Tuesday, November 13, 2012

Disclosure Is the Quid Pro Quo

Teva Canada Ltd v Pfizer Canada Inc2012 SCC 60 LeBel J for a unanimous court,* rev’g 2010 FCA 242 Nadon JA: Blais CJ, Trudel JA rev’g 2009 FC 638 Kelen J
            2,163,446 VIAGRA

I wish I could say that my delay in writing this post has been because I was carefully pondering the implications of the decision, but the truth is that I was away at a conference (in Shanghai – a very interesting and dynamic city), and I have only just gotten to reading the decision. So these are my initial thoughts.

With that caveat, I do not see the SCC Pfizer / sildenafil decision as making a major change in the law. There is a tension in the disclosure requirement between the principle that the patentee must make full and complete disclosure, and the principle that some trials and experiments by the skilled addresses are permissible. The Federal Courts emphasized the latter principle, and the SCC emphasized the former. On the facts, this patent was close to the line, and either emphasis was reasonable. The disclosure was close to the line because the drafter was evidently deliberately pushing the boundaries of the disclosure requirement. Pfizer was apparently trying to high get a patent while trying to hide the effective compound from competitors; as Kelen J put it in “[t]e patent plays ‘hide and seek’ with the reader” [FC 135]. In my view, the SCC decision does not move the line, but rather clarifies where it lies. The decision should not have a broad impact, unless there are many other patents out there that are similarly trying to “‘game’ the system” [80]. In the end, the SCC holding is straightforward: the quid pro quo for a patent is disclosure. The patentee cannot have its cake and eat it too; it cannot keep its secret and its monopoly as well. The patentee tried in this case to minimize its disclosure, presumably to delay competitors from entering the market with patentable improvements. This is simply contrary to the disclosure requirement. Yes, competitors may be able to use the information disclosed, even prior to the end of the patent term. But disseminating information about the invention so others may make use of it, during the term as well as after, is the very purpose of the disclosure requirement.

Turning to the specifics, the patent disclosure starts by noting that the invention relates to compounds for the treatment of impotence. It discloses compounds of “formula I” which encompasses 260 quintillion individual compounds. It then specifies “preferred, “more preferred,” “particularly preferred” and "especially preferred” individual compounds. The last category lists nine compounds. The disclosure then states “Moreover, patient studies conducted thus far have confirmed that one of the especially preferred compounds induces penile erection in impotent males". However, the disclosure does not specify which of the especially preferred compounds this is [FC 114]. The patent includes 27 claims. The first seven claims are cascading claims to the use of a specified compound to treat erectile dysfunction. Claim 1 claims compounds of formula I, with the subsequent claims to selections from that class, down to claim 5 which specifies the nine especially preferred compounds, and claim 6 and 7 each claiming one compound. The compound in claim 7 is sildenafil.

The basic problem is that the patent did not expressly disclose that sildenafil was the compound that had been tested and found to work. Kelen J put it this way:

135 I prefer to express in obiter my discomfort with the existing jurisprudence which condones a patent description by way of cascading claims for groups of compounds such that the skilled reader must undertake a minor research project to determine which claim is the true invention. In my mind, the disclosure plays games with the reader. Why did the disclosure not simply state that that compound in Claim 7 was sildenafil? The patent plays "hide and seek" with the reader. The reader is expected to look for the "needle in the haystack", or "the tree in the forest". Remember, Claim 1 is for a range of compounds which includes 260 quintillion compounds.

136 By withholding from the public the identity of the only compound tested and found to work, sildenafil, the patent did not fully describe the invention. Obviously Pfizer made a conscious choice not to disclose the identity of the only compound found to work, and left the skilled reader guessing. This is contrary to the statutory requirement to fully disclose the invention.

This effectively captures the heart of the SCC reasoning. All that the SCC did was to turn Kelen J’s obiter into the central holding.

The counter-argument, which Kelen J felt bound to apply, was that sildenafil was specifically identified, in claim 7. At a technical level, Kelen J’s holding that the disclosure was sufficient was driven by the proposition that every claim is treated as a separate invention. If that is so, then “the skilled reader would know from looking at Claim 7 that sildenafil is the invention” [FC 130]. At a most pragmatic level, a skilled reader would know that the most important compounds are individually claimed, so that the compound that was tested and worked must be the compound of either claim 6 or 7, and it would require only routine tests to find out which of these actually worked [FC 146, FCA 78].

The tension, then, is that the disclosure did not expressly disclose that sildenafil was the compound that had been tested and found to work, but the claims, which are part of the specification, narrowed it down to compound 6 or 7, and a skilled person would have choose between the two with only routine experimentation.  

A similar tension has long been evident in the disclosure requirement. Some very early cases suggested that the specification had to disclose everything necessary for the invention to be worked without the need for any experiments at all: see eg Turner v. Winter (1787) 99 Eng. Rep. 1274, 1277 1 WPC 77 (KB); R v Arkwright (1785) 1 WPC 64 at 66; Morgan v Seaward (1836) 1 WPC 170, 175. The practical difficulty with such a rule is that in almost any field there are some steps that require some degree of routine trial and error, and to require a patentee to describe all steps with sufficient particularly to avoid such trials would leave the patent open to technical attacks and in some case might make it impossible to frame a valid claim, even though a skilled person would easily be able to work the invention on the basis of the disclosure. The principled objection is that the specification is read by a person skilled in the art, with a mind willing to understand, and not by an automaton blindly following the instructions of the specification. As Jessel MR said inOtto v Linford (1882) 46 L.T.(NS) 35 at 41, “[t]he first thing to be remembered, in specifications of patents, is that they are addressed to those who know something about that matter.” Consequently, the stringency of the earliest cases was soon rejected, and it has long been clear that the specification is not insufficient simply because trials or experiments are required to work the invention: Wandscheer v Sicard Ltd [1948] SCR 1, 17; Smith Incubator Co v Seiling, [1937] SCR 251 at 254.

The question then is how much experimentation is acceptable. The formulation applied by Canadian Patent Office, in common with UK, EPO and US law, is that the disclosure will be insufficient if “undue experimentation,” (or undue effort or burden) is required to work the invention. One formulation often used by the Federal Court is that “a specification is sufficient if a skilled worker can practice the invention, even if routine trials and experiments not amounting to invention might be necessary to arrive at the desired result”(Cabot Corp et al. v. 318602 Ontario Ltd. et al. (1988), 20 C.P.R. (3d) 132 (FCTD). This implies that the test is whether the effort required amounts to invention. The view that invention sets the standard is supported by statements in many cases of good authority: see egWandscheer ibid, remarking that the disclosure will be good “ “so long as the task involved does not require invention.” Nonetheless, it is not entirely clear that this should be considered a strict test. Certainly it is clear that if invention is required, then the disclosure will be insufficient, but it was never entirely clear that the disclosure will be sufficient even if very extensive trials are required, so long as these do not require invention. While the courts often comment that invention was not required to work the invention when holding a specification sufficient, decisions holding the specification insufficient do not generally hold explicitly that an inventive step would have been required to overcome the difficulty. Further, in many of the cases noting that invention was not required, the experimentation required was minimal on the facts.

But even if experimentation not amounting to invention might be considered "undue," on the particular facts, all that was required was “a minor research project” [FC 135, SCC 75]. Surely, this is not an “undue” burden. What, then, is the difference between Pfizer / Viagra and cases in which the burden was not undue? At one level, the distinction may be that even though the trials required were minor, so to was the effort required to make the disclosure: “Pfizer had the information needed to disclose the useful compound and chose not to release it. . . .Even now, in its factum to this Court, Pfizer offers no explanation as to why — knowing that Claim 7 contained the tested and thus, the useful, compound — it elected to withhold that information.” [76]. Recall that the rationale for allowing minor experimentation is to avoid the need to write a lengthy technical manual explaining how to carry out routine steps that would already be known to the skilled reader. In this case, Pfizer did not have to write an operations manual. All it had to do was add one sentence to the disclosure stating which particular compound had been tested.

At another level, I suggest that was distinguishes Pfizer / Viagra is that the failure to disclose went to the heart of the invention. Pfizer did not omit some routine step in the preparation of the compound. What was omitted was central to the inventive concept. As the SCC noted, quoting Kelen J, “By withholding from the public the identity of the only compound tested and found to work, sildenafil, the patent did not fully describe the invention” [77, emphasis added by SCC]. It is “the invention” itself which was not disclosed, not steps to its preparation.

The SCC’s insistence that the invention itself be disclosed could be seen as a kind of best mode requirement. This was argued in the courts below, but dismissed on the basis that the statutory best mode requirement applies only to machines [FCA 72]. While this is true, the best mode requirement itself is really only a specific aspect of the requirement to make full disclosure. The best mode requirement was a codification of the case law elaboration of the disclosure requirement: Morgan v Seaward (1836) 1 WPC 170, 174-75. That is, even in the absence of a statutory best mode requirement, the requirement to make full disclosure may, in some cases, require the patentee to disclose its best mode. This does not amount to a general best mode requirement. For example, if all nine especially preferred compounds worked well, it is unlikely that the patent would have been invalidated for failure to disclose that one worked marginally better. But on the facts of this case, where only one had been tested and known to work, it was necessary to disclose that fact.

To summarize, there is a tension between the principle that the patentee must make full disclosure and the principle that some trials are permitted. As compared with the FC and FCA, the SCC decision inPfizer / Viagra put more emphasis on the former principle. However, this is a matter of emphasis, rather than a matter of a basic change in the law. If the Federal Court judges choose to interpret Pfizer / Viagra aggressively, then this might lead to an secular shift in the balance. But I do not think this is likely. This case was close to the line of a valid disclosure, not because the law was shifting, but because the patentee was testing the boundaries. This was a dangerous game, which did not pay off. There is no reason to think that the courts will be aggressive when faced with a reasonable, good faith disclosure.

At the level of legal doctrine, the debate centered on the rule that each claim is to be treated as a separate invention. This post is long enough already, so I will discuss that point in tomorrow’s post.

*McLachlin C.J. and LeBel, Deschamps, Abella, Rothstein, Cromwell and Moldaver JJ.


  1. What is your take on the fact that the SCC voided the patent? This seems to at least partly overturn multiple FCA decisions related to assessing validity in PMNOC cases.

  2. Could you give me an example of a particular FCA decision which is called into question? This is not to dispute your assertion, but I can give more thought to it in the context of a particular example.

  3. I believe that any of the following are good examples:

    Merck Frosst Canada Inc. v. Canada (Minister of National Health and Welfare), [1994] F.C.J. No. 662 at paras. 25-26, 55 C.P.R. (3d) 302 (C.A.) (QL), Hugessen J.A., leave to appeal to S.C.C. refused, [1994] S.C.C.A. No. 330; Wyeth Canada v. Ratiopharm Inc. (2007), 60 C.P.R. (4th) 375 at para. 21, 2007 FCA 264, leave to appeal to S.C.C. refused, [2007] S.C.C.A. No. 572; SmithKline Beecham Pharma Inc. v. Apotex Inc., [2003] 1 F.C. 118 at para. 23, 2002 FCA 216, leave to appeal to S.C.C. refused, [2002] S.C.C.A. No. 324; Pfizer Canada Inc. v. Apotex Inc. (2001), 11 C.P.R. (4th) 245 at para. 17 (F.C.A.), leave to appeal to S.C.C refused, [2001] S.C.C.A. No. 111; Eli Lilly Canada Inc. v. Novopharm Ltd. (2007), 62 C.P.R. (4th) 161 at paras. 40-41, 2007 FCA 359, leave to appeal to S.C.C. requested, [2008] S.C.C.A. No. 9.

  4. I understand. I had completely missed that point. The only statement that the patent is void is in the final sentence “I would therefore allow the appeal with costs and hold that Patent 2,163,446 is void” [91]. I think this must be interpreted as meaning “invalid for the purposes of this proceeding.” The rule that validity in NOC proceedings is not in rem was not in issue – it was not even mentioned at any level – and I don’t think the SCC can be taken to have overturned such a well-established rule without any discussion at all. Note that the courts below routinely referred to the question being whether the patent is “invalid.” They do not explicitly say that the invalidity is effective only for the purposes of this proceeding, but it is clear that this is what they meant, since the FC and FCA are well aware of their on jurisprudence on this point. In its substantive reasoning, the SCC at [83] concludes that the patent is “invalid” and here I think the SCC was using the term in the same way as the courts below. I think that in the concluding sentence the SCC was using “void” as a synonym for “invalid” without appreciating that there might be any substantive difference. The appellants in their factum did not request a declaration of invalidity, but only that "this appeal be allowed" with costs.

  5. That seems like a good interpretation. Thanks. It will be interesting to see how this is treated if there is a corresponding infringement action.