Tuesday, October 3, 2017

Is it "More or less Self-evident That What Is Being Tested Ought to Work"?

Pfizer Canada Inc v Apotex Inc 2017 FC 774 Brown J
2,436,668 / desvenlafaxine (ODV) / PRISTIQ / NOC

Bristol-Myers Squibb Canada Co v Apotex Inc 2017 FCA 190 [Dasatinib FCA] Gleason JA: Webb, Near JJA var’g 2017 FC 296 [Dasatinib FC] Manson J
            2,519,898 / dasatinib / SPRYCEL / NOC

Desvenlafaxine and Dasatinib both apply the “obvious to try” test – endorsed in Sanofi 2008 SCC 61, as “one factor to assist in the obviousness inquiry” [64] – but to opposite effect, with Brown J holding the claimed invention to be non-obvious, while Manson J, now affirmed by the FCA, holding the invention at issue to be obvious. In my view, both decisions applied the test correctly, given the facts as found. And indeed, the FC has almost always applied the test correctly in the years since Sanofi was decided. Perhaps it should not be surprising that the courts apply the law correctly. But I do find it surprising, because there is continuing doctrinal confusion as to the nature of the test, stemming ultimately from Sanofi itself. That confusion has surfaced before, and has now surfaced again in both these recent decisions.

In Sanofi, the SCC stated that:

[65] I am of the opinion that the “obvious to try” test will work only where it is very plain or, to use the words of Jacob L.J., more or less self-evident that what is being tested ought to work.

This lends itself to the interpretation that it must be self-evident, prior to any experimentation, that the claimed invention will solve the problem at hand. But that is difficult to reconcile with the Court’s observation that the obvious to try test is appropriate precisely in “areas of endeavour where advances are often won by experimentation” (Sanofi [68]), which implies that the test applies particularly when it is not self-evident, prior to any experimentation, that the claimed invention will solve the problem at hand.

Consider the seminal case of Johns-Manville Corporation's Patent [1967] RPC 479 (CA), which concerned the use of a particular compound, polyacrylamide, as a flocculating agent for filtration of cement. Flocculating agents were well known as being useful in filtration processes generally and the possibility of using such an agent to improve filtration of cement had long been appreciated, but all those previously tried had decreased the quality of the final cement product. Polyacrylamide had been recently developed by a third party as a general flocculating agent and its use in cement filtration had been suggested. Because it was particularly effective and could be used at low concentrations, there was reason to think that it would prove useful where other agents had not. On the facts, a skilled person would have no difficulty in actually making it work.* On these facts, the Court of Appeal affirmed the trial judge’s holding that the claimed invention was obvious. But if we ask the whether it was “more or less self-evident that what is being tested [polyacrylamide] ought to work [as a flocculating agent for filtration of cement],” the answer is also no, because the nature of cement making is such that nothing is certain until it is tried.

It has always seemed to me that the obvious-to-try analysis is easier to understand if we approach by asking whether there was an inventive step. The traditional formulation of the question in Canada, as in the UK, was whether the claimed invention lacked “inventive ingenuity” or an “inventive step,” though alternative formulations, such as whether the invention would “occur to any ordinary mind,” were also used. The Cripps question was originally merely a way of setting the standard – how much inventive ingenuity. When framed in terms of inventive ingenuity, Johns-Manville is straightforward: was there any inventive ingenuity involved in selecting polyacrylamide to try as a flocculating agent? On the facts, the answer was no. The need for a better flocculating agent was known, polyacrylamide was a new agent with good properties, so of course it should be tried. Was there any inventive ingenuity involved in getting polyacrylamide to work? Again, on the facts, the answer was no. Putting these together, there was no inventive step involved in going from the problem to the solution, and consequently the claimed invention was invalid. (If we must frame this in terms of obviousness, there are two questions: was the claimed invention obvious to try? If no, it is not obvious. Was it obvious to succeed? If no, it is not obvious. It is only if the answer to both these question is “yes” that the invention is obvious.)

In summary, in areas in which advances are won through experimentation, if we ask whether it would have been very plain, prior to any experimentation, that what was being tested would solve the problem, the answer will always be “no – we won’t know until we try.” But if that were the right question, patents for putative inventions which involve no inventive step would nonetheless be valid. The obvious to try test was developed precisely to ask a different question in these kinds of cases.

I won’t try to parse Sanofi in this blog post, except to note that it is doubtful that the Court held that the test is whether it is “more or less self-evident that what is being tested ought to work” [65]. In [65], that statement appears as though it were a test, but in the very next paragraph, the Court states what also appears to be a test, namely (my emphasis)

[66] For a finding that an invention was “obvious to try”, there must be evidence to convince a judge on a balance of probabilities that it was more or less self-evident to try to obtain the invention.

Asking whether it is self-evident to try is very different from asking whether it is self-evident that it will succeed. Note also that this statement is not exhaustive: there must be evidence that it was obvious to try, but not everything that is obvious to try is necessarily obvious, and it might have been obvious to try, but requiring invention to actually implement. Moreover, in [69] the “ought to work” statement appears, not as a test, but as one of several factors.

So, the approach to obvious-to-try set out in Sanofi is somewhat confusing. Or maybe it is just me that is confused, as the Federal Court has, in my view, consistently applied the obvious-to-try test appropriately on the facts.

Consider Desvenlafaxine. Claims 8 and 9 of the ‘668 patent claimed Form I ODV succinate, which is a particular crystal form of a particular salt of ODV [7], [168], [175]. (ODV is an abbreviation for o-desmethyl-venlafaxine or desvenlafaxine). (Other claims at issue claimed the same compound for the treatment of depression and a sustained release (oral) form.) Venlafaxine was a known treatment for depression, and it was also known that ODV was the active metabolite of venlafaxine responsible for almost all the pharmacological effect. Consequently ODV itself and its use to treat depression were previously known [8], [9]. Development of a drug form of ODV itself was desirable because, inter alia, it was thought that it might be more potent, and less variable in its effects, than venlafaxine [32]-[33]. It appears that neither the succinate salt nor any crystalline form of ODV succinate had ever been made before [13], [14], and it was agreed that Form I ODV succinate was a new compound that had never before been made or disclosed [27]. While crystallinity is generally highly desirable for a drug candidate, not every crystalline salt will have suitable properties for drug development (e.g. solubility, physical and chemical stability, and bioavailability) [67]. Wyeth and scientists at SSCI, a specialist firm contracted by Wyeth, eventually developed Form I ODV succinate.

On the question of obviousness, Brown J started by considering the definition of obviousness as it had been defined in Beloit (1986) 8 CPR(3d) 289 (FCA). The question, under Beloit, is whether the POSITA:

would, in the light of the state of the art and of common general knowledge as at the claimed date of invention, have come directly and without difficulty to the solution taught by the patent. [220]

In the context of this case, that question becomes whether the Skilled Person would have come directly and without difficulty to Form I ODV succinate [277]. Evidence accepted by Brown J established that, prior to actual synthesis and testing, it is not possible to predict “the properties of either ODV succinate generally, or Form I ODV succinate, or in particular, whether those properties would be amenable to formulation as a sustained release dosage form” [235]; (and see evidence emphasized in [234]). Consequently, the answer to that question was no: [278], referring to evidence reviewed in [234]; [279].

The difficulty is that this analysis seems to suggest that the fact that the properties were unpredictable in the absence of experiment was in itself sufficient to establish non-obviousness. But because those properties are generally unpredictable, that implies that any salt or crystalline form is non-obvious. For example, the fumarate salt of ODV had been prepared and was crystalline [300.l]. Since it was known that crystallinity is generally highly desirable for a drug candidate, and crystalline ODV fumarate was already prepared and sitting on the shelf, it would have been – dare I say – obvious to try the fumarate salt. And indeed, it was one of the first things that Wyeth tried [43]. It did not work, but if it had worked without difficulty after being tried, there would be a good argument that it was obvious, even though the properties of the fumarate salt are no more predictable in advance than those of the succinate.

This is not to suggest Brown J’s conclusion was wrong. There was also considerable evidence, which he also emphasized, that there was an inventive step both in choice of the succinate salt and in crystallizing it: see e.g. [305]. That is, crystalline succinate was neither obvious to try, nor obvious to succeed. This was emphasized when he went on to consider the Sanofi test for obviousness ([282ff]), before arriving at his ultimate conclusion that the invention was not obvious [330]. Thus, on the facts, there was ample evidence to support his conclusion that the claimed invention was not obvious; my point is that the fact that the properties of a salt or crystal form cannot be predicted in advance, should not be determinative. Further, the fact that Brown J first undertook an extensive analysis of obviousness under the Beloit standard, before turning to the Sanofi analysis, illustrates a degree of understandable discomfort with the Sanofi analysis.

Consider in contrast the facts in Dasatinib concerning the ‘898 patent. The claims at issue were to the oral use of dasatinib (and related compounds) for treating chronic myelogenous leukemia (CML) [26]-[27]. At first instance Manson J held the ‘898 patent invalid for obviousness as being obvious to try. The prior art included the ‘788 Application, which matured to the ‘932 patent. As discussed in last week’s post, the ‘932 patent claimed the compound dasatinib as such, and also disclosed that the claimed compounds were useful in treating cancers of the CML type: Dasatinib FC [149], [176]. Oral administration is evidently desirable, and had at least been suggested in the ‘788 application, but, on the facts, “the effectiveness of oral administration could not be predicted prior to performing clinical tests” [181]. Therefore, if the question is whether it was “more or less self-evident that what is being tested [oral administration] ought to work,” the answer was no. But Manson J held that “I do not consider this to be dispositive of whether an invention was obvious to try” [181], and considering, inter alia, that the patentee “provided no evidence suggesting that the work they did to reach the invention in claim 1 (i.e., oral administration of dasatinib to treat CML) was long or arduous” [185], Manson J held the claimed invention to be obvious [186]. In other words, there was no inventive step in deciding to try oral administration, and there was no inventive step involved in getting oral administration to work. There being no inventive step in going from the prior art to the solution, the invention was therefore invalid for obviousness. In my view, Manson J applied the obvious to try test correctly to conclude that the invention was obvious, notwithstanding that it was not self-evident that oral administration would work, prior to the actual experiment.

On appeal, BMS argued that it was an error for Manson J to have applied the obvious to try test given that “Both Parties agree that it was not obvious at the relevant date that dasatinib would be an effective oral treatment for CML and/or imatinib-resistant CML” [173]. The FCA affirmed on the basis that on the facts it was open to Manson J to apply the obvious to try test, though “it is difficult to understand what the Federal Court meant in the first sentence of paragraph [173]" [57]. While the parties would know better than me what they agreed to, what I understood Manson J to be saying is that it was not obvious in the sense that the effectiveness of oral administration could not be predicted prior to performing clinical tests. As Manson J held, that does not conclude the inquiry, but rather implies that the obvious to try test should be applied.

The second ground of appeal was that “the Federal Court misdirected itself by considering whether the experiments to establish that dasatinib was effective to treat CML and imatinib-resistant CML were obvious to try as opposed to considering whether it was more or less self-evident that such experiments would establish the efficacy of dasatinib” [58]. In other words, BMS argued that the correct question was whether it was “more or less self-evident that what is being tested [oral administration of dasatinib] ought to work [to treat CML].” The FCA rejected this ground of appeal on the basis that, on the whole, Manson J had asked “the right question, namely whether it was more or less self-evident that routine experiments would establish that dasatinib was effective to treat CML, including imatinib-resistant CML” [59]. While the FCA upheld Manson J, this statement is problematic: the FCA seems to be accepting that the correct question is whether it was self-evident that the experiments would succeed. As discussed above, if that is the correct question then no inventions would be obvious, no matter how routine, in fields in which advances are won by experimentation, and little can be predicted in advance.

In summary, it seems to me that the law related to the obvious-to-try analysis remains conceptually uncertain. This can be traced back to ambiguity and confusion in Sanofi itself. Despite this doctrinal confusion, the Federal Court has consistently applied the obvious to try analysis to arrive at appropriate conclusions on the facts, both in upholding claims and striking them down as obvious to try, as is illustrated by Desvenlafaxine and Dasatinib respectively. We might recall that the requirement of an inventive step originated as a judge-made doctrine based on the recognition that some advances were so small as to not deserve the reward of a patent: see eg Losh v Hague (1838), 1 WPC 202, 208. Fortunately, despite codifcation and the accretion of doctrine, the courts have not lost sight of that basic purpose.

*The patentee had initially tried it, and failed, but it was held that it would have worked had a person skilled in the art tried it, and it was the patentee’s incompetence that had led to the initial failure. The patentee was obliged to accept this position, as the patent did not disclose the particular procedure the patentee had successfully implemented and if it was not obvious then the patent would have been void for insufficiency.


  1. It was widely felt in the UK that the "obvious to try" test as propounded in Johns-Manville was unfair on Patentees. That is why Jacob L.J. modified the test in the form it was adopted in Sanofi.

  2. Yes, the particular problem with Johns-Manville was that the bar for whether something was “obvious to try” was that it was “well worth trying out". I agree that is too low a standard (ie to easy to invalidate the patent on that basis) for the reasons given by Jacob LJ in Angiotech. However, the point of the example is that the mere fact that the properties are unpredictable prior to experimentation does not in itself make the invention non-obvious, so long as the invention was obvious to try. That point remains even if the standard for obvious to try is whether it is “more or less self-evident” that it ought to be tried.

  3. Once it was determined that the succinate salt of ODV was novel and inventive ( I believe this was correct), why bother with applying Sanofi to Form I of ODV? It would appear any polymorph that originates from this salt would be novel and non-obvious because of the base salt form.

  4. Brown J applied Sanofi to Form I ODV succinate because that was the claimed invention; the claim to the succinate salt was not an issue. at issue. I don't really see a holding that the succinate salt itself was inventive. Can you point me to a specific paragraph?

  5. You are correct in that Form I ODV succinate was the claimed invention according to the judge. I just find it odd that it was not the ODV succinate salt itself.

  6. The succinate salt was claimed in claims 2-5, but they were not in issue