What Is the Threshold for Influence in Inducement?

Apotex Inc v Janssen Inc 2024 FCA 9 Locke JA: de Montigny CJ, Goyette JA affg Janssen Inc v Apotex Inc 2022 FC 107 Manson J

2,655,335 / paliperidone palmitate / INVEGA SUSTENNA / NOC

This decision addresses the second prong of the Corlac 2011 FCA 228 [162] test for inducement in the pharma context. In some ways it is perfectly straightforward. Manson J found as a fact that Apotex’s product monograph (PM) was the ‘but for’ cause of direct infringement and the FCA affirmed on that basis. Locke JA’s decision nonetheless raises two notable issues. First, there seems to be an unresolved tension between this decision and Teva v Janssen 2023 FCA 68 as to the standard for influence on the second prong of the Corlac test; while this decision endorses a high threshold of ‘but for’ causation, Teva v Janssen apparently endorses a lower threshold. Second, while Locke JA explicitly endorsed a test of ‘but for’ causation, some of his comments imply a different approach.

Manson J’s decision was heavily redacted and I did not fully appreciate the issues when I wrote my post on it. The FCA decision has not been redacted and the key issues are much easier to understand. With that said, some aspects of the FCA discussion remain obscure to the extent that they turn on redacted discussion of the facts at trial. In my earlier post I had noted that Manson J’s decision in this case was essentially a companion case to Janssen v Pharmascience 2022 FC 62, which involved the same patent and the same type of generic product. However, the issues raised on appeal were quite different, so I’ll deal with the Pharmascience decision in a separate post.

Janssen’s 335 Patent relates to a dosing regimen for long-acting injectable paliperidone palmitate formulations for the treatment of schizophrenia and related disorders. It claims a dosage regimen comprising two loading doses of 150 mg-eq and 100 mg-eq, and subsequent monthly maintenance doses of 75 mg-eq [2]. In this NOC proceeding, Apotex brought a motion for summary trial on the basis that it would not infringe because it would not provide the 75 mg-eq dose, which is an essential element of the claimed invention [3]. That is, Apotex intended to market the loading doses but not the maintenance doses, which would be supplied by Janssen: see Pharmascience [8]. This looks like a strategy to get part of the market while trying to avoid infringing. At first instance, Manson J held in favour of Janssen, finding that Apotex would induce infringement of the 335 Patent with its generic version of INVEGA SUSTENNA [3].

The direct infringers would be the prescriber (eg a physician or a nurse practitioner) or patient [24] and Janssen therefore relied on inducement. Infringement by inducement is governed by the three part Corlac test, which Manson J correctly summarized as follows [4]:

There is a three “prong” test for inducement: (1) direct infringement by a third party; (2) the inducer influenced the third party to the point that the infringing act would not have occurred without the influence; and (3) the defendant knew that its influence would bring about the infringing act.

All three prongs were disputed at trial, with Janssen prevailing on all three [5]. On appeal, Apotex only took issue with Manson J’s decision on the second prong.

The second prong of the Corlac test is normally satisfied if the defendant supplies a non-infringing product accompanied by instructions to use that product in an infringing manner. So, in a typically skinny label case in the pharma context, the drug itself is unpatented but a use is patented. Direct infringement and knowledge are normally conceded, and the case will turn on the second prong, and in particular whether the generic has managed to scrub its product monograph (PM) clean of any reference to the infringing use. This case is different, as Apotex’s PM “would essentially be a copy of that for INVEGA SUSTENNA” [14]. I take it from all this that Apotex did intend to supply its non-infringing product — the loading doses — accompanied by instructions to use it in an infringing manner, which is to say as part of the patented regimen. The FCA decision isn’t more explicit than the phrase I have just quoted, and the FC decision isn’t clear on this because of the redactions, but on appeal in the Pharmascience case, it was no longer disputed that Pharmascience’s PM contained instructions concerning the use of the 75 mg-eq. dose (Pharmascience [11]) and while that is of course a difference case, my understanding is that the facts were essentially the same. This post therefore proceeds on that assumption.

Given that defendants are regularly found to be inducing infringement by supply of an unpatented product accompanied by instructions to use it in an infringing manner, and Apotex did just that, how did Apotex try to avoid liability?

Apotex’s main argument was to the effect that no one will read Apotex’s PM and therefore no one will be induced by it. This argument has been made before, and always failed: see generally my paper “Is 'But For' Causation Necessary to Establish Inducement?” But the point turns on the evidence, and it is at least plausible — no one is arguing that physicians always consult the generic PM and prescribe strictly according to it—so it is always worthwhile for a generic to try the argument again, to see if it can get a better result on the facts.

The legal aspect of the argument turns on the threshold for causation at the second prong. Corlac [162] itself put the test this way: “the completion of the acts of infringement must be influenced by the acts of the alleged inducer to the point that, without the influence, direct infringement would not take place.” This implies a test of “but for” causation. While there is certainly considerable caselaw that articulates the test this way, I have argued that other caselaw supports a less stringent test, along the lines of “encouragement to infringe,” and that there is very little caselaw in which the result actually turns on the precise test for causation at the second prong: see “Is 'But For' Causation Necessary to Establish Inducement?”

The threshold is important because if it is simply “encouragement to infringe,” then any party who sells a product along with instructions to use it in an infringing manner will be liable for inducement (if the other prongs are satisfied). But if the threshold is “but for” causation, then it matters whether the direct infringer actually pays any attention to the instructions. If no one actually reads the instructions and therefore no one changes their behaviour as a result of the instructions, then any direct infringement would have happened in any event and it cannot be said that “without the influence, the direct infringement would not take place.”

So, consider a hypothetical case where the patent relates to the use of ammonia, an unpatented bulk commodity, for a novel purpose, namely to encourage growth of subterranean bacteria to aid in fracking. The patentee, P, sells ammonia for that purpose. In Scenario 1, the direct infringer, DI, goes to P to buy ammonia for fracking, and discovers it is priced at $2000/t, while the market price for bulk ammonia is $400/t. DI politely walks out of P’s establishment and instead buys ammonia from a different vendor, V, who sells bulk ammonia to clients who mostly use it for fertilizer production. V does not have a licence from P to sell ammonia for fracking. DI nonetheless uses the ammonia purchased from V for fracking. DI is therefore a direct infringer. V did not influence the direct infringer at all. V had no idea what DI planned to do with the ammonia; indeed, V had no idea that ammonia could be used for growing subterranean bacteria and would have mocked the idea. It is clear law that V is not an indirect infringer, even though V supplied the ammonia used by DI to infringe. Doctrinally, it is the second prong—the influence requirement—that relieves V from liability in this example, because V did not induce influence DI’s plan. This would be true even if DI had told V what he was going to use the ammonia for, and V had replied “Sure, buddy, whatever. It’s yours once I sell it to you.” That is why the influence prong is separate from the knowledge prong.

Now, in Scenario 2, suppose that DI did not initially know that ammonia could be used for fracking, but V did. DI goes in to buy ammonia for another purpose, and V says “Do you know ammonia can also be used for fracking.” DI is curious and V hands DI an instruction sheet on how to use ammonia for the infringing purpose. DI says “That sounds great! I can make a bundle” — and buys some ammonia and uses it to infringe. In that case, V is clearly inducing infringement, even on the high “but for” threshold, since DI would not have infringed but for V’s influence.

Finally, in Scenario 3, suppose that both DI and V knew that ammonia could be used for fracking. DI goes in to buy ammonia for that purpose. At the time of sale, V hands DI an instruction sheet entitled “Three great things to do with ammonia.” The first thing on the list is fracking, with detailed instructions. DI takes the sheet, glances at the title, and thinks “I don’t need tips — I know perfectly well what I’m going to do with this ammonia.” DI scrunches the sheet into a ball and throws it away, without reading anything past the title. In that case, the instructions do not actually influence DI. The test from Corlac is “without the influence, direct infringement would not take place.” In this scenario, direct infringement would take place, even without the influence, and so, on its face, the second prong of the Corlac test is not satisfied.

Apotex’s basic argument, as I understand it, is that this is a Scenario 3 situation. Yes, it supplied instructions to infringe, in the form of its PM, but the direct infringers never paid attention. They got their information from Janssen’s PM, or from the scientific literature, but not from Apotex’s PM.

This argument rests on two pillars. The first pillar is legal. The threshold for influence has to be high; on an encouragement standard, it is enough that Apotex sold the product with instructions to use it in an infringement manner and it does not matter if no one paid attention to Apotex’s PM. The second pillar is factual. It has to be established on the evidence that the users did not pay attention to Apotex’s PM. If at least one person was (or would be, in an NOC action) actually influenced to infringe by the PM, then it does not matter whether the threshold is high or low.

Turning to the first pillar, what is the threshold for influence? In this case, Manson J clearly endorsed a strict “but for” causation:

[127] The “but for” influence required in the second prong of the Corlac test sets a high bar – higher than “encouragement to infringe,” a “subtle reference” to the infringing use, or “attempting to induce others to infringe.”

Locke JA affirmed this high threshold:

[13] Apotex urges this Court to treat the threshold as a “but for” test, and it cites jurisprudence that calls this a causation test. I have no objection to Apotex’s characterization of the test, but I do not find it to be more helpful that the “would not have occurred without” and “sine qua non” characterizations provided in the jurisprudence of this Court. In my view, these are merely different ways of describing the same threshold.

This clearly affirms a high “but for” threshold, pointing out some equivalent ways of stating the same test.

While this statement is clear, there is a problem. In Teva v Janssen 2023 FCA 68 revg Manson J’s decision in 2020 FC 593, the FCA apparently endorsed a lower threshold, calling the “but for” threshold which Manson J had endorsed [FC 264] “unduly onerous” [FCA 82]: see here for my discussion. As just noted, in this case Locke JA clearly endorsed a “but for” causation threshold for influence, so it looks to me like there is a tension between these cases. Perhaps I misunderstood Teva v Janssen and read too much into it. I am not going to try to sort this out in this post, except to say that it is not clear to me how reconcile the decisions. Presumably the FCA will clarify matters in due course. I note that the panels were different — Stratas, Gleason, Woods JJA (for the Court) in Teva v Janssen, with Locke JA, de Montigny CJ and Goyette JA concurring, in this case. Locke JA was clearly aware of the Teva v Janssen decision, as he referred to it, albeit not directly on this point [8].

In any event, it is clear that the high “but for” threshold applies for the purposes of this case. One pillar of Apotex’s argument is in place.

What about the second pillar? Apotex argued that the evidence was that the direct infringers did not pay any attention to the PM. This is how Manson J described it:

[132] The crux of Apotex’s argument related to this second prong is that the experts all agree that the ultimate dosing decision is based on physician skill and judgment, not the language in the product monograph. Further, physicians have been steeped in the claimed dosing regimen throughout their training, clinical experience, and use of the INVEGA SUSTENNA® product. As such, the introduction of the product monograph for the APO Product will not influence or change their prescribing practices and will not amount to the level of influence necessary to meet the second prong of the Corlac test.

This argument was rejected by Manson J, who said the following:

[147] Notwithstanding the exercise of skill and judgment by prescribing physicians in selecting the dosing regimen for patients, the evidence before the Court in this case establishes that acts of infringement will be influenced by the acts of the alleged inducer, Apotex, to the point that, without the influence, direct infringement will not take place. Apotex’s product monograph will influence prescribers and patients to implement the claimed dosage regimen, thereby directly infringing the 335 Patent.

[148] I am satisfied, on the evidence before the Court that Janssen has proven, on a balance of probabilities, that at least some prescribers of the impugned APO Product will be sufficiently influenced by the Apotex product monograph to induce infringement by those prescribing physicians.

The details of the evidence are not clear due to the redactions, but this conclusion certainly looks like a finding of fact that some direct infringers would be influenced, even on a ‘but for’ test.

Apotex tried to challenge this finding, arguing that “none of the experts went so far in their testimony as to state explicitly that any direct infringement using Apotex’s product would not occur without Apotex’s influence” [26]. Locke JA rejected this on the straightforward basis that this is a factual finding that Manson J was entitled to make:

[25] As indicated at paragraph 11 above, the Federal Court concluded at paragraphs 147 and 148 of its reasons that such infringement would be influenced by Apotex to the point that, without the influence, direct infringement would not take place. It is in this sense that the Federal Court was entitled to conclude that the second prong of the test for inducing patent infringement was met.

[27] In my view, this is a factually suffused issue on which Apotex would have to establish that the Federal Court made a palpable and overriding error. The Federal Court did not err in law.

That all seems to say that in fact, on the evidence, the direct infringers would be influenced by Apotex to the extent that the influence prong is satisfied even with a legal threshold of “but for” causation. Thus, Apotex succeeded on the legal pillar of its argument, but failed on the factual pillar. That is what I had understood to be the basis for Manson J’s decision when I wrote my post on it.

That much is very straightforward, and I would not normally have devoted so much space to it. But it is necessary context to some further, more puzzling remarks by Locke JA. Apotex’s central argument on appeal was as follows:

[14] Apotex relies on the fact that its product monograph would essentially be a copy of that for INVEGA SUSTENNA, and the assertion that prescribing practices of physicians would not change if Apotex were allowed to market its generic version of INVEGA SUSTENNA. In light of the foregoing, Apotex argues that any influence that its product monograph could have on any ultimate act of direct infringement could not rise to the level required in Corlac.

As we’ve just seen, Locke JA rejected this argument on the facts. Here is the puzzle — he also rejected it on the law:

[23] The main weakness of this argument is that it depends on there being a requirement that prescribing practices of physicians be altered because of Apotex’s activities. In fact, this is not necessary.

I’m not sure how to reconcile this statement with the “but for” threshold which Locke JA endorsed. The test from Corlac is “without the influence, direct infringement would not take place.” This clearly states that the influence must cause a change in outcome — with the influence, there is infringement, without it, there is not. If nothing changes as a result of the influence, how can it be said that the influence caused anything?

Locke JA continued (original emphasis):

[23] What is required is that the ultimate act of direct infringement occur because of Apotex’s activities. . . .

[25] It follows that, even if the practices of prescribing physicians were to remain unchanged following the introduction to the market of Apotex’s generic version of INVEGA SUSTENNA, the fact would remain that activities by patients (and by prescribers) that had previously been non-infringing (because the drug was sourced from Janssen) would be infringing once the drug was sourced from Apotex, an unlicensed supplier.

Here, Locke JA seems to be saying that indirect infringement is established because the direct infringers would use Apotex’s product instead of Janssen’s. That is undoubtedly true. But that is not enough to establish infringement by inducement. It is also true in Scenario 1. DI’s activity, that would have been non-infringing if the ammonia had been sourced from P, is infringing because it was sourced from V instead. But that in itself does not make V liable for inducement. It is uncontroversial and long-established law that a party who does no more than supply a non-infringing product that is used to infringe is not thereby an infringer, even if it causes the direct infringer to change suppliers; that is why the second prong of the Corlac test requires influence by the indirect party. This is entirely apart from the question of the specific threshold for influence.

What was motivating these remarks? Why did Locke JA not simply affirm Manson J on the factual finding of “but for” causation, and leave it at that? I would speculate that Apotex’s argument on the facts was quite strong, notwithstanding Manson J’s finding. (As noted, the evidence supporting his finding is unclear due to the redactions.) My impression from the evidence in the pharma inducement cases generally is that the substantial majority of physicians and pharmacists really do not rely on the generic PM and the courts are struggling with the evidence to make the result come out the way it ‘should’ by finding that at least some do rely on the generic PM. Perhaps Locke JA wanted to provide an alternative basis for the decision that would not turn on the factual finding alone. Perhaps this was to set the stage for the day that a trial judge finally decides on the facts before them that physicians do not read the generic PM. If so, I sympathize with Locke JA’s dilemma, but the statements in these two paragraphs seem to go beyond endorsing a lower threshold and seem tantamount to removing the influence requirement altogether.

With that said, I do not disagree with the result. In my view, there are two sound and principled bases by which the same result could be arrived at, even if it had been established on the facts that the direct infringers did not pay any attention to Apotex’s PM. One would be contributory infringement, a US doctrine under which the mere supply of good especially adapted to infringe, constitutes indirect infringement, without any need for inducement. For reasons I explain in my article “Contributory Infringement in Canadian Law” (2020) 35 CIPR 10, I am of the view that contributory infringement is a sound doctrine which is consistent with the bulk of Canadian caselaw. However, it is not recognized in Canadian law: see MacLennan 2008 FCA 35 [33], [38], [40]. And is most certainly not the basis for Locke JA’s decision in this case. The other basis would be under a lower “encouragement” threshold for influence. But that was not used by Locke JA in this case.

I’ll end this post on a different point entirely. As noted in my post on Manson J’s trial decision, this was essentially a companion case with Janssen v Pharmascience 2022 FC 62, also before Manson J, in which Pharmascience sought to launch the same kind of product as Apotex, ie the loading doses but not the maintenance doses. There was considerable overlap between Manson J’s two decisions — indeed, too much. Parts of para 138 of his Apotex decision were “lifted” from his Pharmascience decision, and consequently contained some factual inaccuracies; for example, para 138 referred to four experts, when there were in fact five in the Apotex case [17]–[18]. While Apotex tried to make something of this on appeal, Locke JA concluded that this was a “slip of the pen” error which did not demonstrate that Manson J had failed to fully consider the issues [20].

Inducement in the Pharma Context is an Inherently Hard Problem

Apotex Inc v Janssen Inc 2023 FCA 220 Locke JA: Mactavish, Monaghan JJA affg 2022 FC 996 (reasons) 2022 FC 995 (judgment) Pallotta J

            2,659,770 / macitentan / OPSUMIT / NOC

This brief decision, affirming Pallotta J’s finding at trial that Apotex’s sale of Apo-Macitentan would induce infringement of Janssen’s 770 patent, deals with a difficult issue of inducement in the pharma context. Two main doctrinal points emerge: first, explicit instructions to infringe are not necessary to establish inducement, and second, it may sometimes be effectively impossible for a generic to avoid inducement by scrubbing its product monograph (PM) clean of references to the infringing use. More broadly, this case illustrates why indirect infringement in the pharma context is an inherently difficult problem that does not appear to have any good solution.

As discussed here, this is in many ways a typical ‘skinny label’ case, in which a generic seeks to sell a drug that is itself unpatented, but which may be used in a manner that is patented. In this case, Janssen’s 770 patent covers the use of macitentan in a combination therapy with a PDE5 inhibitor in the treatment of pulmonary arterial hypertension (PAH). Apotex sought to sell macitentan, which is itself unpatented, for monotherapy treatment of PAH. Combination therapy is about 80% of the market and monotherapy is about 20% [FC 162].

In such cases, the generic is not a direct infringer, so infringement by inducement must be established under the three-part Corlac test, 2011 FCA 228 [162]. This typically reduces to the question of whether the generic’s PM will induce infringement by prescribing physicians, who will read the PM and thereby be induced to prescribe the generic product for use in an infringing manner. A central question is therefore whether the generic’s skinny label has been sufficiently scrubbed clean of any reference to the infringing use. This turns on the details of the generic PM.

In this case, the details are a bit difficult to follow because of redactions, but the key point is that much of the information in the Apo-Macitentan PM is clinical trial data from a landmark SERAPHIN study, which established the efficacy of macitentan for both monotherapy and combination therapy [12], [FC 186]. PAH is a rare disease and macitentan can only be prescribed by about 30 specialists who work in recognized PAH centres [FC 184]. All of these specialists would be aware of the landmark SERAPHIN study and so even though the Apotex PM was scrubbed clean of explicit references to combination treatment, specialists reading the Apotex PM would recognize it referenced the SERAPHIN and so would understand that Apo-Macitentan is also suitable for combination therapy. This was the basis on which Pallotta J found that Apotex’s PM would induce infringement.

The first key doctrinal point is that explicit instructions to infringe are not necessary to establish the second prong of the Corlac inducement test:

[17] The weakness of Apotex’s position in this regard is that it assumes that an absence of explicit instruction and of intention that direct infringement should result equals an absence of influence sufficient to satisfy the second prong. That is not necessarily the case. While explicit instruction and intention may be relevant to the issue of influence, I do not accept that either is required. Even without explicit reference to combination treatment, the Federal Court was entitled to find that the Apo-Macitentan PM would influence use of macitentan in that way.

In Novopharm 2007 FCA 167 [11], the FCA remarked that “an inducement to infringe generally cannot be inferred from a mere reference to the new use in the product monograph, for example, in the course of explaining contraindications or drug interactions, or as part of a list of scientific references.” Locke JA [13]-[14] distinguished this on the basis that in Novopharm the unpatented use of the drug in question was for an entirely different use, whereas in this case, the unpatented use was for the same indication. This is consistent with the qualifier “generally” in Novopharm and this holding may be explained by the unusual expertise of the end-users in this case who would recognize the indirect allusion. Nonetheless, the point remains that in some cases a “mere reference” to an unpatented use may indeed suffice to establish inducement.

The second doctrinal point is implicit. So far as I can tell, there is nothing that Apotex could have done to avoid inducement. Certainly there was no suggestion in either the FC or FCA decision that Apotex could have done more to scrub its PM of references to the combination therapy. And in this case, there was no suggestion that there would have been any way for Apotex to have avoided such a reference. The implication is that it is simply not possible for Apotex to sell macitentan for the unpatented use.

The broader point illustrated by this case is that the problem of inducement in the pharma context is inherently very difficult. Ideally we want to allow free generic entry in the market for the unpatented use while at the same time giving the patentee exclusivity in the market for the patented use. The obvious way to do that is to sue the direct infringer, which allows the patentee to control how the product is used after it is sold. But sometimes it is not practical or desirable to sue the direct infringer, as is particularly the case with respect to pharma. That is when inducement is most important.

The law of inducement tries to target the direct infringer indirectly. The theory underpinning the Corlac test is that the product sold by the defendant will be used in accordance with the instructions, so the product will only be used to infringe if the instructions supplied by the defendant instruct infringement. This is a second-best solution, but it can work tolerably well if the end-users generally read and follow the instructions provided by the defendant. But that theory appears to be substantially wrong in the pharma context. The basic problem is that physicians are to some extent experts, with their own independent sources of information. The situation in this case, where there are only a handful of prescribing physicians who are truly expert in this area, is just an extreme version of a pervasive problem. Of course, there will be some people who are experts in almost any field, but the problem of general end-user (physican) expertise means that the standard theory is particularly weak in the pharma context.

The result under the Corlac approach is that the generic will only be allowed to sell its unpatented product if it scrubs its PM clean of any references to the infringing use. The generic will also argue that even if its PM does instruct an infringing use, that doesn’t constitute infringement because the prescribers don’t pay any attention to the generic PM in any event, though I don’t believe that argument has ever prevailed on the facts. This approach has an air of unreality about it. I can’t help but imagine a scenario in which no doctor ever reads the PM for a generic product, except one scrupulous physician in a rural community where there is nothing else to do, who compulsively reads and follows the PM for any drug she prescribes. If the generic PM instructs infringement, she will be induced, so it is essential to scrub the generic PM clean in order to avoid liability. And then one day she retires, and now the generic PM can say whatever it wants.

In any event, even if there is some substantial number of physicians who pay attention to the generic PM, it seems clear enough that most do not. (And my understanding is that physicians normally prescribe generically and pharmacists do not normally know the indication, so even if the physician knows that the generic drug is not indicated for the patented use, it may be dispensed for that use nonetheless.) This means that if the generic is permitted to sell its product, we will get desirable competition in the unpatented market, but at the price of substantial infringement in the market for the patented use.

What can be done? One approach would be to favour one side or the other. In AB Hassle v Apotex 2002 FCA 421 [57], the FCA gave a strong policy statement suggesting we should favour competition in the unpatented market:

Thus [the defendant] cannot be prevented from obtaining [marketing authorization] solely on the basis that it will sell [the known compound]. If it were otherwise, then serious policy issues would arise. If there was any likelihood that a patient would consume a generic product for a patented use, then the generic product would not be approved. This would prevent new uses from being approved for existing drugs because there is always the possibility that someone somewhere will use the drug for the prohibited, patented purpose. This would result in a real injustice: since a generic company cannot possibly control how everyone in the world uses its product, the prevention of the generic from marketing the product would further fortify and artificially extend the monopoly held by the patent holders. The patent holder would, therefore, effectively control not just the new uses for the old compound, but the compound itself, even though the compound itself is not protected by the patent in the first place. The patent holders, as a result, would obtain a benefit they were not meant to have. In the end, society would be deprived of the benefit of new methods of using existing pharmaceutical medicines at a lower cost.

This logic is sound, but we have to keep in mind that we are striking a balance between the right of the patentee to exclusivity for its patented use and the need to incentivize development of new and useful treatment, and the right of the generic to sell the unpatented product for unpatented purposes and the desirability of competitive prices in the unpatented market. That balance is easy to strike if the only problem is that “someone somewhere will use the drug for the prohibited, patented purpose.” But what if the problem is that everyone everywhere will use the drug for the patented purpose?

Suppose a drug has long been generic, like aspirin, and the patentee invents a new use that represents 0.1% of sales. It seems clearly wrong as a matter of policy to prohibit the generic from selling aspirin, even if we know with certainty that it will be used for the infringing purpose. The benefit to the public of allowing generic prices for 99.9% of uses more than outweighs need for an incentive to innovate for niche uses. But conversely, if the patented use represents almost all of the market, it seems wrong to undermine that exclusivity simply because there is an unpatented niche use. The need to provide an incentive to develop major innovations that constitute the primary use for the compound outweighs the desire to allow generic prices in the niche market. So, if we can’t control direct infringement either through a direct action against the end-user or by restrictions on the product instructions, it is reasonable on policy grounds to say that the generic should be allowed to sell the product if the unpatented uses predominate but should not be permitted to sell the product if the patented uses predominate.

If we apply that reasoning to the facts of this case, in which the unpatented use is only about 20% of the market, it seems reasonable on policy grounds to prohibit the sale of the generic product. Yes, this is a bad solution and it shuts down competition in the market for the unpatented use, but all the solutions are bad—the question is which solution is the least bad. In my post on Pallotta J’s decision, I suggested that her finding may have been motivated by “the fact that combination therapy is the primary use for macitentan, with only 10–30% of patients getting monotherapy [162]. If the opposite were true, so that eg only 10% of the use was in the patented combination, then I wonder if it would have gone the other way.” This view is consistent with the two doctrinal points emerging from Locke JA’s decision on appeal. Taken on its own, the idea that a mere indirect reference to the patented use can support a finding of inducement is novel and somewhat extreme, as is the notion that it may be impossible for Apotex to sell a generic version of this drug. But both are readily understandable if the FCA was stretching existing doctrine in order to strike an appropriate balance by protecting the primary market.

I don’t really know where to go from there. I am reluctant to suggest abandoning the Corlac approach, which focuses on the instructions, even in the pharma context. The Corlac approach is logical and principled, and it works reasonably well in many contexts. The rule just doesn’t work as well in the pharma context, where most physicians do not get their information about how to use a generic drug from the manufacturer’s instructions. But it would be difficult to implement a market share approach in the pharma context, even if this could be done doctrinally. If the market shares for the patented and unpatented uses are 1%/99% or vice versa, it seems easy to say whether we should allow the generic sales. But where is that line to be drawn? Any firm line—50/50? 30/70?—seems arbitrary, but any line that turns on the facts will likely be both arbitrary and unpredictable. All I can say is that this case illustrates that the problem of inducement in the pharma context is very difficult, and I don’t see any easy solution.

Second Prong of Corlac Does Not Require “But For” Causation

Teva Canada Limited v Janssen Inc 2023 FCA 68 Stratas; Gleason; Woods JJA varg Janssen Inc v Teva Canada Ltd 2020 FC 593 Manson J

2,655,335 / paliperidone palmitate / INVEGA SUSTENNA / NOC action /

FC Infringement / FC Inducement

This decision represents a significant and welcome clarification of the law of inducement. It will be the new leading case on the second prong of the Corlac test, making it clear in order to establish inducement is it enough to show that the product was sold along with instructions to use it in an infringing manner. A couple of points regarding the construction of use and Swiss-type claims are also clarified.

The 335 patent relates to a dosing regimen for long acting injectable paliperidone palmitate formulations for treatment of schizophrenia. It has three sets of claims. Claims 1 to 16 are “product” claims that relate to prefilled syringes containing paliperidone palmitate adapted for administration in accordance with the claimed dosing regimens [15]. Claims 17 to 32 are “use” claims directed towards use of a dosage form according to the same regimen [16], and Claims 33–48 are Swiss-type claims to the use of paliperidone “for the preparation [or manufacture] of a medicament” [17] according to the same regimen. All claims were asserted.

In the judgment under appeal, Manson J found the asserted claims to be not obvious and valid. He also found that Teva would directly infringe the product and Swiss-type claims, but not the use claims. However, he found that Teva would not induce infringement of any of the claims [2]. Teva appealed on obviousness and direct infringement, and Janssen cross-appealed the finding of no inducement. Teva’s appeal on obviousness was dismissed for reasons specific to the case: [56]–[67].

Claim Construction

Product claim “for” use

Claim 1, which is representative of the product claims, is to prefilled syringes adapted “for administration” according to the dosage regimen [FC 124]. As discussed here, at first instance Teva had argued that it did not directly infringe the product claims because it did not actually prescribe or administer medications [FC 234]. Manson J rejected this argument on the basis that it is enough that “the capable, approved and intended use” for the Teva product incorporates the essential elements [FC 252]. The FCA affirmed, saying

[77] In the context of product claims like those in claims 1 to 16 of the 335 Patent (i.e., claims to a pharmaceutical preparation for use in the treatment of a condition), evidence that a generic company proposes to make or sell its product for the patented use (even if it is only one use among others) is enough to establish direct infringement in an action brought under section 6 of the PMNOC Regulations

In effect, a claim to a product “for” a use covers a product adapted and intended for that use, and not just the product when actually so used. The clarification is welcome. The FCA cited AB Hassle 2001 FCT 1264 [6], [33], [35–36] 2002 FCA 421 and Lilly v Apotex 2019 FC 884 [24–33] (discussed here). Both do stand for that proposition, but AB Hassle found that the product was not intended for the patented purpose, and Lilly v Apotex only concerned a motion allowing Lilly to amend its pleadings, so this seems to be the first time we have the point confirmed at the FCA level.

Swiss-type claim

In Novartis 2013 FC 985, Hughes J held, in effect, that a Swiss claim should be construed as a use claim, even though a Swiss claim is a product claim on its face: see here. At trial in this case, Teva urged Manson J to adopt the same position: [FC 159]. Relying on Hospira 2018 FC 259 [152–153], [268–323] affd 2020 FCA 30 [16–18], Manson J held that Swiss-type claims should be interpreted as being infringed if the medicament “is adapted for” administration according to the claimed dosing regimen [FC 163]. (So, the use claims and the Swiss claims both encompass products adapted for the infringing use.) The FCA affirmed:

[78] Similarly, in the context of Swiss-type product claims like those in claims 33 to 48 of the 335 Patent (i.e., claims to the use of a drug for the preparation of a medicament for use in treatment of a condition), evidence that a generic company proposes to make or sell its product for the patented use (even if it is only one use among others) is enough to establish direct infringement in an action brought under section 6 of the PMNOC Regulations

While the point was already established by Hospira, this is helpful in succinctly summarizing the law. It is also helpful in reaffirming the law, which can now be considered settled.

Inducement

The most important aspect of the decision relates to infringement by inducement, and in particular the second prong of the test for inducement set out in Corlac 2011 FCA 228, which requires that the act of direct infringement was influenced by the alleged inducer. The second prong was determinative at first instance; Manson J found that Janssen had established the first of Corlac factor, namely direct infringement, but not the second, that the acts of direct infringement would be influenced by the alleged inducer [48]. (He did not go on to consider the third factor, the knowledge requirement.)

The FCA reversed Manson J on the second prong, after a thorough review of the caselaw which greatly clarified the nature of the “influence” requirement. It is also significant that this was a decision by the Court, rather than by a single judge with concurrences, as is more usual. The decision is evidently intended to settle this point.

The second step of the Corlac test requires that [46]

the completion of the act(s) of infringement were influenced by the acts of the alleged inducer to the point that, without the influence, direct infringement would not take place.

On its face this seems to articulate a “but for” test, such that inducement is only established if the direct infringement would not have occurred but for the influence of the indirect party. Accordingly, Manson J explicitly characterized Corlac as requiring “but for” causation [FC 263], as have other cases, such as Bayer 2015 FC 797 [31] and Janssen 2019 FC 1355 [234] (both cited by Manson J). However, in my paper “Is 'But For' Causation Necessary to Establish Inducement?” (available on SSRN), I argued that the jurisprudence is not as clear on the standard for influence as this articulation of the test suggests. There are very few cases which actually apply a “but for” standard. The strongest case supporting a strict “but for” requirement is Slater Steel (1968), 55 CPR 61, 87, which stated that the influence must be the “sine qua non” of the infringement, and even that case is problematic authority, for a variety of reasons addressed in my paper. (And see also Nycomed 2011 FC 1441 affd 2012 FCA 195.) On the other hand, there are other cases, such as Novopharm 2006 FC 1411 affd 2007 FCA 251 and Genpharm 2003 FC 1443 affd 2004 FCA 413, that support a weaker “encouragement” standard.

As discussed here, Manson J acknowledged this apparent split in the caselaw [FC 259–64]. He explained it as a split is between the “earlier cases” [FC 264], and the “more recent cases” [FC 262], saying the earlier cases only required “some nexus” to the generic company [FC 261], while “in more recent cases,” the Court has “scrupulously” applied the Corlac test, “with particular focus on the second prong” [FC 262]. The implication is that Corlac changed the law, or at least the existing standard has been applied more stringently since Corlac. In particular, Manson J stated expressly that “[t]he ‘but for’ influence required in the second prong of the Corlac test requires a higher threshold for establishing inducement than was applied in the earlier cases” [FC 264].

The FCA reversed (all bold is my emphasis):

[82] We agree with Janssen that the Federal Court erred in law in holding that the decision of this Court in Corlac changed the law by incorporating a higher degree of causality at the second step of the analysis for inducing infringement. This error led the Federal Court to incorrectly apply an unduly onerous requirement at the second prong of the analysis for inducement and to incorrectly focus only on the skill and judgement of prescribing physicians to the exclusion of the role played by Teva in inducing infringement of the use claims in suit.

This statement by the FCA is clearly referring to Manson J’s remark at [FC 264], stating that “but for” was a higher standard for causation than had previously been required.

As discussed here, due to redactions in Manson J’s decision, it was not clear to me whether the “but for” element was crucial to his holding, but whether or not it was crucial to the outcome, it is clear that he applied a “but for” standard for causation. The FCA in the above paragraph described this standard as “unduly onerous.” Further, this statement is clearly referring to Manson J’s remark at [FC 264], in which he stated that “but for” was a higher standard than had previously been required. So, while the point might have been made more explicitly, putting these together, this implies that the influence exerted by the inducer need not be the “but for” cause of the direct infringement.

In the course of a thorough review of the caselaw, the FCA also specifically approved two cases articulating an “encouragement” standard. The FCA at [97] quoted with approval Novopharm 2006 FC 1411 [40], [42] affd 2007 FCA 251, in which von Finckenstein found inducement on the basis of a Product Monograph which “would be an encouragement” to infringe. The FCA [98] also quoted a passage from the same decision in which the application judge found the PM would “have the effect of inducing or encouraging” physicians to prescribe for the patented indications. The FCA at [95]–[96] also approved Genpharm 2003 FC 1443 affd 2004 FCA 413, in which “Justice Layden-Stevenson concluded that references (some of them subtle) to the patented use of omeprazole in Genpharm’s PM was sufficient to establish that Genpharm would infringe AB Hassle’s patent if Genpharm’s drug were allowed onto the market” [95] (my emphasis). These references reinforce the view that the FCA is endorsing an “encouragement” standard rather than a “but for” standard at the second prong. Conversely, in discussing Slater Steel, the FCA made no mention of the “sina qua non” passage, but quoted only passages referring to a more ambiguous “induced or procured” test [86].

The FCA also quoted with approval from Hospira 2018 FC 259 [326]–[327], [332]–[333], in which Phelan J held that “Infringement by inducement may be established by inferences reasonably drawn from the contents of the product monograph,” so that the second prong can be established when “the product monograph amounts to instructions or directions for infringement” [333]. In my article, I had suggested that this and other similar cases established a presumption that the physician would follow the directions in the PM, leaving open the possibility that this presumption could be rebutted by evidence showing eg that physicians never read the PM. That is, I took these cases to be saying that inducement “may” be established by showing that the PM instructs infringement, not that it necessarily “would” be established by such a showing. However, the FCA did not make any reference to a presumption, which is noteworthy in itself, as it suggests that these cases are not to be understood as turning on a presumption; the implication is that the instructions to infringe will in themselves establish infringement without the aid of a presumption. The FCA also pointed out that in Windsurfing (1985) 8 CPR(3d) 241 (FCA), the FCA had affirmed a finding of inducement when the inducer had sold a kit accompanied by instructions to assemble the components into a patented product, saying “I think it beyond dispute that the only inference to be drawn from the voluminous evidence in this case is that the respondent knew and intended that the ultimate purchaser would utilize the sailboard parts for the assembly of a usable sailboard which, upon assembly, would infringe the appellant’s patent” [85]. This also implies that strict but for causation is not required; it is enough that the inducer knew and intended the direct infringement.

In light of all this, the FCA reversed Manson J on the basis that he had made an error law at the second step of the Corlac test. The FCA stated:

[110] In the case of a generic drug, inclusion as one of the recommended uses within the PM for the drug of the alleged infringing use, among others, has been found to be sufficient to constitute the requisite encouragement to satisfy the second prong of the test for inducement in Hospira, AB Hassle, and Novopharm. In such circumstances, the infringing use is one of the bases for approval of the generic drug by Health Canada and one of the uses recommended to physicians. . .

This expressly states that all that is required is “encouragement.”

[112] Here, the Federal Court found that the capable, approved and intended use for the Teva product incorporated all the dosing and administration elements of the product claims, including the use of the continuous maintenance doses claimed in the 335 Patent. This finding inevitably leads to the conclusion that Teva would induce infringement of the use claims. Had the Federal Court properly understood and applied the test for induced infringement, no other conclusion was possible.

The reference to the approved use in [112] and [110] evidently reflects the PM. This is an explicit statement that if the PM instructs infringement, the second prong of the Corlac test is established.

The FCA did not explicitly state that “but for” causation is not required at the second prong, and indeed the Court repeated the Corlac test in very similar terms [109]. Nor did the Court expressly adopt an alternative “encouragement” test. Nonetheless, taken as a whole, it is now clear that in order to establish inducement, it is not necessary to establish that the influencer is or would be the “but for” cause of the direct infringement. It is now clear that the sale of a product accompanied by instructions to use the product in an infringing manner will be sufficient to establish the requisite influence. (Indeed, that is ultimately what happened on remand in Corlac itself: Corlac 2012 FC 76 [16]–[17].) This does not turn on any rebuttable presumption that the end-users will read the instructions and use the product accordingly, as no such presumption was mentioned by the FCA.

This will not affect the outcome in most cases; in the past, a party who sold a product accompanied by instructions to infringe would almost invariably be found to have caused the direct infringement. But the clarification should simplify the litigation. In the past we have seen a number of cases debating whether direct infringers, such as physicians and pharmacists, actually read the PM: see Abbott v Novopharm 2006 FC 1411 [40]; Solvay 2008 FC 308 [192]; Aventis 2006 FC 861 [44]; Allergan 2011 FC 1316 [161]. Presumably such evidence is now irrelevant.

I would also say that the FCA in this case truly did clarify the law. Manson J was not off on some frolic of his own; the view that “but for” causation was required at the second prong had a clear basis in Corlac, and was no doubt the most widespread reading of that decision. On the other hand, this is not a case in which the FCA reversed itself in the guise of clarification. The statement in Corlac was not necessary to the result, and was in tension with the holding and results in other cases, creating a real ambiguity in the law, which has now been resolved.

Finally, I will repeat a point I made in my article, which is that there is a separate question of causation with respect to damages. If infringement by inducement is established, the court may, and almost invariably will, grant injunctive relief. But in order to establish damages, it is necessary for the patentee to prove it suffered losses caused by the infringement, and “but for” causation must be established at that stage. (If an accounting is sought, “Nova v Dow” causation must be established.) In other words, encouragement to infringe is enough to get an injunction, but not enough to get damages.

Accounting Elected in the Alternative

UPL NA Inc v AgraCity Crop & Nutrition Ltd 2022 FC 1422 Aylen J

2,346,021 / flucarbazone sodium herbicide / EVEREST / SIERRA / HIMALAYA

I am finally starting to catch up on some of the cases I missed during my blogging break, but the end of term is busy and my blogging will continue to be erratic until late May.

In this decision, released last November, Aylen J found Arysta’s 021 patent to be valid in the face of obviousness and anticipation attacks. (The headline plaintiff, UPL, is a related company.) Infringement by AgraCity was conceded [3], [82]. Aylen J awarded $227k on an accounting of AgraCity’s profits. Relatively little infringing product was sold because an interim injunction had been granted soon after AgraCity had launched: see 2019 FC 530, blogged here and here. The decision turned largely on the facts, but there is an interesting point on the limits of inducement. The trial was not bifurcated, and, unusually, AgraCity was permitted to elect an accounting if its damages award did not meet a certain threshold. This seems to be inconsistent with the rule set out in AlliedSignal (1995), 61 CPR (3d) 417 (FCA), but the point does not appear to have been contested.

The 021 patent relates to a selective herbicide, namely flucarbazone sodium, and its use in weed control. Herbicides may be total herbicides, which kill all plants, or selective herbicides, which kill weeds with minimal injury to crops [118].

The key prior art was US Patent 5534486 and the corresponding Canadian Patent 2,064,636; like the 021 patent, both were originally granted to Bayer [30], [39], and the 486/636 and 021 patents had several overlapping inventors. There was not much difference between the 486 and 636 patents, and Aylen J focused her analysis on the 486 patent [157].

The 486 patent claimed a large genus of compounds and set out 327 preparation examples [33]. Eleven compounds were specifically claimed, including Claim 10 to flucarbazone and its salts [38]. Claim 10 was simply to the compound as such, not to any use. The 486 patent also disclosed that the claimed compounds are herbicides, but it was ambiguous as to whether they were total or selective herbicides [161], [182]. The 486 patent disclosed that some compounds had been tested and found to be effective herbicides, but there were few details: “No specific test results were included. For the tests that were conducted, the 486 Patent does not disclose which weeds were tested or which crops were tested, nor was flucarbazone or flucarbazone sodium included among the compounds that were tested” [182]; and see similarly [34]–[37].

Claim 1 of the 021 patent is to a selective-herbicidal composition, comprising an effective amount of flucarbazone sodium [57]. So, the point of novelty of Claim 1 over Claim 10 of the 486 patent was the identification of flucarbazone as being specifically a selective herbicide. Claim 10 of the 021 patent was to a method of controlling a variety of grassy weeds, including wild oats, in a cereal crop, such as wheat [57]. The point of novelty over Claim 10 of the 486 patent was the identification of flucarbazone as being a selective herbicide for the control of wild oats in crops of wheat [178].

The thrust of the obviousness attack was that in light of the disclosure in the 486 patent that flucarbazone was a herbicide, it would have been obvious to test it to see if it was a selective herbicide. The obvious-to-try argument failed on the facts. In large part this was because Aylen J found that the evidence did not disclose any reason that a skilled person would pick flucarbazone sodium out of the 327 specifically disclosed compounds for further testing as a selective herbicide: the fact that it was specifically claimed was not enough [190]–[192]. Moreover, at the relevant time, “sulfonylurea herbicides [which includes flucarbazone] were not generally known for the control of grassy weeds. Rather, they were known for controlling broadleaf weeds and demonstrating little activity on grassy weeds” [193]. Thus, the CGK would have steered the Skilled Person away from selecting flucarbazone sodium as the candidate compound for further testing [193]. This goes more directly to Claim 10 than to Claim 1, but as I understand it, the skilled person would have been interested in developing selective herbicides for commercially important grassy weeds, such as wild oats, and the fact that flucarbazone was unlikely to be good for that purpose made it unlikely that it would be selected as a candidate compound for further investigation.

This was a key determination. It might have been routine to determine if flucarbazone was a selective herbicide starting with flucarbazone itself as the candidate compound, but once it was determined that the proper field of candidates was all 327 disclosed compounds, the finding on the facts that it would have required prolonged and arduous effort to identify flucarbazone in particular as a selective herbicide, followed [203]–[209]. The same conclusion would have followed even if it could be said that the skilled person would have started with just the eleven claimed compounds [203].

It would be interesting to know specifics about the test data that underpinned the 486. I get the feeling that the drafters of the 486 patent may have been careful not to disclose the specifics of the testing, precisely in order to allow a follow-on patent to be filed on the most promising compounds. That would raise an interesting question as to whether the 486 was insufficient for not making full disclosure, and how that might be proven.

The findings relating to obviousness were fatal to the anticipation attack [224]. A broad disclosure does not anticipate a subsequent more specific claim [220], and in this case there was nothing in the prior art patents providing the requisite “clear and unmistakeable direction” that flucarbazone sodium was a selective herbicide for the control of wild oats in wheat [219], [224].

Inducement

AgraCity is a distributor of generic crop protection products, including a generic flucarbazone sodium herbicide which it sold under the name HIMALAYA. The flucarbazone sold by AgraCity was manufactured abroad by a third party and imported into Canada by AgraCity [243]. AgraCity conceded infringement and inducement [3], [83].

There was, however, an interesting issue regarding the liability of NewAgco. Pursuant to the Pest Control Products Act, companies seeking to market and sell herbicides in Canada must register the proposed herbicide with the Health Canada’s Pest Management Regulatory Agency and obtain approval of the herbicide’s proposed label, which outlines who can use the herbicide and under what circumstances it can be used [11]. NewAgco held the registration for HIMALAYA with the PMRA and its name appears on the end-use label [243]. Other than that, it appears that NewAgco has nothing to do with the manufacture, importation, or sale of HIMALAYA [240]–[246]. Aylen J held that that was not sufficient for NewAgco to be liable for inducement [246].

Remedies

Turning to remedies, Arysta sought damages only if the award would hit a specified target, defined by “displaced sales”:

[254] The Plaintiffs seek damages in the form of lost profits due to the Defendants’ sales of HIMALAYA on the basis that 90% of HIMALAYA sales displaced sales of EVEREST 3.0 AG or SIERRA 3.0 AG. In the alternative, if the Court finds that the evidence does not support a 90% capture rate, then the Plaintiffs elect a disgorgement of the Defendants’ profits on 100% of the HIMALAYA sales.

That is, Arysta sought damages, but only if they could establish on the facts that at least 90% of the sales of the infringing product by AgraCity would have been made by Arysta. This affects the quantum, as Arysta would be entitled to lost profit damages on displaced sales, but only reasonable royalty damages on other infringing sales.

Permitting Arysta to seek an accounting in the alternative in this manner appears to be contrary to AlliedSignal (1995), 61 CPR (3d) 417 (FCA) 444:

While courts of law have, for some time, given the successful party a right to elect one or the other of these two recourses, it seems clear from recent experience that the choice between the two remedies cannot be left entirely to the successful plaintiff. Moreover, it certainly cannot depend on whichever amount would turn out, on inquiry, to be more profitable. An accounting of profits is an equitable remedy which ought to be allowed by the Court in the exercise of its equitable jurisdiction when the circumstances so warrant.

The point does not appear to have been contested, so permitting Arysta to elect an accounting in the alternative is not strictly inconsistent with AlliedSignal as a matter of law. It does, however, illustrate the point that it had become routine to allow the patentee to elect an accounting, almost as a matter of right. This practice will be tested in cases involving patent assertion entities, such as Rovi Guides v BCE 2022 FC 1388, discussed here. Further, in my article on Nova v Dow 2022 SCC 43, forthcoming in the IPJ, I argue that the new approach to an accounting set out in Nova v Dow implies that the courts should take a more restrictive approach to allowing the plaintiff to elect an accounting, to ensure that it is awarded only when necessary in light of the deterrence rationale for an accounting emphasized by Rowe J.

On the facts, Arysta could not establish a 90% capture rate, and so was permitted to elect an accounting.

Skinny Label and Swiss Claims

Janssen Inc v Apotex Inc 2022 FC 996 (reasons) 2022 FC 995 (judgment) Pallotta J

            2,659,770 / macitentan / OPSUMIT / NOC

There are three known biological pathways affecting pulmonary blood pressure, namely the prostacyclin, nitric oxide, and endothelin pathways. PDE5 inhibitors, such as tadalafil, work through the nitric oxide pathway. Macitentan is an endothelin receptor antagonist (ERA), which, as the name suggests, works through the endothelin pathway. The 770 patent relates to the use of combination therapy consisting of macitentan and a PDE5 inhibitor (PDE5-I) to treat diseases involving vasoconstriction, particularly pulmonary arterial hypertension (PAH). Most PAH patients are treated with combination therapy, but a non-trivial number of patients — around 20% — are treated with macitentan monotherapy [162]. Janssen markets OPSUMIT, a 10 mg tablet, for use alone or in combination therapy. Apotex sought to market a 10mg tablet of macitentan, which is not itself patented, and in this NOC action, Pallotta J held that Apotex’s macitentan product would infringe Janssen’s 770 patent. Validity was not at issue [9]. The 770 patent was also at issue in Janssen v Sandoz 2022 FC 715, in which infringement was conceded and validity was at issue: see here.

This is in many ways a typical ‘skinny label’ case, in which a generic seeks to sell a drug that is itself unpatented, but which may be used in a manner that is patented, eg as part of a patented combination, or for a patented indication. The generic in such cases is not normally a direct infringer, so infringement by inducement must be established under the three-part Corlac test, 2011 FCA 228 [162]. This typically reduces to the question of whether the generic’s product monograph (PM) will induce infringement by prescribing physicians, who will read the monograph and thereby be induced to prescribe the generic product for use in an infringing manner. A central question is therefore whether the generic’s skinny label has been sufficiently scrubbed clean of any reference to the infringing use. This inquiry turns on the details of the PM. Sometimes the generic wins, sometimes it loses—as in this case. The generic will also sometimes argue that even if its PM does instruct an infringing use, that doesn’t constitute infringement because the prescribers don’t pay any attention to the generic PM in any event. Whether prescribers heed the generic PM is a fact-specific inquiry, but so far as I know, no generic has ever prevailed on this argument. (Please leave a comment with a case cite if I am wrong about that.) Apotex made that argument in this case, and, unsurprisingly, lost. The result was that Apotex was enjoined from marketing Apo-macitentan until after the expiry of the 770 patent. While the case was typical in broad outlines, there are a few issues of interest in this decision, particularly on the construction of Swiss-type claims.

Three types of claims were at issue in this case [86]:

Product for use

Claims 1–5 eg “A product containing [macitentan] . . . in combination with [a PDE5-I] . . . for therapeutic use. . . in the treatment of [PAH].”

Use for treatment — sometimes referred to as German-style claims

Claims 21–31 eg “A use of [macitentan] in combination with [a PDE5-I] for treating [PAH].”

Use in manufacture of a medicament for — commonly called Swiss-form claims

Claims 10–20 eg “A use of [macitentan] in combination with [a PDE5-I] . . . for the manufacture of a medicament intended to treat [PAH].

A variety of claim types are employed because many jurisdictions have some kind of prohibition on patenting of methods of medical treatment, and consequently a variety of claim forms are used in claiming the use of a compound to treat a disease. A fourth type of claim that is often used in the US is straightforward: “A method of treating disease Y, comprising administering compound X.” This type of claim is not allowed in Canada as it is considered to fall afoul of the prohibition on patenting methods of medical treatment.

A preliminary issue was as to the construction of “combination,” which Pallotta J construed as including any scenario in which macitentan and the PDE5-I work in concert to treat the disease in question. In particular, it does not matter whether the macitentan and PDE5-I are combined in the same dosage form [120]; and it includes a scenario in which the patient is started on macitentan monotherapy and then moved to a combination therapy of macitentan and PDE5-I [116].

The most interesting claim construction issue concerned the interpretation of the Swiss-type claims. Swiss claims were originally developed to avoid the European prohibition on patenting of methods of medical treatment. Since a Swiss claim is, on its face, a claim to the product, not its use, and the physician who prescribes a drug does not manufacture it, physicians are excluded from the scope of infringement. However, in Hoffmann-La Roche v Sandoz 2021 FC 384, [95]–[109], Manson J held that the Swiss claims should be construed as use claims on the basis of a purposive interpretation; in effect, the idea is that what was really discovered was a new use, not a new method of manufacture, and the claim should be construed in light of what was actually discovered [97]. The result was that the generic, Sandoz, could not be liable as a direct infringer. As discussed here, this is a bit of a strange result, because it means that prescribers are direct infringers and the manufacturer is not, even though the original purpose of Swiss claims was to ensure that the manufacturer a direct infringer and the prescriber is not an infringer at all.

In this case, Apotex, relying on Hoffmann-La Roche, argued that the Swiss claims should be construed as use claims, so that Apotex could only be liable, if at all, on the basis of inducement. Pallotta J did not accept this argument. While a purposive construction may consider the nature of the inventive concept, “a purposive construction focuses on the language of the claims,” and “[t]he words chosen by the patentee necessarily play a key role” [128]. The importance of the text is consistent with the general law of statutory interpretation: Canada Trustco 2005 SCC 54 [10]; Canada v Utah 2020 FCA 224 [9]; Biolyse v Bristol-Myers Squibb 2003 FCA 180 [13]. It is also well-established as a principle of claim construction: Free World 2000 SCC 66 [39]–[40]; ABB Technology 2015 FCA 181 [42]–[43] (cited by Pallotta J at [128]. The words of the Swiss claims clearly claim the use “for manufacture of a medicament” [128].

While I’ve waffled a bit on this (see my comments on Warner-Lambert) I think Pallotta J is right on this. The key doctrinal point is that the words play a “dominant role” in the interpretive process, as the SCC put it in Canada Trustco. This is important in the context of statutory interpretation because the legislature must be able to rely on the courts to carry out the legislature’s intent, and the words, when clear, are the best reflection of that intent. The parallel point in the context of claim construction is that the patentee should be able to decide what it wants to claim—taking the risk that the claim may be invalid—and again, the words it uses are the best guide to that intent. If there is some public policy reason that Swiss form claims should not be permitted at all, then this should be implemented through a substantive legal rule, which can then be debated and subject to appeal.

Even though Pallotta J came to a different conclusion than Manson J, she did not have to expressly disagree with his analysis, as Manson J held that the meaning was context specific [103] and his holding of law was only that Swiss claims do not “automatically benefit[] from a literal construction” [102]. I must say I am not very enthused about that approach. Patent agents take great pains with the words used in a claim. While some terms used will be specific to the invention, and so may require considerable context, there are some terms, such as “comprising,” which are effectively terms of art with a specific meaning, and which should accordingly be given a consistent meaning. To my mind, Swiss claims are sufficiently common and established as a claim form, that they should be given a standard construction, so that patent agents will be able to predict the consequences of using Swiss claims.

Even though Palotta J construed the Swiss claims as product claims, she nonetheless held that neither the Swiss claims, nor the product-for-use claims, were directly infringed by Apotex. She noted that Apotex only intended to sell macitentan itself, not a product that contains macitentan in combination with a PDE5-I. Consequently, she held that Apotex did not infringe the product-for-use claims, which claim ‘A product containing [macitentan] in combination with [PDE5-I]’ because the product sold by Apotex would not contain the drugs in combination [144]. This reasoning seems right to me. She held that the Swiss claims were not infringed for the same reason [144]. This also seems right, given the structure of the Swiss claims at issue: “A use of [macitentan] in combination with [a PDE5-I] . . . for the manufacture of a medicament intended to treat [PAH].” This indicates that macitentan and the PDE5-I are used in combination for manufacture of the medicament.

As an aside, it occurs to me that a more difficult question would arise if the Swiss claims were of the form “A use of [macitentan] for the manufacture of a medicament intended to treat [PAH] in combination with [a PDE5-I].” In that claim, the question is whether the macitentan is intended for combination use, which, given Pallotta J’s holding that “in combination” encompasses any scenario in which the drugs work in concert, would include administration of separate pills in combination therapy. In that hypothetical, the question would be whether the generic product was intended for use in combination therapy. As discussed here, in Warner-Lambert v Generics (UK) [2018] UKSC 56 the UKSC split three ways in trying to decide what “for” means, with two variations on an objective intent test, plus a subjective intent test, not to mention the modified objective foreseeability test endorsed by the EWCA. Using the hypothetical claim, the generic might be a direct infringer on either the objective or subjective intent approach, depending on whether, on the facts, it subjectively or objectively intended macitentan to be administered in combination with a PDE5-I. I would stress that I am not endorsing either of these approaches, but simply pointing out that a more difficult issue would have arisen had the Swiss claims been framed differently.

With construction settled, Apotex argued that its PM did not instruct the use of macitentan in combination therapy. Pallotta J rejected this argument on the facts. It’s a bit hard to follow because of redactions, which, though brief, obscure the precise nature of the references. If I understand correctly, the main trial which established the safety and efficacy of macitentan as a monotherapy – “SERAPHIN” — also established its efficacy as a combination therapy [153]. The Apotex PM contained clinical trial data from SERAPHIN [192], [193], and SERAPHIN was so well known [187] that the references to it in the Apotex PM would be enough to alert physicians to the fact that macitentan is effective in combination therapy, and so would induce infringement [192], [199]. This is even though the APO-MACITENTAN PM “removed all mentions to the use of macitentan as a combination therapy that are present in the OPSUMIT PM, and only mentions SERAPHIN results that reported macitentan was useful as a monotherapy” [164]. It is true that the courts have been stringent about ensuring that the generic PM is scrubbed clean of references to the infringing use, but this seems pretty harsh on Apotex, given that it had to include the clinical trial data to establish efficacy as a monotherapy. I must say that I suspect that one factor in Pallotta J’s analysis is the fact that combination therapy is the primary use for macitentan, with only 10–30% of patients getting monotherapy [162]. If the opposite were true, so that eg only 10% of the use was in the patented combination, then I wonder if it would have gone the other way. Of course, in that scenario, it is perhaps unlikely that the main clinical trial data would be for the combination therapy.

Apotex also argued that even if the PM did instruct infringing combination therapy, this did not matter because prescribers don’t pay attention to the PM anyway: “In the field of PAH, the physicians know SERAPHIN very well and it does not matter what is contained in APO-MACITENTAN PM” [166]. This argument failed on the facts [197], as it always does. (If anyone knows a case where this argument has succeeded, please let me know in the comments.) The argument makes sense in theory, given the requirement that the influence be the “but for” cause of direct infringement, but if successful, it would mean that generic would in theory be able to sell a drug with the infringing use plastered all over the PM, on the basis that the PM is irrelevant anyway. I suspect that a desire to avoid this strongly counter-intuitive result might help explain why the courts invariably come to the conclusion that at least some prescribers are influenced by the PM. (Of course, it’s also possible that it’s simply true that at least some prescribers are influenced by the PM.)

Finally, Pallotta J emphasized that in light of this influence, Apotex would induce infringement “even if a physician applies their own skill and judgment to the decision to prescribe combination therapy” [156]: see similarly [181]. Pallotta J remarked that “[i]f it were otherwise, inducing infringement could never be found in the context of pharmaceutical patents” [156]. In any event, I don’t really see any difficulty with this position. As Pallotta J also noted, “a ‘but for’ test does not mean that Apotex’s activities must be the sole cause of the infringement” [156]. A physician who learns from the PM that macitentan is useful in combination therapy to treat PAH must still use their skill and judgment to determine that their patient is suffering from PAH. Conversely, a physician who has determined that their patient is suffering from PAH will not prescribe macitentan combination therapy unless they have learned somewhere that it is effective in treating PAH; and if they learn that from the PM, the information in the PM is the ‘but for’ cause of the infringement, even if it is not the sole cause of the infringement.