“[T]he soundness of a prediction is a question of fact”

Sanofi-Aventis Canada Inc. v. Apotex Inc. / ramipril, 2011 FCA 300 Layden-Stevenson JA: Noël, Pelletier JJA aff’g 2009 FC 676 Snider J

In this brief decision, the FCA affirmed Snider J’s finding that Sanofi’s claims to ramipril were invalid on the basis that the inventors could not soundly predict the utility promised by the patent as of the filing date. The FCA stated at [5] that “the soundness of a prediction is a question of fact” which was therefore owed deference unless the trial judge had applied the wrong legal test. The FCA held that Snider J has applied the correct test, and her findings of fact were amply supported by her detailed review of the evidence.

While this conclusion is no doubt reasonable on the facts (which I have not reviewed in detail), I am concerned by the apparently blanket statement that “the soundness of a prediction is a question of fact.” If taken to an extreme, so that every case turns on its own facts, it will not be possible to develop guidelines that provide the legal predictability and certainty that is so important to an industry such as biotechnology, which requires massive investments in reliance on patent protection. This statement is also doubtful as a matter of principle. There is no sharp dividing line between questions of law and questions of fact. Legal principles may be stated at different levels of generality, and as principles become more detailed, they shade more closely into questions of fact.

Both of these points are illustrated by the recent decision of the UKSC in HGS v Eli Lilly [2011] UKSC 51 rev’g [2010] EWCA Civ 33 (blogged here, here and here), in which the EWCA distinguished a number of decisions from the EPO Technical Boards of Appeal dealing with very similar facts, essentially on the basis that each case turned on its own facts. That would have made is makes it difficult for patentees to determine whether their patent is valid, and more importantly, to determine how soon they can apply for a patent with confidence that its validity will be upheld. As explained by Lord Neuberger in HGS v Eli Lilly at [98], this predictability is very important in the pharmaceutical industry, which relies on patent protection to fund research. The patentee is caught in a squeeze between wanting to patent early, to protect its innovation in a competitive industry, and not patenting too early in order to ensure validity. Whether the line is drawn early or late is arguably less important than that it is drawn clearly, so that inventors can know, rather than guess, what point they have to reach in order to obtain a valid patent. The UKSC, which strongly emphasized the need for legal certainty, felt able to draw principles from the TBA decisions which the EWCA had viewed as turning on their own facts. Notably, the UKSC stated the governing principles at three different levels of generality: general principles; principles applicable to new proteins and their encoding genes; and principles applicable where the protein is said to be a family or superfamily member [107]. It was the willingness to set out principles applicable to a narrow class of cases, rather than simply saying that the cases turn on their own facts, that mark the difference between the EWCA and the UKSC, and which allowed the UKSC to state guidelines that enhance predictability.

I am not arguing that the FCA was wrong in holding that the question on appeal in the ramipril case was one of fact. It is not possible to state principles that will provide clear answers in all cases; to some degree each case must turn on its own facts. My concern is that a blanket statement that the soundness of a prediction is a matter of fact reflects an approach similar to that taken by the EWCA in HGS v Eli Lilly, and may consequently impede the development of detailed principles that would improve predictability.