but many of the facts refer to steps taken by Pfizer rather than Warner-Lambert.

By the time the case arrived at the UKSC, there were four issues to be addressed [14]. These (in the Court’s numbering), were:

(1) Claim construction. The key issue was whether Claim 3, to “neuropathic pain” includes all neuropathic pain, or only peripheral neuropathic pain. The question was important because Arnold J had found that the disclosure of the specification was sufficient to support the claim that pregabalin was efficacious in the treatment of peripheral neuropathic pain, but not central neuropathic pain [10]. If the claim was limited to peripheral neuropathic pain, then, in Arnold J’s view, it would have survived this validity attack. All levels of court were agreed that Claim 3 extended to both central and peripheral neuropathic pain [15(1)]. The claim construction point was moot at the UKSC level, which held the disclosure insufficient even in respect of peripheral neuropathic pain. This aspect of the discussion largely turns on the facts and doesn’t have significant implications for Canadian law.

(3) Amendment. After Arnold J’s decision holding the claims at issue invalid, Warner-Lambert sought to amend Claim 3 to limit it to peripheral neuropathic pain, pursuant to s 75 of the Patents Act 1977. Arnold J refused to allow the amendment, and this decision was unanimously affirmed by both the EWCA and the UKSC: [15(1)]. As this type of amendment during litigation is not permitted in Canadian law, this raises no issues of interest.

(2) Sufficiency of the disclosure in the specification. This is the issue of “plausibility” which I will address elsewhere.

(4) Infringement
Infringement was argued on the basis that Claims 1 and 3 were valid, but that Arnold J’s construction — that it encompassed all neuropathic pain — was correct. Because the infringement action concerned the skinny label launch by Actavis, the question is the extent to which sale for non-patented uses would infringe. It was established on the facts that some use for the patented indications was inevitable, even with a skinny label, essentially because the prescribing doctor will normally prescribe generically, without stating the indication [Pat 376-80], and the pharmacist will not normally know what indication the drug is prescribed for [Pat 384-89].

As an aside, we may note that if the drug is prescribed by brand name, the pharmacist would not substitute [Pat 384], so in theory, if physicians could be sufficiently educated to prescribe only by brand for the patented indications, this would solve the problem. On an application by Pfizer, Arnold J issued an order requiring NHS England to issue guidance to GPs that only Lyrica branded pregabalin should be prescribed for neuropathic pain [Pat 510]. This order apparently had some effect, but it did not completely solve the problem, and it remained foreseeable that substantial use for the patented indications would continue [66]. This order was not relevant to the test for infringement applied by Arnold J [661] or at the UKSC level, but would have been relevant to the test endorsed by the EWCA [CA 217] [Pat 667]. The UKSC also noted that it would not always be appropriate to address the problem with orders of this type, as prescribing practices have been developed for good reason – presumably medical – and should not lightly be discarded [66].

Thus, in the end, the facts were quite a standard scenario of a skinny label launch where the product information does not encourage infringement. (Warner-Lambert objected to one aspect of the label, but Arnold J dismissed the objection [Pat 443], and it played no role in the reasoning at any level.) In Canada, these types of actions are normally brought on the basis of inducement. The parallel to inducement in the UK is s 60(2),(3) of the Act (though the law is different in several respects from Canadian inducement). This was run only as a secondary argument in Warner-Lambert. The infringement argument was primarily based on s 60(1)(c), which provides that a person infringes if “where the invention is a process, he disposes of. . .any product obtained directly by means of that process.” This codifies the Saccharine doctrine – the product of an infringing process is infringing. The argument was run that way because it was common ground that Swiss-form claims are process claims, albeit “purpose-limited process claims” [63], and it was undisputed that the generic pregabalin was the direct product of the process set out [Pat 601]. Since the claims at issue were Swiss-form claims, they were process claims, so the product itself is infringing, if it is made “for” treating neuropathic pain.

Infringement therefore turned not on the interpretation of s 60(1)(c), but rather on the construction of “for” in a Swiss-form claim [71]. This is a crucial point, which was emphasized at all levels of court [88], [121], [EWCA 189]: [Pat 92]; Swiss form claims are no longer granted by the EPO (G 2/08), and instead claims are now normally granted as "product X for treating indication Y," which are “purpose-limited product claim” [121], [Pat 91], not, like a Swiss form claim, a purpose-limited process claim. Consequently the reasoning and result might well be different under the form of claim now permitted by the EPO.

While Court was unanimous that there was no infringement on the facts, the reasons were very different. Lord Sumption (Lord Reed concurring) adopted an objective test, the “outward presentation” test, under which
[84] the badge of purpose is the physical characteristics of the product as it emerges from the relevant process, including its formulation and dosage, packaging and labelling and the patient information leaflet which in EU (and other) countries will identify the conditions for whose treatment the product is intended.
That is, so long as the product packaging and accompanying information do not instruct infringement, there is no infringement.

Remarkably, on Lord Sumption’s approach, there is no infringement even if the generic subsequently directly markets the product to dealers and pharmacists for the treatment of pain [86]. Lord Sumption recognized this consequence as “imperfect” but felt it justified by the need for an objective test. (Lord Mance also adopted an objective test, but couldn’t bring himself to fully commit to this rather extreme position [217].) Moreover, Lord Sumption held that neither would liability arise by way of indirect infringement under s 60(2). That subsection provides that it is an infringement to supply an essential element of the invention to a direct infringer, knowing it will be used to infringe (broadly corresponding to infringement by inducement in Canada) [87]. 60(2) is not infringed because Claim 3, as a Swiss form claim, is to the pregabalin itself, and not to its use for treating neuropathic pain, so there is no direct infringement to support the claim for indirect infringement [88].

Lord Briggs (with Lord Hodge concurring on this point), agreed with Arnold J that the appropriate test is whether the alleged infringer subjectively intended to target the patent-protected market. Subjective intent could be established by “all forensic means whereby a purpose of the generic manufacturer to serve (and profit from) the market for neuropathic pain could be proved, including but not limited to the packaging on the product. Anything from which the court could properly find that the manufacturer had such a purpose could be relied upon, including targeted disclosure, during litigation, of documentary records of the manufacturer’s decision-making processes” [172]. (On the facts, Actavis did not have the requisite intent: [Pat 661].

Implications for Canadian law
In Canadian law this type of scenario is dealt with under the law of inducement, which (more or less) boils down to saying that a generic is permitted to sell the product for the unpatented purposes, but cannot encourage the use of the product for infringing purposes, whether by the product monograph or by direct marketing. The Canadian approach is not perfect. (A perfect rule would ensure that the generic product is not used at all for patented indication, while remaining freely available for unpatented indications.) But perfection is probably unattainable in this area, and the Canadian approach is, in my view, clearly superior to any of the approaches adopted by the UKSC in Warner-Lambert. The Canadian approach avoids the glaring “imperfection” of Lord Sumption’s approach, which would allow a generic to market with a skinny label and then use direct marketing to encourage infringing use. It is also preferable to Lord Briggs’ approach which has a number of problems, pointed out by Lord Sumption [75]-[78], and acknowledged by Lord Hodge [188]. One is that whether the product is infringing depends on matters which may not be within the knowledge of the downstream parties, such as physicians and pharmacists (as it may turn on the manufacturer’s intent as revealed by internal documents) who might therefore infringe without the possibility of knowing they were doing so. Moreover, it would seem that because the generic cannot know which particular pill might be dispensed for the patented purpose, if the generic has the requisite intent, the entire production run will be infringing product. This means that a downstream party, such as a pharmacist who dispenses it, will infringe, even if the product is dispensed for a non-infringing use.

The approach of the EWCA is much more attractive from a policy perspective than either of the approaches endorsed by the Lords. The EWCA, per Floyd LJ, Patten and Kitchin LLJ concurring, applied an objective foreseeability approach, with a very important qualification. More specifically, the EWCA started with the proposition that the generic drug would infringe if it was foreseeable that the generic product would be used for infringing purposes [EWCA 206]. (Doctrinally, this was based on the view that a person intends the reasonably foreseeable consequences of their actions.) This in itself goes far beyond Canadian law, as it would effectively preclude the generic from marketing the product for non-infringing use if any substantial degree of infringing use was inevitable. But the important qualification is that this imputed intention could be negatived “where the manufacturer has taken all reasonable steps within his power to prevent the consequences occurring” [EWCA 208]. This brings the EWCA approach much closer to the Canadian approach: in the Canadian approach the generic can sell so long as it does not encourage infringement, while under the EWCA approach, the generic can sell so long as it discourages infringement. Depending on what “reasonable steps” might be on the facts, these approaches might even amount to the same thing. I note also that the EWCA, in particular Floyd LJ and Kitchin LJ, are more experienced in patent matters than any members of the panel in the UKSC.

The UKSC unanimously rejected the EWCA approach, essentially on the basis that it was impermissible judicial legislation, effectively creating a defence to infringement, and so going beyond what could be done without a statutory amendment [81], [160]. Perhaps that is true as a matter of UK law, but I would suggest that the distance between the current Canadian position and the EWCA position is not so great that it would be impossible to bridge through an evolution of the law relating to inducing infringement. Of course, even if possible, that would only be desirable if the EWCA position were actually preferable to the Canadian position as a matter of policy. I won’t explore those questions further in this post.

The proximate cause of these difficulties at the UKSC is the Swiss-form claim, which, as we have seen, was treated by the Court as a process claim, with substantively different scope from either a purpose-limited product claim, or a use claim. In one way, this is sensible; on its face, as a matter of form, a Swiss-form claim looks like a process claim. But in Canada, nothing is normally made of the distinction: see eg AB Hassle 2002 FCA 421 [5], re 2025668 treating Swiss-form claim 1 and use claim 2 on the same basis. Treating the different forms equivalently is also entirely sensible, and arguably more sensible, as it is universally acknowledged that Swiss-form claims were accepted by the Enlarged Bord of Appeal in G 5/83 [1985] OJ EPO 64 as a means for avoiding the restriction on patenting of methods of medical treatment set out in EPC 1973 Art 52(4), specifically in the context of second medical use claims: [2011] EWHC 1831 (Pat) [44] -[54]; [2008] EWCA Civ 444[14]-[29]. (To make things more convoluted, a first medical use of a known compound could be claimed as a use claim, and Swiss form claims were only strictly necessary for a second medical use: [2011] EWHC 1831 (Pat) [46].) This restriction was removed, or at least modified, by the EPC 2000, so that a claim to a use of a known compound for a treating a disease is now acceptable. In so holding, the Enlarged Board of Appeal in G 2/08 noted that the decision of the EBA in G 5/83 approving the use of the Swiss-type format, “had filled a gap in the legal provisions,” and “Under the new law, EPC 2000, the lacuna in the old provisions, which had been closed [by G 5/83] no longer exists” [5.10.1-2]. The relevant explanatory notes confirmed that the new provision “is intended to match as closely as possible the scope of protection to the scope provided by a 'Swiss-type claim'.” [5.10.4]. Because “the loophole existing in the provisions of the EPC 1973 was closed,” [7.1.2] there is no reason for the rule allowing Swiss-form claims, and such claims are therefore no longer permitted by the EPO [7.1.3]. If the Swiss-form claims were always intended to close a loophole, it seems formalistic to treat them as substantively different from the claims which are now used to fill that role directly, and which were intended to match the effect of Swiss form claims as closely as possible. I acknowledge that this formalistic approach was not unique to the UKSC: the EPO had always recognized that the effect of the new claims might be broader: G 2/08 [6.5]; T 1780/12 [19]-[24]. Nonetheless, I hope that Canadian courts will continue to ignore these formalistic distinctions in an area which is already more complex and formalistic than it needs to be.

Of course, Canada has its own prohibition on methods of medical treatment, albeit one which is judge-created rather than statutory, which is somewhat different in effect from the European version. The US has no such prohibition. This results in the following types of claims, all aimed at fundamentally the same thing, a second medical use:

(1)       Swiss form claims: “Use of compound X in the manufacture of a medicament for the treatment of indication Y”
No longer accepted in the EPO, accepted but not necessary in Canada, not used in the US.
(2)       "Compound X for use in the treatment of disease Y"
Accepted by the EPO (under EPC 2000).
(3)       “Use of compound X for the treatment of disease Y"
                        Accepted in Canada; not accepted by the EPO;
(4)       "A method of treating disease Y, comprising administering compound X."
Accepted in the US, not accepted in Canada or the EPO:

If the proximate cause of the difficulties on display in Warner-Lambert is the Swiss-form claim, the ultimate cause of this morass is the prohibition on patenting of methods of medical treatment itself. It is widely acknowledged, in European law at least, that the rationale for this prohibition is to “to leave the physician free to act unfettered” G 2/08 [5.7] and to “protect the autonomy of clinical judgments” [82.4]. That is, a physician should be permitted to treat her patient according to her best medical judgment, without having to worry about the threat of a patent infringement lawsuit.

If it is necessary to protect the autonomy of physicians, the appropriate way to do so is with a defence, which protects the physician directly in appropriate circumstances. A prohibition on patenting of methods of medical treatment is grossly overbroad as a means for protecting physician autonomy. It is widely acknowledged that patents should be granted for new medical uses of a known compound, as new uses, such as for AZT in treating HIV/AIDS, or sildenafil in treating ED, can be very important advances in which the patent incentive plays a major role: [Pat 88]. It is the need to reconcile this important patent incentive with the prohibition on direct patenting of methods of medical treatment that has led to the endless doctrinal contortions we have witnessed in European law, and, to a somewhat lesser extent, in Canada. Moreover, the difficulties are not over. Actavis asserted that the new claims allowed by the EPC 2000 will cure most of the problems associated with the Swiss-form in the longer term, but “Whether that is right remains to be seen” [123]. It is not clear to me that the problems will be solved, as such a claim is still not a use claim: it is a product claim, rather than a process claim [3], [121], and I’m not sure why the problems under 60(1)(a) will be any less intractable than those under 60(1)(c).

At the same time as it is overbroad, the prohibition on patenting methods of medical treatment is also grossly underinclusive as a means of protecting physician autonomy. Because it does not protect the physician directly, if the product is indeed infringing, a physician who uses it will be an infringer.

So, the prohibition on patenting methods of medical treatment is completely inadequate as a cure for the problem of protecting physician autonomy. What is worse, the problem itself does not exist. Physicians are not protected in the US, and yet it is unheard of for a pharmaceutical company to bring an infringement action against an individual physician. (The defence under 35 USC § 287(c) applies only to surgical techniques.) Nor have I heard it suggested that this was a live problem in European law prior to the EPC 1973. The concern is entirely theoretical.

There would not be anything wrong in preventing a purely theoretical problem, if the cost were low. But it is not. There is a real cost, both in terms of litigation costs, complexity of the legal system, and uncertainty for the parties, arising from the doctrinal minefield of the restrictions on patenting of methods of medical treatment.

What Warner-Lambert really shows is that the prohibition of patenting of methods of medical treatment is a cure that is much, much worse than the disease. And indeed, it is not even a cure at all. It is as if, in order to guard against the possible release of the last smallpox sample from the US government lab, everyone in the world was vaccinated with a vaccine which caused serious side-effects – and which was not even effective against smallpox.

The problem of infringement of second medical use claims is truly difficult. How can the law balance adequate protection for the patentee in respect of the new use, while allowing unrestricted sale for the old use within the context of our medical system? There is probably no perfect solution to this problem, and there is certainly no easy solution. Tackling this difficult problem will only be more difficult if the law is hamstrung by the baroque distinctions and restrictions illustrated by Warner-Lambert.

1 comment: