Tension in the Sanofi Disclosure Analysis

Takeda Canada Inc v Apotex Inc 2024 FC 106 Furlanetto J

2,570,916 / dexlansoprazole / DEXILANT / NOC

Proton pump inhibitors (PPIs) are a class of compounds that decrease gastric acid level in the stomach [5]. PPIs are now a well-established treatment for GI disorders related to increased gastric acid [7]. It is desirable to have a once daily dosing regimen, but the formulations known at the time of the patent could result in nocturnal breakthrough events if used on a once-daily basis, resulting increased gastric acid production [12]. Takeda’s 916 patent aimed to address this problem with a “pulsatile” release formulation, comprising a PPI with “a first and a second dose,” which are released from the dosage form as “discrete pulses,” resulting in specified blood plasma concentrations [90], [92]. Takeda’s DEXILANT product is a pulsatile release formulation of the claimed type, which includes two types of delayed-release beads containing dexlansoprazole [6]. Apotex sought to sell a dexlansoprazole oral dose capsule product and Takeda brought this NOC action in response [2].

Furlanetto J addressed three claim construction issues, which all turned on the evidence. One question in particular turned on the meaning of “a first and a second dose” in the claims. Apotex argued that this means distinct amounts of PPI in the dosage form itself [97], as with the two types of delayed-release beads in DEXILANT. Takeda argued that the reference to two doses was not the amount of PPI in the formulation, but the amount of PPI released from the formulation [97]. In other words Takeda argued that even if there was only one form of PPI in the product, if it was released in two waves resulting in two distinct blood plasma peaks, that would fall within the claims. Furlanetto J accepted Apotex’s construction, for reasons that strike me as being straightforward and compelling [98]–[104]. (Her conclusions on the other two claim construction issues also strike me as being sound.)

In light of that construction, Apotex prevailed on infringement because it does not contain two different types of dexlansoprazole: “Characteristic of a single continuous, delayed release dosage form, each of the mini-tablets in the Apotex Product are identical” [119]. There was also an extended discussion of various labs tests which was a bit difficult to follow because of redactions. My impression is that Takeda was arguing that if you squint really hard, you can see two peaks in the plasma concentration curves. In any event, in the end Furlanetto J held that Apotex’s product would not infringe as Takeda failed to established that Apotex’s product comprised a first and a second dose of dexlansoprazole from the dosage form as two discrete pulses: “Rather, the evidence indicates that there is only one dose of dexlansoprazole in each 30 mg or 60 mg capsule of the Apotex Product that is released from the Apotex Product in a single, continuous delayed release fashion” [147].

To this point, the decision turned on the facts with no novel points of law. The conclusion on claim construction and infringement would have been sufficient to dismiss the action [148]. Furlanetto J nonetheless went on to address Apotex’s validity arguments, “which formed a significant portion of the parties’ arguments at trial” [148].

Anticipation

The anticipation argument raised a tricky point which turns ultimately on a tension in Sanofi 2008 SCC 61.

The focus was solely on the disclosure branch of the two part test for anticipation [151]. The key piece of prior art was Application 2,499,574, and in particular Example 57 which “provides a formulation that is close to DEXILANT® except that it includes the excipient PEG as part of the formulation” [168]. Apotex introduced evidence that the PEG would not affect the pharmacokinetics and that the blood plasma reached would exceed the threshold specified in the claims [169]–[170]. Furlanetto J held that Example 57 did not anticipate because “the heart of the invention of the 916 Patent is the recognition of the threshold plasma concentration and how to achieve and maintain it,” and “There was no suggestion in the 574 Application of any recognition of these properties, nor of their significance” [176]. Moreover, Example 57 did not disclose the blood plasma concentrations that would result, and for that to be determined, the capsules of example 57 would have had to be administered to humans and the PK properties would have to be identified and studied” [176]. It is not entirely clear to me if Example 57 comprised two different types of granules (see [202], suggesting it did), but it is clear that Example 11 did have two types [161], and Furlanetto J held that Example 11 did not anticipate because, inter alia, “there was no disclosure of any plasma concentration being a threshold which must be surpassed for pharmacological effect” [166].

So, Furlanetto J is saying that these examples did not anticipate because they did not disclose the properties of the compound described in the examples. This is contrary to the basic rule that “what infringes if later, anticipates if earlier”: Consolboard [1981] 1 SCR 504, 534. There is no requirement that anyone at the time would know of all the properties. As Lord Hoffmann said in Synthon [2005] UKHL 59 [22], in a passage quoted with approval in Sanofi 2008 SCC 61 [25], “whether or not it would be apparent to anyone at the time, whenever subject matter described in the prior disclosure is capable of being performed and is such that, if performed, it must result in the patent being infringed, the disclosure condition is satisfied.” So, if a pharmaceutical company were to market a pill that in fact had two types of PPI and resulted in plasma concentrations that were in fact above the specified thresholds, that pill would infringe if sold after the patent issued, regardless of whether anyone had measured the plasma concentrations prior to litigation. By the same token, such a pill would anticipate if sold prior to the date of the patent, also regardless of whether anyone had ever measured the plasma levels. As Lord Hoffmann explained in Merrell Dow [1995] UKHL 14 [47], “whether or not a person is working a product invention is an objective fact independent of what he knows or thinks about what he is doing”; similarly in Synthon [2005] UKHL 59 [22], quoting Merrell Dow, Lord Hoffmann noted that “[t]he flag has been planted, even though the author or maker of the prior art was not aware that he was doing so.”

All of this is settled law, with Sanofi as the leading case, and indeed, it was expressly recognized as such by Furlanetto J [153]. Why then did she hold that the plasma concentrations had to be disclosed in order to anticipate? The problem lies in Sanofi itself. The quotes from Sanofi that I have given above are all from the discussion of general principles relating to anticipation. When the SCC in Sanofi discussed the disclosure requirement in particular, and in applying those principles to the facts, the SCC did indeed imply that the properties have to be disclosed in order to anticipate.

Recall that the question in Sanofi was whether the prior art disclosure of the racemate anticipated the claim to the enantiomer. The answer was no. But why? Under the general principles set out in Sanofi, the answer would have been that making the racemate would not infringe a claim to the enantiomer, and so the test that “the matter relied upon as prior art must disclose subject-matter which, if performed, would necessarily result in an infringement of the patent” (Sanofi [25], quoting Synthon [2005] UKHL 59 [22]) would not be satisfied. More broadly, the question in Sanofi was framed as being whether disclosure of the genus can anticipate the species. The answer is no, because someone practising the genus would not inevitably infringe, as the genus might be practised without infringing, by selecting another species from the pool: see Sanofi [21].

But while that answer would have been straightforward on the general principles set out in Sanofi, the SCC changed its approach when it came to the discussion of the disclosure requirement specifically. The SCC said:

[31] Section 27(1) of the Act requires as a condition for obtaining a patent that the invention was not “known or used” and was not “described” in any patent or any publication more than two years before the patent application was filed. In the context of genus and selection patents, in E. I. Du Pont de Nemours & Co. (Witsiepe’s) Application, [1982] F.S.R. 303 (H.L.), Lord Wilberforce stated, at p. 311:

It is the absence of the discovery of the special advantages, as well as the fact of non-making, that makes it possible for such persons to make an invention related to a member of the class.

The compound made for the selection patent was only soundly predicted at the time of the genus patent. It was not made and its special advantages were not known. It is for those reasons that a patent should not be denied to the inventor who made and discovered the special advantages of the selection compound for the first time.

This is all true, but it runs together anticipation and obviousness.

The general point of Witsiepe’s Application is that a species may be patented over the genus. It is now understood that two separate hurdles must be overcome. The prior art disclosing the genus must not have specifically disclosed the species, or it will be anticipated. This is true even if it turns out that the species had special properties that were not appreciated in the prior art. Suppose that the genus patent specifically disclosed five especially preferred compounds by structure, and disclosed in vitro tests showing they were all effective in treating a disease. One of those, the lead compound, was commercialized and sold. It would not be possible to get a subsequent species patent on that compound, even if it was subsequently discovered, surprisingly and unexpectedly, that it had no side effects at all, while every other known member of the genus had significant side effects. That would be true even if no one had known that this was the only compound with no side effects (perhaps because the lead compound was the only one which was commercialized and the others had never been tested in humans). The same is true if the especially preferred compound had not been commercialized, but merely specifically disclosed, as there is no difference in principle between prior sale and prior publication. (If it were surprisingly and unexpectedly discovered that the compound was useful in treating an entirely different disease, it would be possible to get a patent on the use to treat that disease. That is not because the prior art must disclose the properties to anticipate; it is because making and using the compound for uses known in the prior art would not infringe the new use patent.) On the other hand, even if the species was not specifically disclosed, it must also have special properties. There is no “invention” in picking one member of the genus at random and finding that it has exactly the same properties as all the other known members. If it does not have special properties, it will be obvious, even if it is not specifically disclosed.

The quoted passage from Witsiepe’s Application’s runs these together: “It is the absence of the discovery of the special advantages [non-obviousness], as well as the fact of non-making [anticipation], that makes it possible for such persons to make an invention [new, useful and non-obvious subject matter] related to a member of the class.” And the following sentence from the SCC is to the same effect: “It was not made [not anticipated] and its special advantages were not known [not obvious].” It is true that both of these requirements must be satisfied for a selection patent to be valid. The problem is that in the heading to that section — “i. Disclosure” — and in the following paragraph, the SCC in Sanofi identified both of these separate requirements with the disclosure branch of anticipation:

[32] Where there is no such disclosure [of the special advantages], there is no discovery of the special advantages of the selection patent as compared to the genus patent, and the disclosure requirement to prove anticipation fails.

The SCC also ran obviousness and anticipation together in dealing with the disclosure requirement when applying the law to the facts. The special properties of the enantiomer were better activity and lower toxicity. The SCC said there was no anticipation because “[t]here was no evidence that the person skilled in the art would know from reading the [prior art genus patent] that the more active dextro-rotatory isomer would be less toxic than the racemate or levo-rotatory isomer or any of the other compounds made and tested” [40] and “[s]ince the ‘875 patent did not disclose the special advantages of the dextro-rotatory isomer and of its bisulfate salt, as compared to the levo-rotatory isomer or the racemate and their salts, or the other compounds made and tested or otherwise referred to in the ‘875 patent, the invention of the ‘777 patent cannot be said to have been disclosed” [41]. The approach set out on the facts in Sanofi was picked up in comments in Novo Nordisk 2010 FC 746 [170]–[174] and in Allergan 2022 FC 260, which were relied on by Takeda and quoted by Furlanetto J at [174], [175].

So, the reason that Furlanetto J’s analysis of disclosure in this case was inconsistent with the general test set out in Sanofi, is that the SCC’s analysis of disclosure in Sanofi was also inconsistent with the general test that it had just set out. The reason for this internal inconsistency in Sanofi is that it was a selection patent case. When the SCC addressed the broad principles of anticipation, it relied on the leading cases on anticipation generally. But when it got to the specifics, it turned to cases on selection patents specifically, which run anticipation and obviousness together.

I can’t fault Furlanetto J for following the SCC’s lead. But this case illustrates the problem with the view that the special properties must be disclosed in order to anticipate. The focus on disclosure of the properties didn’t make any difference in Sanofi itself because the enantiomers were not specifically described in the prior art in any event. But in this case, the focus on the properties may have made the difference. Because Furlanetto J held that the claims were not anticipated because the plasma concentrations were not disclosed, it is not entirely clear to me whether she would have held them to be anticipated on the “what infringes if later, anticipates if earlier” test. So suppose that the prior art did indeed disclose a formulation that would indeed infringe, so that the claims would be anticipated on that test. In particular, suppose that even though the plasma levels were not disclosed, the formulation disclosed in the prior art would in fact result in clinically effective plasma levels, exceeding the threshold set out in the claims. And since there is no difference between a prior publication and prior practice, suppose that the prior patentee had marketed its product to thousands of patients, who had taken it to successfully treat their GI disorders, all without measuring the plasma levels. Would the 916 patent be valid, so that the prior manufacturer would have to withdraw its product and turn the market over to Takeda? That can’t be.

When can a range anticipate a point?

AbbVie Corporation v JAMP Pharma Corporation 2023 FC 1520 McVeigh J

2,504,868 / 2,801,917 / 2,904,458 / adalimumab / HUMIRA / SIMLANDI

The most important aspect of this case was McVeigh J’s refusal to grant a permanent injunction to the successful patentee. I will discuss that in a separate post. McVeigh J also had a brief discussion of double patenting, which I discussed in a previous post that focused on NCS v Kobold 2023 FC 1486. This post provides an overview and addresses some miscellaneous issues, including McVeigh J’s brief but helpful discussion of as to when prior art that discloses a range can anticipate a point within a range.

This case involved three patents related to adalimumab, which is well known under the brand name HUMIRA for treatment of rheumatoid arthritis, inflammatory bowel disease (IBD) and psoriasis. The 868 patent relates to a dosing regimen for a known use (the treatment of IBD). McVeigh held the 868 patent to be invalid as being obvious to try [303], though an attack based on patentable subject matter failed [339]. The 917 patent relates to a second medical use — the treatment of hidradenitis suppurativa (HS) — and a dosing regimen. It seems that the use of adalimumab to treat HS was known [400], so the key issue, as with the 868 patent, was whether the dosing regimen was obvious. McVeigh J held on the facts that it was [402]. An attack based on patentable subject matter failed [413]–[415]. There was also an unusual attack based on “Claim term not described” which I will come back to. The 458 patent related to an adalimumab formulation, which McVeigh J found on the facts to be not anticipated [559] and not obvious [591]. Attacks based on overbreadth [592]–[603] and double patenting [604]–[620], also failed.

Identical Witness Statements

As discussed here, in dTechs 2023 FCA 115 [32], [34], the FCA addressed the role of counsel in the preparation of expert reports. Gauthier JA noted that while it is not unusual for expert reports to be prepared “in close collaboration with counsel,” this is not normally a significant problem, so long as the Court is ultimately “presented with the substantive and objective opinion of the expert.” Gauther JA also noted that “While counsel may make mistakes and overstep the bounds of what is permissible involvement, this will normally be revealed on cross-examination at trial, and will be considered by trial courts in assessing the evidence” [34]. Something along those lines happened here, albeit with a fact witness:

[210] Dr. Noertersheuser suffered significant credibility issues. It became apparent during his cross-examination that much of his statement was nearly identical, or word-for-word the same, as a witness statement made by Dr. George Richard Granneman on December 18, 2017 in US litigation. While this is not, in and of itself, hugely problematic, his responses on cross-examination reduced his credibility. He attempted to explain that perhaps he and Dr. Granneman had written identical statements because of how closely they had worked together in the past. He said they worked side by side and discussed things back then, and that must have been the reason they wrote identical statements. He was asked several times and confirmed that he independently came up with the exact same words but that he never copied. This is clearly inaccurate, and he would have better served the Court by admitting the statements were copied, instead of trying to claim that it was sheer coincidence.

[212] Unlike in Rovi, Dr. Noertersheuser was not an expert witness. Nonetheless, the word-for-word copying of Dr. Granneman’s report raised significant concerns about Dr. Noertersheuser’s impartiality. His inability to accept or acknowledge the paragraphs were word-for-word the same greatly impugned his credibility and reliability.

[215] The similarity between Dr. Noertersheuser (fact witness) and Dr. Granneman’s expert report went well beyond the appropriate limits. I will assess Dr. Noertersheuser’s evidence with some caution when it is in disagreement with other witnesses.

Treating his evidence with “some caution” is a very measured response.

Blinding the Witness

Over the past decade there have been various attempts to enhance the credibility of an expert on issues such as claim construction or obviousness by not informing the expert of the party’s position when they write their report. While some earlier decisions supported this practice, the trend more recently has been to be skeptical, generally on the view that the substance of the report is more important than whether the witness was blinded. Another concern is that it may impossible to truly blind a person who is an expert keeping abreast of developments in a particular field: see eg Hospira 2018 FC 259 [203]. That problem arose in a particularly acute form in this case, when one of the experts who had been ‘blinded’, forgot that she had previously provided expert evidence in another matter involving the 868 patent [247]–[249], [252]. McVeigh J expressed skepticism of the worth of the blinding process, saying “I rely on Dr. Baughman’s evidence, not for the fact that it was blinded but for the reasoning it provides” [250]. McVeigh J did not fault Dr. Baughman, whom McVeigh J found to have sincerely failed to recall her prior involvement”; nonetheless, while she did accept the evidence as credible and reliable, she only assigned a moderate weight to it [252].

When can a range anticipate a point?

The parties disagreed “as to whether prior art that discloses a range can anticipate a point within a range or embodiment. For example, has disclosure occurred if the range is 0.2-400 mg in the prior art, and the subsequent patent claims about 10 mg or a narrower range such as 0.10-100 mg” [157]. JAMP relied on several recent cases in which “the Federal Court held a range or a broad disclosure can anticipate a point within a range or an embodiment” [158]. After reviewing those cases, and discussing several blog posts in which I have argued that these decisions were clearly wrong, McVeigh J concluded as follows:

[173] I find that the law has developed that, where the evidence presented at trial shows that the range is narrow enough, such that a flag can be planted based on the context and examples given, then it is anticipated. If the evidence shows a very broad range that the Judge, with the experts’ assistance (if needed), does not see it as a precise enough to plant the flag before proceeding to the enablement stage, then it fails at the disclosure stage.

This statement of the law strikes me as sound and helpful (though I’m not sure it actually explains the prior cases or whether McVeigh J intended it as doing so). It is consistent with the EPC Examination Guidelines G.VI.8, which also strike me as sound:

A sub-range selected from a broader numerical range of the prior art is considered novel if both of the following two criteria are satisfied (see T 261/15):

(a) the selected sub-range is narrow compared to the known range;

(b) the selected sub-range is sufficiently far removed from any specific examples disclosed in the prior art.

The meaning of "narrow" and "sufficiently far removed" has to be decided on a case by case basis.

McVeigh J also applied her analysis to the facts in two instances, both holding that there was no anticipation.

JAMP argued that the asserted claims of the 917 Patent were anticipated by the 868 application [383]. Recall that the 917 patent relates to the use of adalimumab, a TNFα inhibitor, to treat HS according to a specified dosing regimen. The 868 application disclosed the use of TNFα inhibitors as a means of treating 16 broad categories of disorders, including HS, and presenting a large class of dosing elements with numerous choices for the specific TNFα inhibitor, the number of loading doses, the amount for each loading dose, the interval between loading doses, the interval between loading and maintenance dose (one hour, one day, one week, two weeks) and the amount of maintenance dose [387]. So, “a skilled person reading the 868 Application would have a range of dose amounts, dosing intervals, and durations of treatment to choose from when creating a multiple variable dosing regimen to treat an inflammatory-related disorder” [388]. In contrast, the 917 Patent claimed a much more specific dosing regime [389]. McVeigh J invoked her para 173 analysis [385] and concluded:

[390] Given the number of dosing elements disclosed in the 868 Application, it is not clear that a POSITA would know to select adalimumab and to administer it using the specific dosing regimen claimed in the 917 Patent. If the POSITA reading the prior art reference must adopt a specific way forward in order to infringe, yet there are numerous other ways to perform the prior art without necessarily infringing, then there is no disclosure: Shire [2021 FCA 52] at para 50.

This strikes me as straightforwardly correct, and consistent both binding jurisprudence, such as Shire and with McVeigh J’s summary statement at [173].

JAMP also argued that the asserted claims of the 458 Patent, claiming a formulation of adalimumab, were anticipated by the 181 Application, titled “Self-Buffering Protein Formulations” [539]. The 181 Application listed adalimumab among the proteins:

[545] The most preferred protein concentration range in the 181 Application is 20-150 mg/mL. This is a very large range. For the four proteins tested whose test results appeared in the 181 Application, there were concentrations of 90 mg/mL and 110 mg/mL which proved to be self-buffering within pH ranges which included 5.2. But there were other ranges tested too, and none of these proteins were adalimumab. Moreover, in the claims section of the 181 Application, these proteins are not introduced in the same claim as adalimumab and there is no indication of dependency between the claims that introduce these proteins and the claim which refers to adalimumab.

[546] As mentioned above, a range does not always anticipate a point, but it could if dependant on the facts of the particular patent. This is even truer where the range is very large or broad. Accordingly, I do not see how a range of protein concentration from 20-150 mg/mL can plant a flag at the specific concentration of 100 mg/mL, especially where specific examples of formulations given do not include the relevant protein.

Accordingly, she held the 181 application did not anticipate. Again, this strikes me as straightforwardly correct, and consistent both the binding jurisprudence and with McViegh J’s summary statement at [173].

Methods of medical treatment

It is still black letter law that methods of medical treatment are not patentable subject matter, though the jurisprudential basis for that proposition is shaky: see Hospira 2020 FCA 30 [48]–[49] (discussed here). However, no one really knows what a “method of medical treatment” is. That’s because there is no clear principle underpinning the rule. The rule was originally set out by the SCC in Tennessee Eastman [1974] SCR 111, primarily on the very logical basis that allowing patentability of methods of medical treatment would allow an end-run around what was then s 41, which restricted patents for medicine to product-by-process claims. Section 41 has since been repealed, but Tennessee Eastman had a tantalizing dictum — “I do not think so” — suggesting there might be more to it than that. Subsequent decisions of the SCC suggested that the bar might turn on the “essentially non-economic” nature of the methods of medical treatment: see Hospira 2020 FCA 30 [49], discussing Shell Oil [1982] 2 SCR 536, 554 and Wellcome / AZT 2002 SCC 77 [49]. But no one really knows what it means to say that professional skills are essentially non-economic. (Try telling that to a lawyer.)

In any event, the best current theory is that a claim that encompasses the skill of a medical professional is unpatentable. To say it’s the best theory is not to say it’s a correct theory, or even a good theory — see Hospira [51]–[52] — but that’s more or less where we are now. The problem is that no one really knows what it means for a claim to encompass the skill of a medical professional. As a result, we are all flailing in trying to give some kind of content to the putative rule that methods of medical treatment are unpatentable.

In this case, JAMP argued that the claims of the 868 and 917 patents, both directed to dosage regimens, were invalid as being directed to a method of medical treatment that requires the exercise of professional skill or judgment [332] saying that:

[333] where there is evidence that the dose would be changed over time in response to a patient’s needs, then that regimen is vulnerable to an attack on the basis that is an unpatentable method of medical treatment. In JAMP’s view, the dosing regimen must be appropriate for all.

McVeigh J rejected this argument, quoting the decision of Manson J in Janssen 2022 FC 1218 [171] (discussed here) to the effect that if a physician departs from the claimed dosage regimen “they would no longer be practicing the claimed invention” [338]. She followed this reasoning, concluding that “to the extent a minority of physicians wish to deviate from the claimed regimens at some point, this does not render the claims unpatentable; rather, it takes their conduct outside the scope of the patents like in Janssen” [339]. See similarly [413]–[415] respecting the 917 patent.

More broadly, this suggests that the incoherent state of the law has made the Federal Court wary of striking down patents on the basis that they claim unpatentable methods of medical treatment. This wariness strikes me as being entirely appropriate. This bar on patentability has no basis in the Act. While the courts can and do read text into a statute, this is normally done only after a careful purposive interpretation shows that adhering to the text would lead to an absurd or otherwise unacceptable result. The courts are right to be reluctant to use a judge-made doctrine to strike down a patent for an invention which is otherwise new, useful and non-obvious, without some clear principled basis for doing so.

Doses not subject to discretionary adjustment

The 917 patent claimed a dosage regimen “wherein said multiple doses are not subject to any discretionary adjustment by a physician or medical practitioner” [380]. This phrase was added by AbbVie during prosecution to overcome an examiner objection that the claim was to a method of medical treatment because it limits the professional skill or judgment of a physician [441]. The issue came up as part of an added matter attack, which I will return to below. For now, I’ll just point out McVeigh J’s remark that “AbbVie’s addition did not change claim 1, nor does it broaden the claim. Claim 1 would have had the same effect, with or without the additional statement” [443]. This is a relief. I don’t want to get too far into the weeds of how this phrase might be construed. (Surely any dosage regimen is subject to discretionary adjustment in some cases, eg if the patient has an unexpected life-threatening reaction to the first dose—would that mean all claims with that phrase would be invalid?) The larger point is that McVeigh J’s statement suggests that the issue of patentability of methods of medical treatment will be decided on substantive grounds, not by trying to make the issue disappear with drafting slight of hand. When the law is settled, but unworkably, it may be necessary to allow the use of drafting methods to get around the black letter law, as was done with Swiss form claims. But that is a second best solution. It is better to get the substantive law right in the first place, and the law on patentability of methods of medical treatment is currently sufficiently unsettled to allow us to try to do that.

Added matter

JAMP attacked the 917 patent for failure to comply with s 38.2(2): see [192]–[197], reviewing the law, and [416]–[444], further reviewing the law and applying it to the facts. This attack was referred to as a “Claim Term Not Described,” but I’ll call it an “added matter” attack, as is this the conventional term for the parallel issue in UK law. The issue arises regularly in UK law, because claim amendments are permitted in litigation, and the question is whether the amended claim impermissibly adds new matter. There is limited Canadian caselaw on this provision, with the main authority being Western Oilfield 2021 FCA 24 (discussed here).

38.2(2) provides as follows:

The specification and drawings contained in an application, other than a divisional application, may not be amended to add matter that cannot reasonably be inferred from the specification or drawings contained in the application on its filing date.

One issue that was raised was whether this provision can be the basis for an invalidity attack, or whether it is purely procedural.

After reviewing the authorities, including Western Oilfield, McVeigh J concluded that “the law currently stands, it is unclear whether a patent is invalid when a party adds a claim term that cannot be reasonably inferred from the specification or drawings” [197]. Western Oilfield is perhaps a bit ambiguous. The main issue was the nature of the test for whether the new matter could be inferred, and specifically whether the Canadian courts should follow the “strict” UK test. One reason for wariness given by the FCA was that “the U.K. provision provides explicitly for patent revocation. Section 38.2 does not” [143]. McVeigh J took this as a hint that perhaps invalidity is not a remedy for failure to comply with s 38.2(2), as I did I in my post on that aspect of Western Oilfield, where I said that Locke JA’s statement “raises the question of whether added matter contrary to s 38.2 is a basis for invalidating a claim in a granted patent, or whether it is only ground for refusing an amendment during prosecution.”

On re-reading, I don’t think that is what Locke JA was suggesting. As noted, the discussion was in the context of whether the strict UK test for inferability should be followed. The full paragraph was:

[143] The third reason to be wary of the strict U.K. test is that the U.K. provision provides explicitly for patent revocation. Section 38.2 does not. On the contrary, subsection 38.2(1) provides that, subject to certain limitations, a patent application may be amended. The provision of particular interest in the present appeal, subsection 38.2(2), provides for one of the contemplated limitations.

I think Locke JA’s point is that because an amendment is explicitly permitted, we should be wary of being too restrictive in permitting such amendments. That does not imply that invalidity is not the remedy when the amendment is not permitted, even under a more relaxed test. The FCA in Western Oilfield went on to hold that the amendment was reasonably inferable [147], so the Court did not explicitly address the remedy for failure to comply with this provision.

I’d also note that s 72 of the UK Act exhaustively states the grounds for invalidity of a patent, including that the invention is not patentable (72(1)(a)), which includes anticipation or obviousness, as well as added matter (72(1)(d)). The Canadian Act has no equivalent. Section 60 simply provides that patent may be declared invalid or void, but without listing the specific grounds, and s 59 is the same. And s 28.2, for example, dealing with novelty, only says that subject matter “must not” have been disclosed, but does not say explicitly that a claim may be invalidated on that basis. So, given that there is no provision in the Canadian Act expressly stating that a patent may be held invalid for anticipation, obviousness, inutility or insufficiency, the fact that there is no provision expressly providing that a patent may be held invalid on the basis of 38.2 does mean much. This is another reason for thinking that this comment by Locke JA was not intended to question whether invalidity was the appropriate remedy for failure to comply with 38.2(2).

In any event, while McVeigh J did not consider the matter settled by Western Oilfield, she went on to hold that when an amendment is made in violation of 38.2(2), a claim is indeed invalid [434]. This seems to me to be clearly correct. Suppose I invent the wheel and file an application fully disclosing and claiming the wheel. Then, after I filed, but before my patent is granted, someone else publicly discloses how to make and use mRNA vaccines. I then amend my application to disclose and claim how to make and use mRNA vaccines. My claim can’t possibly be valid, even if it somehow slips through the patent office. There isn’t any other provision that would clearly prohibit this kind of amendment, since the mRNA vaccine is new and non-obvious as of my filing date. Section 38.2(2) provides a straightforward basis for invalidating such a patent.

The issue arose on the facts because of the addition of the phrase “wherein said multiple doses are not subject to any discretionary adjustment by a physician or medical practitioner” to the 917 patent during prosecution, as discussed above. McVeigh J held that amendment did not result in invalidity:

[443] In my view, AbbVie’s addition did not change claim 1, nor does it broaden the claim. Claim 1 would have had the same effect, with or without the additional statement.

[444] However, that is not the issue. The issue is whether the added claim term was reasonably inferable from the 917 Patent disclosure. I find that though it does conflict with the disclosure, AbbVie has not gained anything more than it originally had.

It’s not clear to me what McVeigh J meant when she said the added term “does conflict with the disclosure” but the main point seems to be that since the amendment did not change the claim scope, nothing at all was added, so it follows that nothing that was not reasonably inferable was added.

Claim construction

The only real dispute between the parties was the definition of “treating” in Claim 1

[296] I agree and adopt JAMP’s experts’ construction, as, even given its ordinary meaning, treating does not always achieve a meaningful result. Treating means attaining some therapeutic results but does not demand a particular duration or result.

[297] Treating does not mean achieving a meaningful clinical response and I do not accept Dr. Marshall’s definition that there must be a “certain therapeutic effect” (Dr. Marshall Report at para 91). “Treating” means a physician administering or prescribing adalimumab to an IBD patient according to the dosing regimen of the patent.

This seems right to me, not just on the text.

The question goes to utility. Generally, clinical efficacy does not have to be established at the time of filing. If the research has proceeded to the point where there is sufficient information to support a valid patent, the patent agent will want to draft a valid claim. The drafter will never specify “clinical efficacy” in the claim unless it is crucial to validity, because doing so needlessly raises the bar for utility. It seems to be consistent with a purposive interpretation to construe the claims on the presumption that the drafter did not intend to draft the claim in a way that would undermine its validity.

Is it Necessary to Disclose the Special Advantage of a Selection Patent?

Pharmascience Inc v Bristol-Myers Squibb Canada Co 2022 FCA 142 Locke JA: de Montigny, Monaghan JJA affg 2021 FC 1 Zinn J FC Selection

2,461,202 / 2,791,171 / apixaban / ELIQUIS

This decision raises some issues related to selection patents and the date for assessing sufficiency. I’ll deal with these in separate posts, starting with selection patents.

The patents in suit relate to the anticoagulant compound apixaban which is used in treating thromboembolic disorders, including stroke. The 202 patent claims the compound apixaban as such, as well as its use in the treatment of thromboembolic disorders [4]. The 171 patent claims various formulations of apixaban. Only validity was at issue. Four related actions, two each involving Pharmascience and Sandoz as defendants, were tried “on a coordinated basis” at first instance. Only Pharmascience pursued its appeal [1].

The more interesting issues relate to the 202 patent and its validity over the prior art patent 2,349,330. Apixaban is a selective inhibitor of the enzyme Factor Xa (FXa) and it works by blocking certain clotting proteins in the blood. The 330 patent disclosed and claimed a large class of compounds which were disclosed as FXa inhibitors and so potentially useful in the treatment and prevention of thromboembolic disorders [FC 79]. The class is extremely large: “more possibilities than stars in the universe” [FC 80]. On the evidence provided by Sandoz’s experts, the class encompassed apixaban [FC 77], but did not specifically disclose it, or describe how to make it [FC 82]. Thus, the 202 patent is what would traditionally be called a selection patent over the 330 patent. A three-part test for the validity of selection patents was set out in IG Farbenindustrie (1930) 47 RPC 289 (Ch) 322-23 and endorsed in Sanofi 2008 SCC 61 [9]-[11]. The factors, in summary, are: (1) a valid selection must possess some substantive advantage over the genus; (2) all of the selected members must possess the advantage; (3) the advantage must be peculiar to the selected group.

There has been considerable debate as to the role of the IG Farbenindustrie factors and their relationship to the standard grounds of invalidity. In Olanzapine 2010 FCA 197 [33] the FCA stated that an assessment of whether the IG Farbenindustrie conditions have been met “does not constitute an independent basis upon which to attack the validity of a patent,” and “A selection patent is the same as any other patent. Its validity is vulnerable to attack on any of the grounds set out in the Act.” This implies that the IG Farbenindustrie factors are a reflection of one or more of the standard grounds of invalidity. As discussed in my post on Zinn J’s decision, he was quite skeptical of the IG Farbenindustrie factors. He stated that failure to meet those requirements is not a stand-alone basis to find a selection patent invalid [FC 104], and he rejected anticipation [FC 83] and obviousness [FC 88] attacks without reference to the factors, though he did go on to hold that if compliance with the IG Farbenindustrie factors were required, the 202 Patent meets that test [FC 106–08].

So, the continued vitality of the IG Farbenindustrie factors and their relationship to the standard grounds of invalidity remains unclear. The FCA will no doubt clarify that relationship in due course, but chose not to do so in this decision. On appeal, Pharmascience argued that Zinn J erred because he never made a clear finding that the 202 patent is a selection patent. Locke JA noted that Zinn J had nonetheless clearly held that the 202 patent satisfied the IG Farbenindustrie criteria and in particular that it disclosed the special advantages of apixaban (the first factor, and typically the most important) [12]. The FCA then went on to separately affirm Zinn J on anticipation and obviousness, without reference to the IG Farbenindustrie factors as such [16]–[21]. Thus, it is not clear whether Locke JA considered it necessary for Zinn J to address the IG Farbenindustrie factors.

Requirement to disclose the special advantage

The most commonly applied IG Farbenindustrie factor is the first, requiring that “There must be a substantial advantage to be secured or disadvantage to be avoided by the use of the selected member” Sanofi [10.1]. This is readily seen as a reflection of the non-obviousness requirement in the context of selection patents. If the prior art discloses a large genus of compounds and discloses that all members of the genus possess a certain property, eg all the compounds treat cancer, then it is not inventive to select one species from that genus and show that it treats cancer, since we already knew that all members of the genus treat cancer. To select one species from a genus is an “arbitrary selection” discussed here and here. A selection must therefore have some special advantage—it cures cancer with fewer side effects, or with higher efficacy—compared with what was known about the genus as a whole.

In this case, Pharmascience argued that Zinn J “erred in failing to recognize (i) that the 202 Patent did not indicate a special advantage, and (ii) the requirement that a selection patent disclose the special advantage of the selection (because the special advantage is the invention, and a patent must disclose its invention” [11]. That is, the argument is not just that apixaban must have a special advantage over the compounds of the 330 patent, but that the special advantage must be disclosed in the 202 patent. This argument seems to be ultimately derived from Sanofi-Aventis v Apotex 2013 FCA 186 [51] (though that case was not cited), where the FCA said “In the case of selection patents, as we have seen, the novelty of the selection and its advantages (including disadvantages to be avoided) are the invention and must be described in the patent,” relying at [45] on Sanofi 2008 SCC 61 [114] (and also Olanzapine 2010 FCA 197 [78], which cited the same paragraph of Sanofi.) Sanofi in turn stated

[114] While double patenting requires a comparison of the claims of a genus and selection patent, it is necessary that the specification of the selection patent define in clear terms the nature of the characteristic which the patentee alleges to be possessed by the selection for which he claims a monopoly. See I. G. Farbenindustrie, at p. 323.

The SCC was evidently referring to the following passage from IG Farbenindustrie:

It should be obvious, after what I have said as to the essence of the inventive step, that it is necessary for the patentee to define in clear terms the nature of the characteristic which he alleges to be possessed by the selection for which he claims a monopoly. He has in truth disclosed no invention whatever if he merely says that the selected group possesses advantages. Apart altogether from the question of what is called sufficiency, he must disclose an invention; he fails to do this in the ease of a selection for special characteristics, if he does not adequately define them. The cautions repeatedly expressed in the House of Lords as regards ambiguity have, I think, special weight in relation to selection patents.

This all looks like good authority for the proposition that a valid selection patent must disclose the special advantages. (Though the reference to ambiguity is a bit mystifying.)

The difficulty with this proposition is that it is well-established that the inventive concept that renders the invention non-obvious need not be disclosed: see Raleigh v Miller (1948) 65 RPC 141 (HL) 161 (Lord MacDermott) stating “a patentee is not bound to include in his specification a statement of the inventive step”; Consolboard [1981] 1 SCR 504, 532-33; BUSM v Fussell (1908) 25 RPC 631 (CA) 649-653. The specification must disclose an invention—which is to say, subject-matter that is in fact new, useful, and non-obvious—but it is not necessary for the specification to disclose what makes it an invention. There has been some confusion on this point because it was at one time necessary in English law that the patent should distinguish what is new from what is old, on the view that the specification ought to inform the public of the bounds of the monopoly. But that function is now served by the claims, which serve that purpose by defining the scope of the monopoly directly, rather than by distinguishing old from new: see BUSM v Fussell (1908) 25 RPC 631 (CA) 650-51, explaining that the rule became outdated with the advent of claims. So, it is now well-established that it is not necessary to describe in what respect the invention is new. So long as the claimed invention is in fact novel and sufficiently disclosed, “There is no obligation to go further, and to state why it is novel, or what in it is novel”: Consolboard [1981] 1 SCR 504, 531-32, quoting with approval BUSM v Fussell (1908) 25 RPC 631 (CA) 651. The same principle applies to obviousness.

Given that there is no general requirement to disclose the inventive concept, either there is no requirement to disclose the special advantage of a selection, or there is some special rule that applies only to selection patents, and the latter is contrary to the Olanzapine principle that “A selection patent is the same as any other patent. Its validity is vulnerable to attack on any of the grounds set out in the Act.”

As the SCC pointed out in Consolboard at 532, a contrary rule would “impose[] an impossible burden on the patentee. Suppose he was not aware of another invention that was very similar to his own. If he failed to specify the differences between the two inventions, the patent would be invalid notwithstanding that his invention was new and different.” It would be “grossly unjust” to require the inventor to keep abreast of every development in her field, nor would such a requirement benefit the public: BUSM at 652, quoted in Consolboard at 532. So, suppose that the inventor in this case was not aware of the 330 patent, notwithstanding that it was prior art; it would then be impossible for the patent to specify the special advantages of apixaban over the compounds disclosed in the 330 patent.

In any event, Locke JA did not have to address these issues because he held that the patent did in fact disclose special advantages of apixaban [12], though he did state that “There was no need for the 202 Patent to provide an explicit comparison of apixaban to any other particular compound falling within the scope of the 330 Patent” [14]. He also stated that “There is no requirement for a selection patent to discuss the special advantages of the selection over the genus in any particular way” [38]. This is ambiguous, as it is consistent with the view that it is necessary to disclose the special advantages, just not in a particular way, such as explicit comparison with the genus compounds.

Usefulness as the special advantage

The decision also raises a different issue. Locke JA noted that the special advantage of apixaban is that “it was useful” [12, quoting FC 107, original emphasis]. The fact that apixaban is useful is a special advantage over the compounds of the 330 patent because “the skilled person reading the 330 Patent would understand that not all of the claimed compounds would be useful in treating and preventing thromboembolic disorders” [12, quoting FC 107]. This is consistent with the view that the first IG Farbenindustrie factor is a reflection of the non-obviousness requirement, as Locke JA made the same point in the context of the obviousness discussion. He noted that Zinn J identified the inventive concept as the selection of apixaban from the other compounds of the 330 patent, and that the difference between the inventive concept and the state of the art, reflected in the 330 patent, was that apixaban was “effective” in treating thromboembolic disorders, as opposed to the compounds of the 330 patent, which merely had the “potential” to be useful [20].

This is a bit odd on its face. If the 330 patent were valid, then we would know that the compounds encompassed by it, including apixaban, are sufficiently useful to satisfy the utility requirement. If the genus is useful and we pick a species out and say it’s useful too, isn’t that just arbitrary selection? As the defendants argued “It is obvious that apixaban can be used to treat thromboembolic disorders as the 330 Patent states that the bicyclic compounds claimed therein (which include apixaban) are effective FXa inhibitors and are useful for the treatment of a thromboembolic disorder” [FC 85]. This may be a reference to the statement in the 330 disclosure that “These and other objects, which will become apparent during the following detailed description, have been achieved by the inventors' discovery that the presently claimed bicyclic compounds, or pharmaceutically acceptable salt or prodrug forms thereof, are effective factor Xa inhibitors.”

There are a couple of answers. First, assuming the 330 patent is valid, the bar for patentable utility is low, requiring merely a sound prediction of utility for the claimed compounds. Selecting a compound that is actually useful from among many that are merely predicted to be useful is arguably inventive, assuming, as Zinn J found at [FC 87], that the search is difficult enough. This is not very different from the advantages of the selected compound in Sanofi, which had “greater therapeutic effect and less toxicity” [78] than the other compounds of the genus. (It seems clear that there is nothing special about lower toxicity; no doubt greater therapeutic effect alone would suffice.) While the 330 compounds and apixaban both satisfy the utility requirement, maybe the 330 compounds barely clear the hurdle and apixaban cleared it easily. That is the thrust of both Zinn J and Locke JA’s reasoning on this point, though the point might perhaps have been made more clearly. Zinn J did not draw a direct comparison between the utility of the genus and apixaban, and while he did find that apixaban was useful, he did not state explicitly that it surpassed the standard easily. With that said, the evidence he accepted that the studies “really leaves no doubt” that it would be therapeutically useful in humans certainly indicates that it substantially surpassed the relatively low bar for patentable utility.

In any event, even if we don’t know whether the 330 patent is valid. We know that many granted patents are invalid, and it is possible that most of the compounds are actually not useful at all, even on the basis of sound prediction. In that case, selecting a truly useful compound could well be a special advantage. One way or the other, the bottom line is that just because the 330 patent says the disclosed compounds are effective, there is no particular reason to believe that statement.

The apparent puzzle largely disappears if we simply ignore the fact that the patent is a selection and treat the genus patent as prior art with no special status. If the genus patent in fact establishes that all compounds withing its scope will actually be effective, then selecting one from the group will be obvious. But if it simply indicates a direction to be explored, and actually finding a useful compound is arduous, then there should be no objection to granting a patent to the successful explorer. On that view, the key aspect of the decision is Zinn J’s statement that the invention was “the result of hard work, innovative thinking and a bit of good luck” [87], though that is somewhat conclusory; see also the evidence, accepted by Zinn J in the context of anticipation, to the effect that it would be only a small percentage of the genus compounds that would be safe and effective in treating thromboembolic disorders [FC 81]. I wonder if the question of whether apixaban had unexpected utility over the genus patent might have been addressed more explicitly if the genus patent had simply been treated as part of the prior art, rather than framing the issue in terms of a selection patent.

Arbitrary Selection?

Merck Sharp & Dohme Corp v Pharmascience Inc 2022 FC 417 Furlanetto J

            2,529,400 / sitagliptin phosphate salt form / JANUVIA / NOC

The key claim at issue in Merck’s 400 patent, Claim 4, is to the crystalline monohydrate form of the dihydrogenphosphate [DHP] salt of sitagliptin, which is used to treat type 2 diabetes [1], [12]. The closest prior art is PCT application WO 03/004498, which disclosed a genus of compounds, specifically exemplifying sitagliptin, among other compounds, both as a free base and hydrochloride salt [11]. The crystalline monohydrate DHP salt was evidently not specifically disclosed.

Pharmascience argued that the 400 Patent is invalid for obviousness and/or insufficiency [3], [64]. The insufficiency attack was a straightforward “how to make” argument, alleging that the disclosed process details were not sufficient to allow a skilled person to make the crystalline monohydrate [255]. This attack failed on the facts: [277–78].

The more interesting attack was obviousness. Given that sitagliptin was specifically disclosed in WO498, the inventive step lay in selecting sitagliptin as the lead compound from the other compounds, or in selecting the salt form, or in the combination of those decisions. As I see it—and this is not the way Pharmascience expressed it—the basic case for obviousness is that Merck just chose sitagliptin arbitrarily from the compounds disclosed in WO498, then did more or less routine salt screening to come up with a salt form with reasonable stability and manufacturing properties.

Suppose a genus patent discloses a class of compounds as having certain useful properties, such as utility in treating diabetes. There is no invention in picking one compound at random from the genus and showing that it is useful in treating diabetes; the genus patent told of us that all members of the genus are useful for that purpose, and the compound is a member of the genus, so of course it is useful in treating diabetes. This is known in European law as an “arbitrary selection”: see Dr Reddy's [2009] EWCA Civ 1362 [44]–[52]; and see Actavis v ICOS [2019] UKSC 15 discussed here. While the term “arbitrary selection” is not as commonly used in Canadian law, the same principle has been recognized in Canadian law: see Teva v Janssen 2018 FC 754 [198], Locke J, expressly noting that arbitrariness is not inventive; Millennium v Teva 2019 FCA 273, applying the principle; Janssen v Teva 2015 FC 247, also applying the principle. This is why a selection must have some special property over what is already known about the genus: this is not a matter of special rules applicable to selection patents, but simply a reflection of the fact that if the selected compound has no advantage, then it is not inventive to select it. Supose I hand you a bag, saying “This is full of red rocks” and I pull out a handful of rust-coloured pebbles of various shapes and sizes to show you. If you reach into the bag yourself and pull out another rusty pebble, no one will be surprised. Now, I can’t predict in advance exactly which rusty pebble you will take from the 100s in the bag—precisely because the selection is arbitrary. An arbitrary selection is both uninventive and unpredictable at the same time.

I don’t see much in Furlanetto J’s decision to suggest that the selection of sitagliptin was anything other than arbitrary. She states:

[203] The evidence indicates that the [skilled person] would not have any specific motivation arising from WO498 to focus on the particular crystalline form of a salt of sitagliptin over the other compounds disclosed within WO498.

But an arbitrary selection is exactly one where there is no particular motivation to pick the compound over the other compounds of the genus, precisely because one is as good as the other.

Furlanetto J also noted that “In June 2003, the only publication on sitagliptin was WO498. There was no indication as to its activity, efficacy, toxicity or tolerability. Further, the general level of potency reported in WO498 indicated a level of potency that was 45 to 167 times less than [another compound known in the prior art]" [210]. This was generally in a support of a “teaching away” argument. But teaching away is evidence of non-obviousness when the compound that was disparaged is subsequently discovered to have exceptional properties. I don’t see anything to that effect in Furlanetto J’s discussion.

Furlanetto J concluded

[212] There is no evidence that the PSA would be drawn to sitagliptin as a starting point for further development. On the basis of the prior art, it cannot be concluded that there was motivation for anyone other than Merck to move forward with sitagliptin as a lead compound.

Again, this is consistent with saying that Merck selected sitagliptin arbitrarily.

I’m not saying that the selection was arbitrary; only that the decision doesn’t identify this possibility and explain why it’s not. I think part of the reason for this is the focus on the inventive concept. Of course, the Windsurfing / Polozzi framework invites us to identify the inventive concept, but it is now recognized that this can be treacherous, with the FCA having wrestled with the question of whether a focus on the inventive concept is unhelpful, helpful or necessary: see here, here and here. This is a context where it seems to have been unhelpful. Pharmascience argued that “the choice of sitagliptin as the starting point does not form part of the inventive concept” [207]. Once we decide that it is, the question becomes whether that was obvious, and that leads to the asking whether it was predictable, which, in my view, is the wrong question.

The question might have been addressed with an objective problem-and-solution approach, in which case the issue of arbitrary selection would have been addressed in identifying the objective problem. Or it might have been addressed simply by asking is whether there was an inventive step involved in choosing sitagliptin.

There is also the old IG Farbenindustrie (1930) 47 RPC 289 (Ch) in Sanofi 2008 SCC 61 [10]–[11] approach to selection patents. Furlanetto J had a fairly long discussion of whether the claimed invention was a “selection patent” over WO498 invention of [77]–[95]. While she recognized that the validity requirements are the same either way, she expressed the view that this assists the Court “in understanding ‘the nature of the beast’” and makes it “easier to compare the facts of the particular case before the court with other previous fact scenarios” [80]. Ultimately she decided that it was a selection, but this doesn’t seem to have played any important role in the obviousness analysis.

In particular, she didn’t mention the IG Farben factors. There is good reason for this omission. The IG Farben factors have fallen out of favour, for two main reasons. First, as Rennie JA pointed out in Apotex v Shire 2021 FCA 52 [32] “A selection patent is subject to the same requirements and vulnerable to the same attacks as any other patent, including attacks based on anticipation and obviousness.” The IG Farben factors must therefore either be duplicative of standard grounds of attack, or wrong. And, as discussed here, the second and third IG Farben factors are arguably wrong, or at least problematic—though these factors were rarely if ever actually used to strike down a patent.

So there are good reasons to avoid the IG Farben approach. With that said, there is a reason the IG Farben factors were used for so long. It is the first factor that was almost always the crux of the analysis:

1. There must be a substantial advantage to be secured or disadvantage to be avoided by the use of the selected members.

This factor is a convenient way of expressing the inventive step requirement in the context of selection patents. It directly addresses the problem of arbitrary selection. As just discussed, if an entire genus has certain properties, it is obvious that any member of the genus will also have those properties. There is no invention in reaching into the bag of rocks and pulling out a rusty pebble; but if you rummage around and pull out a ruby, to everyone’s astonishment, that deserves a reward. The “advantage to be secured or disadvantage to be avoided” is the extra ingredient ensures the patentee has extracted a ruby from the bag, and not another red rock.

So, while the first factor is strictly redundant, in the sense that it simply encapsulates the inventive step requirement, it nonetheless poses that requirement in the context of selection patents in a way that is perhaps more helpful than the Windsurfing / Pozzoli focus on the inventive concept.

There seems to be more of a case that the development of the crystalline monohydrate form of the DHP salt was inventive, as there was a significant amount of work involved. But it seems that it was routine to develop the DHP salt [226]–[229], which was suitable for tableting. Developing the particular crystalline monohydrate form was challenging, but it is not clear to me that it offered significant advantages over the DHP. What was the problem solved by the crystalline monohydrate? It was difficult to develop, but if it did not solve any problem, in what way was it inventive? I have to admit that I came away with the nagging feeling that the inventors developed the DHP salt, and its properties were just fine, but they kept on going to develop something more difficult, just because they knew that the routine work to develop the DHP salt would not support a patent. There’s nothing in the decision that I can point at to support that notion, but I don’t see anything to rule it out, either. If that’s right, this is different from arbitrary selection, since the crystalline monohydrate was difficult to develop, but it’s still not clear to me that it’s inventive. Arbitrary selection is reaching into the bag and extracting a random rock that has the same properties as the others that were already known; inventive selection is rummaging painstakingly through the bag to find the single ruby; developing the crystalline monohydrate seems to have been like rummaging painstakingly through the bag to find the rusty rock that is most nearly spherical, even though being spherical offers no practical advantage over any other shape. One may at least question whether that is deserving of a patent.

Ultimately, even accepting all of Furlanetto J’s factual findings, I am having difficulty putting my finger on the inventive step in this invention. Assuming I’m not just missing the something, this strikes me as problematic. We need a doctrinal framework to guide the analysis, but in the end, a good framework should do more than simply generate an answer—it should help us understand the nature of the contribution that deserves the reward of a patent. I don’t see anything specifically wrong doctrinally with Furlanetto J’s analysis, and I suspect that ultimately my difficulty lies with the way the problem has been framed, with the focus on identifying the inventive concept.

A Selection Patent Does Not Differ in Substance or Form from Other Patents

Apotex Inc v Shire LLC 2021 FCA 52 Rennie JA: de Montigny, Gleason JJA affg 2018 FC 637 Fothergill J

            2,527,646 / lisdexamfetamine [LDX] / VYVANSE / NOC

Rennie JA’s decision for the FCA is noteworthy in three respects. First, Rennie JA held in the clearest of terms that nothing turns on whether a patent is characterized as a selection patent. This is a welcome reaffirmation of prior holdings to much the same effect, but this decision is important because the issue was central to the appeal and the FCA’s holding was so unequivocal. The issue can now be considered settled, as I’ll discuss in this post. Second, Rennie JA provided an extensive discussion of the role of the inventive concept in the obviousness analysis. In many ways this discussion is a helpful clarification of the law, but I must admit that there are some aspects of the decision that make me nervous. I’ll have another post on that issue.

Third, Rennie JA reaffirmed that the various “obvious to try” considerations factors set out in Sanofi 2008 SCC 61 [69]–[70] are factors, not requirements, and they need not all be met [106]–[107[. Normally I would devote a post to this aspect of the decision, but I discussed the issue at length in my very recent post on the decision of Locke JA (Webb, Boivin JJA concurring) in Zytiga 2021 FCA 45, which addressed the point at greater length. The decision of Rennie JA in this case appears to me to be entirely consistent with that in Zytiga, so I won’t blog separately on this aspect of the decision. The important point is that we now have two separate panels of the FCA expressing the same view. In Zytiga Locke JA stated that “I maintain my view that ‘more or less self-evident that it ought to work’ should be treated as a factor in the obviousness to try analysis, and not as a requirement” [36]. This way of putting it suggested that there might be some question as to how widely that view is shared. In light of Rennie JA’s decision in this case, I think this point too can now be considered settled.

As discussed in my post on Fothergill J’s decision, the 646 patent relates to the compound LDX and its use for treating ADHD. Amphetamines have long been used to treat ADHD, but they can be abused by snorting or injection. LDX provides a solution to that problem in the form of an abuse-resistant sustained release prodrug [31]. The main validity attack was based on AU Patent No 54168/65, which disclosed a very large class of related compounds. None of the specific examples disclosed LDX itself, but LDX was included in as one of the many possible configurations in the “advantageous” class described on page 4 of AU 168 [FC 104]. The advantages of LDX as a solution to the problem of amphetamine abuse were not described in AU 168 [FC 108], and the compounds were not even identified as prodrugs [FC 107]. The thrust of AU 168 was that the disclosed compounds might be useful as appetite suppressants that are substantially free of CNS stimulatory activity [FC 106].

There was a question as to whether the 646 patent was a selection patent over AU 168. Fothergill J found it unnecessary to decide this question [29], [FC 98], essentially on the view that nothing turned on that point:

[FC 97] By statute, the basis for assessing anticipation cannot depend on whether the patent is a selection patent or not. The jurisprudence does not imply that anticipation and obviousness in respect of a selection patent are to be assessed over the genus patent from which it is a selection, rather than over the prior art read as a whole. A selection patent “does not in its nature differ from any other patent” (Sanofi-Synthelabo at para 9), there is no reference to selection patents in the Patent Act, and the conditions for a valid selection patent do not constitute an independent basis upon which to attack the validity of a patent. A selection patent, like any other patent, is therefore vulnerable to any attack set out in the Patent Act, but no other.

Apotex argued that this was an error, and that it does matter whether the 646 patent is a selection patent. In particular, Apotex argued that the advantageous properties of a compound over the prior art may only be considered if the patent is a selection patent over that prior art; otherwise, only the bare chemical formula should be considered [19].

Rennie JA firmly rejected this argument. While he did not specifically refer to Fothergill J’s remarks at [FC 97], his holding is very much to the same effect:

[31] There is no magic to the term “selection patent”. . . . The Patent Act does not refer to selection patents and the jurisprudence is clear that a selection patent does not differ in substance or form from other patents.

[32] A selection patent is subject to the same requirements and vulnerable to the same attacks as any other patent, including attacks based on anticipation and obviousness. [T]he failure of a judge to characterize a patent one way or another . . . is not an error of law. Put otherwise, a finding that the characteristics of a selection patent have, or have not, been met, “does not constitute an independent basis upon which to attack the validity of the patent”

[34] As noted, the validity analysis does not change depending on whether the patent was formally classified as a selection patent or not.

[35] There is no divergence between the requirements for a valid patent claim depending on whether it is found in a selection patent or not.

This seems to me to be entirely right.

But there is a loose end. Rennie JA did not directly explain how his holding could be reconciled with the fact that the SCC in Sanofi 2008 SCC 61 [10]–[11] endorsed the three factors set out by Maugham J in IG Farbenindustrie (1930) 47 RPC 289 (Ch). I’ll make an attempt.

The obvious way to reconcile the IG Farbenindustrie factors with the view that there is nothing special about selection patents is to suggest that those factors are simply an alternative articulation of the standard analysis in the specific context of selection patents. This was evidently Rennie JA’s view: “classification contextualizes what specific claims profess to do while also making it easier to compare the facts of the particular case before the Court to other previous fact scenarios” [33].

This was also apparently the view of Maugham J: “I think it would be unwise to endeavour to state in definite language all the conditions on which a selection patent must depend; for after all a selection patent does not in its nature differ from any other patent and is open to attack on the usual grounds of want of subject-matter, want of utility, want of novelty and so forth” (322)’ the same view was also suggested in Sanofi [9], quoting part of this passage.

With that in mind, consider the factors themselves, as set out in Sanofi [10]:

1. There must be a substantial advantage to be secured or disadvantage to be avoided by the use of the selected members.

While the tripartite IG Farbenindustrie test was often referred to, this first factor was undoubtedly the most important, as it was the factor that was normally actually applied (along with Maugham J’s statement at 321, which is not part of the tripartite test as such, that “the selected compounds have not been made before, or the patent would fail for want of novelty.”) It is therefore fortunate that this factor is easy to reconcile with a standard analysis. It was explicit in both IG Farbenindustrie and Sanofi that this is a reflection of the inventive step requirement: “If the selected compounds. . . possess a special property of unexpected character. . . I cannot see that the inventive step essentially differs from the step involved in producing a new result by a new combination of well-known parts . . . in mechanics in the construction of a new machine” (321), quoted in part in Sanofi [9]; “if for practical purposes it is not obvious to skilled chemists in the state of chemical knowledge existing at the date of a selection patent that the selected components possess a special property, there is then, or at least there may be, a sufficient ‘inventive step’ to support the Patent” (322). The idea is that if the originating patent disclosed a large class of compounds and disclosed that they are all, eg antipsychotics with moderate to severe side-effects, there is nothing inventive in pulling one at random out of the class and establishing that it is an antipsychotic with moderate to severe side-effects. An arbitrary selection is not inventive.

The second factor is as follows:

2. The whole of the selected members . . . possess the advantage in question.

This is trickier. Maugham J stated that if this is not satisfied, “the patent would be misleading and would also fail for insufficiency and non-utility.” Unfortunately, Maugham J did not provide any further details, and it is not immediately clear how the patent would fail in light of English patent law of the day. In modern Canadian law, “misleading” patents are dealt with under s 53, which now has a well-developed body of jurisprudence. To the extent that the second factor was intended to address such problems, it must considered to be displaced by the statutory provision of the Canadian Act. I have to admit it is not clear to me why the patent would be insufficient; a failure of the second factor would not affect the classic “how to make” enablement. There are other aspects of the English sufficiency law of the day which Maugham J might have had in mind, eg ambiguity. The law of utility has also developed considerably since 1930, and arguably if the genus claim in the originating patent was valid, all the claimed compounds which it encompassed must have been useful, and it is hard to see how they would become less useful simply because some other compounds are more useful.

With all that said, it is clear that Maugham J considered his second factor to be nothing more than the application of general principles to the context of selection patents. To the extent that modern Canadian law has diverged from century old English law, then it seems plain that our modern law should govern; it would be wrong to regress—or worse, displace statutory provisions—by adhering to a test for selection patents that was based on century old English law.

On to the third factor:

3. The selection must be in respect of a quality of a special character peculiar to the selected group.

Maugham J at 323 indicated that this was not based on established principles, but was rather his own analysis. It not clear whether this was ever good law. I am not aware of any case which has actually applied this factor to hold a patent invalid. Moreover, this factor was also criticized on substantive grounds by the EWCA in Dr Reddy’s [2009] EWCA Civ 1362 [39], which also held that the whole IG Farbenindustrie test was no longer part of English law.

None of this conflicts with the SCC holding in Sanofi. The only one of the three factors actually applied was the first—see eg [31], [31], [41]— which, as discussed above, is clearly based on the standard non-obviousness requirement. Moreover, while the SCC in Sanofi certainly spoke favourably of Maugham J’s analysis, the Court did not go so far as to say that the tripartite test had to be satisfied as a matter of law. Rather, the SCC stated that it is “a useful starting point for the analysis to be conducted in this case.” [11]. Experience has shown that it is not a particularly useful starting point in most other cases. It is not contrary to Sanofi to recognize that fact.