Monday, April 27, 2015

Blogging Break

I am taking an end of term vacation. I expect to resume blogging sometime on the week of May 11, starting with any decisions that came out while I was away.

Wednesday, April 22, 2015

SCC Affirms FCA on Calculation of Damages under NOC S 8

Sanofi-Aventis Canada Inc v Apotex Inc / ramipril (s 8) SCC 35886 aff’g 2014 FCA 68 Sharlow JA, Pelletier JA concurring, Mainville JA dissenting, var’g 2012 FC 553 Snider J

On Monday the SCC delivered judgment orally, dismissing the appeal in Ramipril s 8:

THE CHIEF JUSTICE — We are all of the view to dismiss the appeal substantially for the reasons of the majority of the Court of Appeal.
           The appeal is dismissed with costs.

My posts on the FCA decision are here, here and here; posts on the FC decision are here and here, as well as here in the related Teva action.

The companion Teva case, which was not appealed to the SCC, also dealt with assessment of damages under s 8: 2014 FCA 67 var’g 2012 FC 552, with an additional blog post here relating to an issue raised solely in Teva. See also the related litigation challenging the validity of s 8: 2014 FCA 69 blogged here, aff’g 2012 FC 551 here.

Friday, April 17, 2015

Should Experts Be “Blinded” as to the Infringing Device When Construing the Claim?

Teva Canada Innovation and Teva Pharmaceutical Industries Ltd v Apotex Inc (NOC) 2014 FC 1070 Gleason J
            2,232,310 / rasagiline / AZILECT

In this NOC decision, which turned entirely on claim construction, Apotex’ Apo-rasagiline product was found not to infringe Teva’s ‘310 patent. The main point of legal interest is Gleason J’s holding that if an expert is “blinded” as to the nature of the putatively infringing product when carrying out their claim construction, this is a reason to prefer that evidence over the evidence of an expert who is not so blinded.

The ‘310 claims a pharmaceutical composition comprising rasagiline and at least one alcohol selected from the group consisting of “pentahydric and hexahydric alcohols” [7]. It was undisputed that the term “pentahydric” refers to five hydroxyl, or “-OH” groups, and that the term “hexahydric” refers to six hydroxyl groups [20]. So, for example, mannitol is pentahydric:

Apotex’s product comprised rasagiline and an alcohol, and the sole question was whether that alcohol was pentahydric or hexahydric. While the exact alcohol used by Apotex was redacted (with the redaction to be removed once Apotex’ NOC is issued) [5], the problem can be understood if we suppose Apotex’ product contained maltitol:
Apotex’ experts opined that in naming an alcohol, all the -OH groups in the molecule should be counted, while Teva’s experts were of the view that only the groups on the alcohol moiety should be counted. So, on the former view maltitol would be a nonahydric alcohol, and on the latter view it would be pentahydric.

Gleason J accepted the view of Apotex’ experts, primarily - or at least "firstly" - as a matter of textual construction of the specification. In particular, the specification uses the term “alcohol” to refer to the entire molecule to be used in the composition, not just the alcohol moiety, and it uses the term “pentahydric or hexahydric” to modify the term “alcohol,” thus implying that all the -OH groups in the molecule must be counted, not just those on the alcohol moiety [89]-[93].

A second reason that Gleason J preferred the evidence of Apotex’ experts is of more general interest. Apotex’ experts had been provided with the ‘310 patent and asked to construe it, and only subsequently were they provided with Apotex’ ANDS and asked to opine as to whether Apo-rasagiline fell within the claims [43]-[46]. Teva’s experts, on the other hand, construed the ‘310 patent with the allegedly infringing substance in mind [34]-[42]. Gleason J held that the way in which they had been instructed was a reason to prefer the evidence of Apotex’ experts [94]. (Though she did not suggest it was a determinative reason.)

As Gleason J acknowledged [96], this raises the tension between the principle that “the construction exercise [should] be uninfluenced by concerns over infringement or invalidity,” and the competing principle that in construing claims "it is essential to see where the shoe pinches” Nokia v Interdigital [2007] EWHC 3077 (Pat) [25], quoted with approval in Qualcomm v Nokia [2008] EWHC 329 (Pat) [7], quoted with approval by Hughes J in Shire Biochem 2008 FC 538, [22] (and see 2009 FC 991 [88]; 2009 FC 676 [82]). Gleason J said that

However, the decision in Shire Biochem does not stand as authority for the proposition that it is proper to construe a patent with the infringing substance in mind, but, rather, only for the common sense notion that to be useful evidence and arguments in a case must be directed toward the issues that arise.

This is not quite right. What Floyd J said in Qualcomm, quoted with approval by Hughes J in Shire Biochem, is that

It is often said that a patent specification should be construed without reference to the infringement. Yet one cannot sensibly identify the point of construction without understanding what it is about the alleged infringement which is said to take it outside the claims.

So, when the claim calls for A, and the allegedly infringing device requires B, “the right question is . . . whether in its context in the specification the skilled man would appreciate that A in the claim encompassed B” (Qualcomm [25]). This necessarily requires focusing on the allegedly infringing device. Similarly, in Technip France SA's Patent (2004) RPC 46, quoted in Shire Biochem, Jacob LJ remarked that “in most sensible discussions of the meaning of language run on the general lines `does it mean this, or that, or the other?' rather than the open-ended `what does it mean'?” In Nokia, Pumfrey J prefaced the quoted remark with the statement that “it seems sensible to turn to the [allegedly infringing product] to identify the points of construction that arise” [25]. He specifically directed the parties to serve statements of the case identifying the relevant aspects of the allegedly infringing product for the experts to focus on in construing the claims, and he indicated that such statements are “essential” [26], at least in a complex case.

Two SCC decisions cited for the proposition that claim construction is independent of validity and infringement are Whirlpool 2000 SCC 67 and Sanofi 2008 SCC 61. In Whirlpool the Court said that “Claims construction is antecedent to consideration of both validity and infringement issues” [43]. This may be taken to say that construction should not be influenced by concerns over infringement or invalidity, but it does not say that claim construction should be undertaken without reference to the infringing product. In holding that the report of the experts who had been “blinded” should be preferred, Gleanson J cited Rennie J’s holding to the same effect in Omeprazole 3 2014 FC 638, [321]. Rennie J held that the “blinded” experts “had mandates that allowed them to opine on the state of the art, in the words of the Supreme Court of Canada, ‘viewed without any knowledge of the alleged invention as claimed’ (Sanofi-Synthelabo Plavix, at para 67).” But that statement in Sanofi was made in the fourth step of the Windsurfing / Pozzoli test for obviousness

(4) Viewed without any knowledge of the alleged invention as claimed, do those differences constitute steps which would have been obvious to the person skilled in the art or do they require any degree of invention?

What this is saying is that the obviousness inquiry must be undertaken without the benefit of hindsight. Claim construction is step (2) of the test, and Sanofi does not say that claim construction should be undertaken without reference to the infringing product.

Apart from these authorities, I see a danger in ignoring the infringing device. Suppose that almost all alcohols were simple linear chain molecules, so that there is normally no difference between counting -OH groups in the alcohol moiety and counting them in the molecule as a whole. If asked the general question of how to define a pentahydric alcohol, a skilled person might say “count the -OH groups in the molecule.” But if then presented with an unusual alcohol with two distinct parts, the skilled person might well say “Well, I never thought of that – of course I meant that when there are two parts to the molecule, you should count the -OH groups on the alcohol moiety.” Imagine trying to give legal advice to a client who insisted on asking “what are the requirements for a valid patent,” while refusing to disclose her invention, and who then insisted on holding you to your original definition after revealing the peculiarities of her own invention.

I take the point that claim construction should not be tendentious. But it seems to me that ignoring the infringing device in claim construction can also be misleading. Either course has its risks. Rather than preferring the evidence of experts who have been blinded, perhaps it would be better to recognize the problem of tendentious testimony on the one hand, and misdirected evidence on the other.

Finally, Gleason J remarked that

[96] Teva could easily have directed its experts’ attention to these issues by posing the question whether the terms “pentahydric or hexahydric alcohols” as used in the 310 Patent would connote a molecule or a moiety to the skilled person, without alerting the experts to the fact that the potentially infringing substance was [redacted].

I don’t find this persuasive. I agree that this would have properly directed the experts’ witness to where the shoe pinches, but I’m not sure this type of instruction is any better than revealing the allegedly infringing molecule itself. The experts knew that all the alcohols specifically mentioned in the specification were linear chain alcohols (mannitol, sorbitol and xylitol), and an expert instructed to consider whether the term connoted a molecule or a moiety could readily have inferred that the infringing substance had two distinct moieties, which would defeat the purpose of the blinding. More importantly, this approach demands an excessive degree of foresight from counsel; how can counsel, who are not normally themselves persons skilled in the art, know in advance which aspects of the allegedly infringing product are relevant to the claim construction? How would they known to present the molecule / moiety distinction before having an opinion from an expert that it was relevant? Perhaps in this case the particular point might have been obvious to a pharmaceutical litigators of ordinary skill, but as a general matter we cannot presume that litigators will know “where the shoe pinches” before they are told by experts.

Tuesday, April 14, 2015

Minister Cannot Issue NOC to Generic Licensee on Basis of Administrative Submission

Actelion Pharmaceuticals Canada Inc v Canada (Attorney General) 2014 FC 1249 Gleason J
            2,071,193 / bosentan / TRACLEER

This decision is a companion to Gleason J’s Exemestane 2014 FC 1243 decision, in which she held that when a first generic has received an NOC, a second generic which licenses from that first generic remains subject to s 5. Consequently, as discussed here, the second generic must serve an NOA on the patentee, and the Minister cannot issue an NOC to the second generic based on an administrative drug submission that cross-references the first generic’s submission.

The Bosentan case raised exactly the same question and Gleason J simply applied Exemestane and held that the patentee was entitled to a declaration that Health Canada had failed to comply with the PMNOC Regulations by failing to require the second generics to address the 193 Patent before issuing them NOCs for their bosentan products [9]. However, in this case the patent had expired by the time of judgment and Gleason J noted that an order quashing the NOC would potentially be moot. Since the mootness question had not been fully argued, she required the patentee to file additional submissions if it wished to pursue an order quashing the NOCs [10].

Monday, April 13, 2015

Arbitrary Selection Is Not Inventive

Janssen Inc v Teva Canada Ltd / bortezomib (NOC) 2015 FC 247 Barnes J
            2,203,936 / bortezomib / VELCADE

In light of the difficulty that is sometimes caused by selection patents, Barnes J’s Bortezomib decision is very welcome as a paradigmatic example of the correct use of selection patent concepts to strike down an invalid selection patent. Of course, invalid selection is not an independent ground for patent invalidity: 2010 FCA 197 [33]. In Canadian law that an invention is a selection is said to inform the analysis of all grounds of invalidity [ibid 27]. In his decision, Barnes J has focused entirely on obviousness; this is how European (EPO Guidelines G.VII.12) and US law (737 F 3d 731, 739 (Fed Cir 2013); 738 F 3d 1337, 1344 (Fed Cir 2013)) deal with selection patents, and it is, in my view, the correct emphasis.

Bortezomib is a proteasome inhibitor which is useful in the treatment of cancer [4]. The asserted claims of the 936 patent are to the compound bortezomib, as well as its use to treat cancer and a dosage form [7]. The prior art 904 patent disclosed a genus of compounds that includes bortezomib [54]. That Patent also disclosed that the disclosed compounds are potent proteasome inhibitors [27]. Barnes J noted that this meant that the 936 patent is a selection from the 904 patent [27], but this holding as such played no role in his analysis. His basic point was that “A person of skill is not doing anything inventive when he chooses options provided in a prior patent to build a molecule that he expects will work” [40]. The 904 patent tells us that all the compounds within the genus are potent proteosome inhibitors; bortezomib is a compound within the genus; therefore any skilled person would have know that bortezomib was a potent proteosome inhibitor, even before the 936 specifically disclosed it as such.

Barnes J’s use of the word “inventive” is helpful here. Much of the confusion related to selection patents because the inventive step requirement is expressed in terms of obviousness. In some sense it was not obvious to select bortezomib out of the myriad compounds in the 904 genus. But the fundamental question is not whether the selection was obvious, but whether it was inventive. (Prior to codification in s 28.3, what is now known as the non-obviousness requirement was normally called the inventive step requirement. Obviousness is only the test for inventiveness. The drafting of the EPC is explicit on this point: see Arts 52(1) and 56.) It might not have been obvious to select bortezomib out of the range of proteasome inhibitors described by the 904 patent, but neither was it inventive; it is what the EWCA has called an “arbitrary” selection: Dr Reddy's [2009] EWCA Civ 1362 [52]. The codification of non-obviousness in s 28.3 in conjunction with the endorsement of the Windsurfing / Pozzoli framework in Sanofi 2008 SCC 61, [67] which focuses on the “inventive concept,” means that in Canada this question is now often framed in terms of defining the inventive concept. So, in Bortezomib there was a question as to whether the inventive concept is only the compound bortezomib, or also the properties of bortezomib [20]. If the inventive concept is only the compound, then it would not be obvious to select the particular compound from the genus; but if the inventive compound includes the properties, it is obvious that bortezomib is useful in the same way that all the members of the genus are useful. In my view it is not particularly helpful in this context to debate what the inventive concept might be. Focusing, as Barnes J did, on whether the patentee had done anything inventive, addresses the same point, but it is more straightforward and less likely to lead to error.

While it is not inventive to discover that bortezomib is a proteasome inhibitor, it might be inventive to discover that it is a particularly effective proteasome inhibitor in comparison with others in the genus. Thus the requirement that a valid selection patent have a “a special property of an unexpected character” or “a substantial advantage over the genus from which it was selected” [44] flows directly from the inventive step requirement. The argument that bortezomib was unexpectedly superior was made by Janssen, but it failed on the facts. There was simply no evidence that bortezomib was particularly effective [44], [46]. Indeed, neither the patent nor Janssen’s experts even asserted that it was particularly effective [47], [48].

Janssen had also argued that bortezomib was not in fact a selection from the 904 genus. If that were true, then it might be inventive to discover an entirely new compound which was a potent proteasome inhibitor; but this argument also failed on the facts: [54].

One point of note is that Barnes J quoted at length from Kane J’s Travoprost 2014 FC 699 decision, which he described as “a very similar situation” in which Kane J applied “classic obviousness principles” in concluding that the patent was invalid. I do not read the Travoprost decision in the same way as Barnes J. As discussed here, as I read it, Kane J’s analysis turned on her construction of the inventive concept. That is, the central question was not whether the species actually had special and unexpected properties, but whether the inventive concept, properly construed, included those properties. I considered Kane J’s analysis on this point to illustrate how a focus on the inventive concept can lead to an erroneous analysis. I do hope that I have misinterpreted her decision, and that Barnes J is right to say that it applied classic obviousness principles in the same way as did Barnes J himself.

Finally, while on the facts as found by Barnes J the identification of bortezomib was not sufficiently inventive to support a patent, we should recognize that the development of bortezomib was an important medical contribution. While the 904 application disclosed a genus of proteasome inhibitors, to actually treat cancer it is necessary to chose one and take it through clinical trials to regulatory approval. This case illustrates that patents alone may not provide a sufficient incentive in this respect. What if all of the members of the 904 genus are in fact equally effective? In that case it would be impossible to get a valid patent, and yet without a valid patent it would not be worthwhile to conduct the clinical trials necessary for marketing authorization. The facts of Bortezomib illustrate that a robust data protection regime is an essential complement to the patent system.

Affirmed that Perfect Match Required for Listing on Patent Register for Medicine as Well as Formulation

ViiV Healthcare ULC v Teva Canada Ltd 2015 FCA 93 Near J: Ryer, Rennie JJA aff’g 2014 FC 893 Hughes J (here) aff’g 2014 FC 328 Milczynski J (here)
            abacavir & lamivudine / KIVEXA / 2,289,753

In this decision the FCA affirmed that the “perfect match” requirement for listing of a patent on the Patent Register applies under all branches of the s 4(2)(a) of the NOC Regulations. That is, each medicinal ingredient must be explicitly named in a claim; it is not enough that the generic product would necessarily infringe the claim and it is not enough that all the ingredients are explicitly listed in the description.

This holding is not a surprise. I have argued that the perfect match requirement is not sound as a matter of policy, but, as discussed in my posts on the decisions under appeal (here and here) and as the FCA held [17], the case at hand is not distinguishable from the leading case, Gilead / COMPLERA 2012 FCA 254 (blogged here).

While the Minister argued in favour of the perfected match requirement in the cases in which the requirement was initially established, it argued against the requirement in this case. Industry Canada has recently indicated that the NOC Regulations will be amended to “confirm established Health Canada practices in relation to the policy intent of the NOC Regulations.” It sounds like this will reverse the rule in ViiV v Teva, but I doubt the amendments will be entirely satisfactory. As discussed here, the Minister's position is that a perfect match is required under para 4(2)(b), which applies to "a claim for the formulation that contains the medicinal ingredient," but not under para 4(2)(a), which applies to "a claim for the medicinal ingredient.” But as I explained in that earlier post, the perfect match requirement may result in pointless formalism even when applied solely under para 4(2)(b).