Discovery of Something Unexpected about Something Obvious

Bristol-Myers Squibb v Teva (NOC) 2016 FC 580 Mactavish J
            2,250,840 / 2,317,736 / atazanavir / REYATAZ

In this NOC proceeding Mactavish J held Novartis’ 840 patent, which claims the compound atazanavir [53], to be valid in the face of obviousness and anticipation attacks, while she held BMS’s 736 patent, which claims the bisulfate salt of atazanavir [273], to be obvious. A central dispute between the parties in respect of both patents was whether the inventive concept of the claimed inventions comprised just the compound itself, as Teva argued, or the compound and all its properties, as the patentees argued. The idea is that the more properties that are part of the inventive concept, the less likely it is that they will all be predictable, and so the less likely it is that the patent will be held obvious. In this post, I will argue that this idea is wrong; the number of properties in the inventive concept does not affect how likely it is that the claimed invention will be held to be obvious. In another post I will suggest that multiplying the inventive concepts might actually make invalidity more likely. The 736 patent also raised a very tricky claim construction problem, which I will discuss in a subsequent post.

Atazanavir is used in the treatment of HIV/AIDS. The first class of anti-HIV drugs were known as nucleotide reverse transcriptase inhibitors (or NRTIs) [13]. While these had potent activity against HIV, they were susceptible to the evolution of drug resistance and had unpleasant side-effects [14], [15]. A new class of anti-HIV drugs was developed, known as protease inhibitors [16]. These has a number of advantages over NRTIs, but the first generation protease inhibitors had poor bioavailability, which resulted in poor compliance and a consequent potential for the development of drug resistance [22]. There was therefore a strong motivation to develop second-generation protease inhibitors [23]. One of these was atazanavir, which came on the market in 2003 and which remains one of the best second generation protease inhibitors [24-25]. The 840 patent, which claimed atazanavir and its salts, disclosed only the free base form. While the free base had good inherent bioavailability, it was not sufficiently orally bioavailable in solid form to be a viable product [266-67]. The problem facing the inventors of the 736 patent was to develop a form of atazanavir having properties suitable for an oral pharmaceutical dosage form, including improved oral bioavailability [268]. They performed a salt screen, and solved the problem with the bisulfate salt which was claimed in the 736 patent. (Ciba-Geigy, which became Novartis, developed atazanavir, and BMS acquired the rights to atazanavir from Novartis and then developed atazanavir bisulfate.)

Obviousness, in particular the obvious-to-try test, was the central issue with respect to both patents. At the end of the day, Mactavish J held that the 736 patent was obvious, but 840 patent was not. In my view, both of these conclusions were amply warranted on the evidence reviewed by Mactavish J. She also affirmed that the correct standard is whether it is “more or less self-evident” to try to obtain the invention, not whether there was a fair expectation of success [458]. The obvious-to-try test wasn’t articulated as precisely as it might have been by the SCC in Sanofi 2008 SCC 61, but this is another case in which the FC has applied it sensibly to arrive at an appropriate conclusion on the facts. On the whole, the obvious-to-try test seems to be working quite well.

The more interesting issue, which was also the focus of the argument, was on the proper characterization of the inventive concept of the patent [104]-[146], [413]-[436]. This reflects a trend that has been building ever since Sanofi 2008 SCC 61, [67] held that the second step of the test for obviousness was to “identify the inventive concept of the claim in question or if that cannot readily be done, construe it.” We now commonly see extended discussions of the identification of the inventive concept which rarely appeared in earlier cases. This decision raised several issues on this point. Is identification of the inventive concept mandatory or optional? (Answer: mandatory [119].) When is it appropriate to look at the disclosure rather than the claims to identify the inventive concept? (Answer: when the inventive concept is not readily discernable from the claims themselves [126]). In this post I address a more basic question: why does the identification of the inventive concept matter in the first place? (At least I will make a start on addressing it – I think the role of the inventive concept is one which deserves an article rather than a blog post.)

For both patents Teva argued for a narrow inventive concept, limited to the compound itself (or atazanavir and its most basic property, protease inhibition) [104], [110], [122], while the patentee argued that the inventive concept included the compound’s advantageous properties [104], [121], [416], [417]. The idea, evidently, is that the more attributes comprising the inventive concept, the less likely it is that they will all be obvious.

So, for the 736 patent, BMS argued that it had to be obvious that by performing a salt screen, a salt would be created that had all the properties that are part of the inventive concept [387], [451]. Teva, on the other hand, argued that the law does not require that a POSITA be able to predict all the advantageous properties of the claimed compound [380], [450]. BMS argued that the inventive concept also included its properties, in particular (1) crystallinity, (2) oral bioavailability, (3) stability, and (4) what was referred to as in situ transformation of the salts [416]. Mactavish J ultimately concluded that the inventive concept included the first three of these properties but not the fourth [446].

Mactavish J’s holding that three of the four properties were part of the inventive concept might seem to bode well for the patentee. But it didn’t turn out that way. Mactavish J held that it was obvious to try a salt screen in order to improve the oral bioavailability of atazanavir [496], and in conducting a salt screen the POSITA would have come directly and without difficulty to the bisulfate salts of atazanavir [501]. Consequently, “to the extent that the inventive concept of the ’736 patent was the improved bioavailability in animals of . . . atazanavir bisulfate salts over the free base of atazanavir, it was obvious” [505]. With respect to the other two properties that were part of the inventive concept:

[501] Standard techniques for characterizing the properties of these salts would then have disclosed the existence of . . . atazanavir bisulfate salts and the[ir] properties . . including the anhydrous non-hygroscopic crystallinity and solid state stability. . . While the POSITA would not have known in advance what the properties of each of the bisulfate salts would be, I do not understand the applicants to suggest that there was anything inventive about the techniques used to identify the characteristics of the Type-I atazanavir bisulfate salt.

And similarly:

[507] That said, the fact that . . . atazanavir bisulfate salts were stable, non-hygroscopic, anhydrous solid form crystals was easily discovered using routine techniques, without undue effort. The discovery of these inherent characteristics of the Type-I atazanavir bisulfate salt did not add anything inventive.

From this, I take two points. First, the fact that two properties were unpredictable was beside the point if one property (bioavailability) was obvious, and the others were easily verified. The second, which is implied rather than stated directly, is that if verification of the other properties had involved inventive ingenuity, the invention would not have been obvious, notwithstanding that improved bioavailability was obvious – or, more precisely, that improved bioavailability would have rendered the invention obvious. The first point is part of the ratio, while the second point, to the extent that it was intended by Mactavish J at all, is strictly obiter, given that on the facts of this case verification of the other properties was not inventive.

I will argue that the second proposition is not correct; the invention would have been obvious even if inventive ingenuity had been required to verify the other properties.

My starting point is that “[m]erely knowing something new about something old is not enough” to support a patent: Genentech Inc.'s Patent [1989] RPC 147 (CA) at 208, ¶ 12.10. In Canadian law this is reflected in the proposition that discovery of a mechanism of action of a known product will not support a patent: Alcon / olopatadine 2012 FC 410 Barnes J (blogged here); Lundbeck / memantine 2009 FC 1102 at [20], [149], [189-92] Mactavish J; Astrazeneca AB v Apotex Inc / omeprazole (NOC), 2007 FC 688 [103] Barnes J; Abbott Laboratories v Sandoz Canada Inc 2008 FC 1359 [71] Hughes J. Similarly, discovery of new properties of a known compound cannot support a patent for the compound itself (though it may support a patent for a new use of the known compound): see In re Spada, 911 F2d 705, 708 (Fed Cir 1990). The reason is that even though the discovery may mark a significant and inventive contribution to the technical knowledge, the claimed invention is not novel. So, in Memantine 2009 FC 1102 the inventors had discovered memantine’s previously unknown mechanism of action, but memantine itself was anticipated: [190-92]. (Though if that discovery leads to a novel application, such as a treatment of a different disease, or a new method of administration, that application may be patentable, even though the application is obvious once the mechanism was known, because in that case the claimed invention would be new: Genentech Inc.'s Patent [1989] RPC 147 (CA) at 208, ¶ 12.11; Olopatadine 2012 FC 410 [38].)

With that in mind, consider these scenarios: 

(1) Atazanavir bisulfate is a known compound, which is known to have good bioavailability. The patent has expired. A researcher discovers, non-obviously, that it has exceptionally good stability.

(2) Atazanavir is a known compound, with poor bioavailability. A researcher discovers, non-obviously, that atazanavir bisulfate, a new compound, has good bioavailability. The researcher gets a valid patent for atazanavir bisulfate. The patent expires. The researcher later discovers, non-obviously, that atazanavir bisulfate has exceptionally good stability.

(3) Atazanavir is a known compound, with poor bioavailability. A researcher discovers that atazanavir bisulfate, a new compound, has good bioavailability. The researcher gets a patent for atazanavir bisulfate. The patent is challenged in litigation and is held to be invalid on the basis that it was obvious to try atazanavir bisulfate in order to improve bioavailability. The researcher later discovers, non-obviously, that atazanavir bisulfate has exceptionally good stability.

In all these scenarios it is clear that the researcher cannot get a second patent for atazanavir bisulfate based on the discovery of its exceptional stability, because the claimed invention, atazanavir bisulfate, is not new.

Now consider a fourth scenario:

(4) Atazanavir is a known compound, with poor bioavailability. A researcher discovers that atazanavir bisulfate, a new compound, has good bioavailability and that it has exceptionally good stability. The researcher gets a patent for atazanavir bisulfate. The patent is challenged in litigation and it is held that it was obvious to try atazanavir bisulfate in order to improve bioavailability. However, it was not obvious that it would have exceptionally good stability.

Is the patent in scenario (4) valid? In my view the answer is no. There is no difference in principle between scenarios (3) and (4). Whether the patent is held invalid before or after the discovery of the unexpected stability does not change the fact that the claimed subject-matter was obvious. Discovery of something new about something old does not make that old thing new; and discovery of something new about something obvious does not make that obvious thing inventive.

That point is well settled in the UK, where it is known as the “golden bonus” rule. This rule, which is usually traced back to Hallen v Brabantia [1991] RPC 195, 216 (CA), says that if an invention is obvious for one reason, it does not become non-obvious merely because it has some non-obvious benefits (the "golden bonus"). (The same is apparently true in Germany.) While the rule is not as well established in Canada, it was applied in substance by Barnes J in Janssen / bortezomib 2015 FC 184 (blogged here).

The statutory reason that “knowing something new about something old” is not enough, is that according to s 28.2 it is “the subject-matter defined by a claim” that must be new; in scenarios (1), (2) & (3) the claimed invention, atazanavir bisulfate, was not new at the time of the second patent. Similarly, according to s 28.3 it is the subject-matter defined by a claim that must be non-obvious, not the inventive concept. In scenario (4) the subject-matter defined by the claim was obvious, even though its stability was not.

Despite the extensive discussion of the inventive concept, Mactavish J’s factual findings directly addressed the claimed subject-matter, namely atazanavir bisulfate. Applying the criteria set out in Sanofi, she found:

[501] In conducting a salt screen of atazanavir, the POSITA would, therefore, have come directly and without difficulty to the bisulfate salts of atazanavir.

[502] That the POSITA would have come directly and without difficulty to the . . . bisulfate salt of atazanavir is confirmed by what actually happened in this case.

[503] Indeed, [the lead inventor’s] team was able to get to the . . . atazanavir bisulfate salt “quickly, easily, directly and relatively inexpensively, in light of the prior art and common general knowledge”

From this it follows directly that the subject-matter defined by the claim would have been obvious. But Mactavish J did not quite come to this conclusion. She held (my emphasis):

[505] As a consequence, I find that to the extent that the inventive concept of the ’736 patent was the improved bioavailability in animals of Type-I atazanavir bisulfate salts over the free base of atazanavir, it was obvious.

She then went on to find that identification of the other properties “was a serendipitous discovery made without prolonged or arduous work, and not an invention” [508]. The phrase I have emphasized in [505] suggests that if she had come to the opposite conclusion, that the discovery of those other properties had involved invention, she would have held atazanavir bisulfate to be non-obvious. But having previously concluded that a POSITA would have come directly and without difficulty to the claimed invention, what does it matter whether identification of some of its properties would have required ingenuity? It is clear law that discovery of those properties after the 736 patent had issued would not have supported a patent; what difference does it make that the non-obvious properties of an obvious subject-matter were discovered before the patent was issued rather than after? That can no more support a valid patent than discovery of non-obvious properties of old subject-matter.

In summary, I would argue that increasing the number of properties comprising the inventive concept does not reduce the chance that it will be held invalid, because it is enough if the claimed subject matter is obvious for any one reason. Adding non-obvious properties to obvious subject-matter does not make the subject-matter non-obvious. This is because the question isn’t whether the property of the invention is obvious, it is whether the property would make the claimed subject-matter obvious. If one property would make the invention obvious, then it is obvious whether or not other properties might be unexpected. Conversely, though, limiting the number of inventive concepts does not necessarily make it more likely that the claimed invention will be held invalid; but that is an argument for another day.