When Is an Invention Made?

Les Laboratoires Servier v Apotex Inc 2019 FC 616 Roy J
            2,423,825 / perindopril arginine / VIACORAM / NOC

This decision, as well another released last week, Aux Sable 2019 FC 581, invalidated significant claims on the basis of overbreadth. In both these cases the reasoning was, with respect, dubious, though in different ways. I suggest this is primarily due to defects in the doctrine itself. While the principle is clear enough – a patentee may not claim more than it has invented and disclosed – there is no clear methodology associated with the overbreadth doctrine. In particular, there is no well-established way of determining what is “the invention” which is to be compared with the claims. (Though that is not the main problem with the reasoning in the case at hand.) I have a draft paper on overbreadth which I hope to post later this week or early next week exploring this issue in more depth.

Turning to the case at hand, perindopril is used for treating hypertension and related cardiovascular disorders. Servier invented perindopril and received Canadian patent 1,341,196 for the compound per se [9], which it marketed as COVERSYL [10]. The 196 patent expired in March of 2018. In the meantime, Servier developed and received marketing authorization for the arginine salt (marketed as VIACORAM). This is protected by the 825 patent, which claims the arginine salt of perindopril and its hydrates as Claim 1 (“Sel d'arginine de périndopril ainsi que ses hydrates”). Apotex, wishing to sell a generic version of perindopril arginine, which it concedes will infringe the 825 patent, attacked the validity on a variety of grounds, including overbreadth [5]-[6].

A preliminary point of claim construction is helpful in understanding the overbreadth issue. Arginine is an amino acid, and it is a chiral compound which exists as both dextro- and levo-rotatory, forms, as well as a racemic mixture. L-arginine is the naturally occurring form [19]. While the disclosure identified the preferable form as L-arginine [19], [69], Claim 1 simply claimed “arginine.” Servier argued that this should be construed as L-arginine [159]. Roy J ultimately rejected this, essentially on the basis that the patent itself acknowledged the difference between the various forms of arginine, and to construe the claim as referring to L-arginine would read a limitation from the disclosure into the claim [172]-[179]. There was also an issue as to whether the reference in the claim to “hydrates” was restricted to crystal hydrates, or included any form incorporating water into the salt. Roy J again concluded the broader construction was correct [179].

The key reasoning on overbreadth is brief (my emphasis):

[205] The principle is simple enough on its face. A patent cannot validly claim more than has been invented. The claims may not be broader than the invention (Schering-Plough Canada Inc. v Pharmascience Inc., 2009 FC 1128). As usual, Professor Vaver captures in a few words what is at stake in his Intellectual Property Law, Irwin Law, 2nd Edition (2011):

The game for patent holders, especially in highly competitive industries, is to reveal as little and to claim as much as possible. The less disclosed, the more that can be retained as competitive edge. The wider one claims, the tougher it is for imitators and competitors. But the specification must stay clear of the known and the obvious. It must demonstrate and claim only something over and above existing technology. Much patent drafting involves trying simultaneously to achieve these aims. Along the way several obstacles must be avoided, lest the claims or the whole patent end up invalid. [p. 346]

This is all unobjectionable, but it does not provide any test for overbreadth. Schering-Plough 2009 FC 1128 has only a very brief discussion at [211]-[215], dismissing the attack on the facts, and Professor Vaver’s discussion is clearly not a legal test, but rather a discussion of the problem. No other authority is cited.

The key paragraphs follow:

[206] Apotex challenges the validity of the '825 on the ground that its claims are broader of the invention in two aspects: first, Servier claimed the arginine salt of perindopril, which includes the D-arginine and the racemate arginine salt of perindopril; second, Servier claimed the hydrates of perindopril arginine. In both cases, Apotex argues that there is no evidence in this whole record that Servier made the D-arginine or the racemate arginine salt of perindopril or that Servier made a hydrate of perindopril arginine.

[207] If either one of those obligations is accepted, it will be fatal to the validity of '825.

This statement is quite perplexing. On its face, Roy J is saying that if Servier claimed the D-arginine salt but never physically made it, then the claim is, for that reason, overbroad. This suggests that the test for overbreadth which he is applying is that a claim is overbroad if the patentee did not make every embodiment falling within its scope. This is confirmed particularly by the discussion on the facts of the hydrates, where Roy J emphasized that “The evidence in this case did not disclose that a hydrate of perindopril arginine was ever made" [229], and “there is no proof that Servier ever made a hydrate” [230]. The discussion of the D-arginine was concerned entirely with the claim construction point [210]-[222], of whether the claim to “arginine” encompassed D-arginine at all. It appears that Servier did not really address the overbreadth point as such [210] – perhaps implicitly conceding that the claim was overbroad if D-arginine had not been physically made?

I’m really not sure where this comes from. While overbreadth doctrine is not very well developed, the most prominent test is that a claim is overbroad if an essential element of the invention is omitted from the claim: see eg Amfac 80 CPR(2d) 59 (FCTD) at 80 aff’d 12 CPR(3d) 193 (FCA) at 196, 204. This was not the approach applied by Roy J; there was no discussion of the essential elements of the invention before concluding the claims were too broad. It is also very common that if a claim is overbroad, it is also invalid for some other reason, perhaps most often lack of utility, because in omitting an essential element, the claim encompasses something old, or useless. While Roy J did go on to discuss sound prediction of utility, this was after he had already concluded that the claims were too broad [222 re D-arginine, 231 re hydrates]. Moreover, the discussion of sound prediction was largely an aside, as “[i]t is not completely clear that Servier invoked the doctrine of sound prediction to justify the hydrates being included in claim 1" [241].

Perhaps the idea that a claim is invalid if it encompasses something which has not been physically made is derived from an overly literal reading of the famous statement by Thurlow J in Farbwerke Hoechst AG v. Commissioner of Patents, [1966] Ex CR 91 at 106:

There are two fundamental limitations on the extent of the monopoly which an inventor may validly claim. One is that it must not exceed the invention which he has made, the other is that it must not exceed the invention he has described in his specification.

But “made” in this statement does not mean “physically synthesized.” If it did, every genus patent would be invalid. A genus patent claims a broad range of related compounds, the vast majority of which will never have been synthesized, yet it is well established that a genus claim will be valid if there is a sound prediction of utility across its full scope, based on an extrapolation from the few compounds which were made and tested. Indeed, selection patents are possible only because so many compounds, and often the most desirable, falling within the genus were never made, so that the species selected is new, notwithstanding the genus. And every patent is a genus patent in some sense: a claim to “the wheel” encompasses not just the particular wheel that the inventor might have made and tested, but all wheels, including those made well after by some other person out of different materials. Saying a claim to arginine is overbroad because it encompasses D-arginine when the inventor had only made L-arginine, is like saying that a claim to the wheel is overbroad because it encompasses a wheel made out of birch when the inventor had made his prototype out of maple.

The broader point is that an invention is not a thing, but information: see Aerotel [2006] EWCA Civ 1371 [32]. The physical embodiments are merely a reflection of that information. As the SCC stated long ago, in Ball v Crompton Corset Co (1886), 13 SCR 469 at 477, quoting with approval the Supreme Court of the United States in Smith v Nicholls 21 Wall. 118:

A patentable invention is a mental result. It must be new and shown to be of practical utility. Everything within the domain of the conception belongs to him who conceived it. The machine, process or product is but its material reflex and embodiment.

Since an invention is information, making an invention is not a physical act. The question of when an invention was made arose regularly under the first-to-invent scheme of the old Act, and the invention was considered to have been first made by the person who “formulated, either in writing or verbally, a description which affords the means of making that which is invented”: Christiani and Nielsen v Rice [1930] SCR 443 at 456; and see also Ernest Scragg & Sons Ltd v Leesona Corp (1964) 45 CPR 1 at 31-32. That is, an invention is “made” when it is described – when the information has been provided which allows a skilled person to make and use it. This is why the disclosure, and not the physical embodiment, is the quid pro quo for the patent monopoly: Consolboard [1981] 1 SCR 504 at 517.

Perhaps I have misunderstood Roy J’s reasoning on this issue, but the proposition that a claim is overbroad if it encompasses embodiments which have not been physically made is supported neither by the cases nor by sound principle. This would not affect the result: as discussed in yesterday’s post, the claims at issue were also held invalid for obviousness.