Friday, August 14, 2015

Disagreement on Date for Assessing Obviousness-type Double Patenting

Eli Lilly Canada Inc v Apotex Inc (NOC) 2015 FC 875 Gleason J
            2,226,784 / tadalafil /CIALIS

In this decision dealing with Lilly’s CIALIS patent, Gleason J disagreed with de Montigny J’s earlier Mylan / tadalafil decision 2015 FC 17 as to the appropriate date for assessing obviousness-type double patenting. In my blog post on Mylan / tadalafil I had agreed with de Montigny J’s reasoning, but on reviewing Gleason J’s decision, I have realized that there are some holes in that logic. At the end of the day, this point was not determinative and so Gleason J expressly declined to come to any firm conclusion on this legal issue. I have spent considerably longer on the blog post than I usually do, and I haven’t come to any firm conclusion either; all I can say is that the law is unsettled, and that is just as well, as the point is a difficult one.

The judicially created doctrine of obviousness-type double patenting precludes an inventor from obtaining a second patent for an invention which is an obvious modification of an invention for which the same inventor holds a prior patent. While the “obviousness” aspect of this doctrine is the essentially same as the obviousness analysis generally, it is a distinct doctrine because it may be raised in circumstances in which the first patent would not constitute part of the state of the art for the purposes of an obviousness attack on the second patent. In particular, obviousness-type double patenting essentially permits a prior unpublished patent application to used as prior art against a subsequently filed application, even though the unpublished application is not part of the statutorily defined state of the art: [60]; s 28.3. Prior unpublished applications were particularly problematic under the old Act, under which patents were not published until issued: for example, in Whirlpool 2000 SCC 67, decided under the old Act, the second patent (734) was filed 2 ½ years after the first patent (803) application was filed, but two years before it was granted. The problem is reduced under the new Act, but it may still arise because patents are not published until 18 months after the priority date. So, in this case the priority date of the 784 patent was two weeks before the publication date of the earlier 377 patent which was the basis for the double patenting attack.

The date for assessing obviousness-type double patenting is important if the common general knowledge has advanced in the interim. Suppose the inventor invents the wheel and files an application at a time when wood is the only known material. Bronze then becomes a widely known material. The same inventor then develops an improved wheel made of bronze and files a second application, before the first application is published. If the date for assessing obviousness is the filing date of the first application, then the bronze wheel is non-obvious; but if the crucial date is the second filing date, then it is obvious over the wheel claimed in the first application. The basic dilemma is that the inventor has already received a reward for inventing the wheel, in the form of the first patent, and it seems wrong that she should be rewarded with an additional term of protection for a bronze wheel, when she did not invent bronze. But on the other hand, the improved bronze wheel would not be obvious to a skilled person even at the time of the second application; the skilled person would know nothing about the wheel, as the first application was still unpublished. If the inventor had waited to file her application until the later date, she would have been entitled to claim both the wheel and the bronze wheel, because neither would be obvious to a skilled person at that time; it seems perverse to say she should not be permitted to obtain a patent for the bronze wheel for the sole reason that she had previously filed the application for the wheel.

This case, and de Montigny J’s earlier Mylan / tadalafil decision 2015 FC 17, both deal with this problem in the context of two Lilly patents. The earlier 2,181,377 patent discloses tadalafil and discloses that it is a PDE V inhibitor. It also claims tadalafil and its use in treatment of certain specified disorders, not including erectile dysfunction (ED). (While not explicitly stated in the decision, I take it that but for this disclosure, the fact that tadalafil is a PDE V inhibitor would not have been obvious to a person skilled in the art at any later date.) But by some later date, it would have been obvious to a skilled person that any PDE V inhibitor would be effective in treating ED. Thus the knowledge disclosed in 377 patent, that tadalafil is a PDE V inhibitor, is equivalent to the invention of the wheel. The subsequent advance in the common general knowledge, that PDE V inhibitors can be used to treat ED, corresponds to the invention of bronze. Both together would make it obvious that tadalafil could be used to treat ED. The key question as a matter of law is when the crucial date is for assessing obviousness. The important question as a matter of fact is when it became part of the common general knowledge that any PDE V inhibitor would treat ED.

In Mylan / tadalafil, Mylan argued that the key date for assessing obviousness-type double-patenting is the priority date of the later (784) patent, and that by that time it would have been generally known that any PDE V inhibitor would be effective in treating ED [132]. de Montigny J rejected this [134], and held instead that the appropriate date is the priority date of the earlier (377) patent [135], [143]. He held that the use of tadalafil to treat ED would not have been obvious over the 377 patent as of that date [142]. However, his holding regarding the appropriate date was not determinative, as he held that on the facts the 784 patent would not have been obvious even as of the priority date of the 784 patent [143]. [149]. (While I have referred to filing dates by way of illustration, and the decisions at hand refer mostly to the priority date, as both patents claimed priority from a foreign filing, but it seems implicit that it is really the choice between the earlier or later claim date that is at issue.)

Gleason J commented on this holding as follows:

[126] Further, I agree with the policy reason advanced by Justice de Montigny in Mylan Tadalafil as being the basis for the rejection of the Canadian filing or publication dates of the 784 Patent (July 11, 1996 and February 6, 1997, respectively) for the conduct of the obvious-type double patenting analysis as selection of either of these dates would circumvent the temporal limitations in section 28.3 of the Patent Act and expose the patentee to prior art that arises after the claim date of the later patent.

While Gleason J is apparently agreeing with de Montigny J, this is actually a very polite expression of disagreement. This rationale described by Gleason J was the reason given by de Montigny J [134], for rejecting the priority date of the 784 patent, not the filing or publication date. Of course, because the filing and publication dates are after the priority date, de Montigny J also implicitly rejected those latter dates as well. But the point is important because Gleason J went on to hold that “a sound argument could be made for the selection of July 14, 1995, the priority date of the 784 Patent, as the appropriate date for the conduct of the obviousness-type double patenting analysis” [129] (and see also [133]). Thus Gleason J only partially accepted de Montigny J’s reasoning.

With that said, her conclusion that de Montigny J was wrong in rejecting the priority date of the later patent was tentative. She held that even using the 784 priority date, on the facts the 784 patent would not be invalid for obviousness-type double patenting. It follows that the same result would also obtain if the earlier priority date of the 377 patent was used. Consequently, despite her apparent preference for the claim date of the second patent, out of respect for comity with de Montingny J’s decision, Gleason J declined to make a “firm determination” as to which date is appropriate [135].

To summarize the state of the law, on the one hand, de Montigny J clearly held that the appropriate date is the priority date of the earlier patent, while Gleason J tentatively held that the appropriate date is the priority date of the later patent, and neither holding was determinative on the facts. At this point, all we can say is that the law is unsettled.

To step back to the policy point, Gleason J’s reason for rejecting the Canadian filing date or the publication date of the later patent is that using either of these dates would:

expose the patentee to prior art that arises after the claim date of the later patent. This would effectively place an inventor with an earlier unpublished patent at a substantial disadvantage in respect of the date for assessing prior art as compared to any other person who might have made the same invention. I agree with Lilly that such a result should not be countenanced as it would discourage inventors from publicizing their inventions after the claim date while their applications are still pending [126].

There are three different arguments in this short passage: (1) the patentee would be exposed to prior art arising after the claim date; (2) it is not fair for two inventors with the same invention to be treated differently; and (3) the differential treatment would discourage publicizing inventions. Point (1) strikes me as sound, and it allows a distinction between the claim date of the second patent and its filing or publication date; point (2) also seems sound, but it does not seem to me to support the distinction between the claim date of the second patent and its filing or publication date; and I’m not sure I understand point (3).

To take point (2) first, suppose Lilly filed its application for tadalafil per se, but prior to its publication, another party, say Merck, independently discovered tadalafil and that it was an PDE V inhibitor. After that, but still prior to the publication of Lilly’s tadalafil patent, it becomes common general knowledge that PDE V inhibitors are effective in treating ED. Merck would be able to file an application claiming tadalafil and tadalafil for the treatment of ED. Merck’s claim to tadalafil would be anticipated by Lilly’s earlier application, but Merck’s claim to the use of tadalafil for the treatment of ED would be valid, because that fact that tadalafil is a PDE V inhibitor is not part of the state of the art against Merck. However, Lilly itself would not be able to obtain a patent for tadalafil for the treatment of ED because the fact that tadalafil is a PDE V inhibitor would effectively be part of the state of the art as against Lilly. Thus Lilly and Merck are differently treated even though both have made the same invention and neither is responsible for the knowledge that PDE V inhibitors generally can treat ED.

That point, if I have understood it correctly, seems sound, but I don’t see why it isn’t also a good argument against using the claim date of the later patent. Take the key date to be the claim date of the second patent. Suppose that the fact that PDE V inhibitors are effective in treating ED becomes common general knowledge only after that claim date. In that case Lilly or Merck would be treated equally, as either could get a valid patent for the use of tadalafil for treating ED. Now suppose that the fact that PDE V inhibitors are effective in treating ED becomes common general knowledge before the claim date. Lilly would not be able to get a patent for the use of tadalafil to treat ED, because the knowledge that tadalafil is a PDE V inhibitor could be used in the obviousness attack. On the other hand, Merck would be able to get a patent for the use of tadalafil to treat ED, because without using Lilly’s first patent, it is not obvious that tadalafil is a PDE V inhibitor.

The third point is that this differential treatment would discourage Lilly from publicizing its invention early. I’m not sure I understand this argument. It seems to me that on the contrary, Lilly would be encouraged to publicize it invention early – if Lilly’s invention is going to be prior art against itself, it might as well be prior art against everyone. Put another way, if Lilly has a compound patent for tadalafil, it is better for Lilly if no one has a patent for its use in treating ED, than if Merck does. Perhaps the argument is that Lilly would have an incentive to delay applying for its first patent, so that it could not be used against Lilly in its subsequent application for tadalafil for the treatment of ED. (A delay in applying would also lead to delaying publication.) While I see this point as a matter of logic, it does not strike me as compelling. Pharmaceutical patentees have a strong incentive to file early to avoid being scooped by a competitor, and it seems to me this incentive would outweigh any possible benefits from delaying in order to reap the benefits of advances in the art that could not be anticipated by the patentee. Maybe I’m missing something on this point.

The first point, that using the filing or publication date of the second patent exposes the patentee to pior art arising after the claim date, contrary to s 28.3, is clearly correct and justifies a distinction between the claim date and later dates such as the filing or publication date. By the same token, it does not rule out the claim date of the second patent. Thus it allows us to reject the filing or publication dates of the second patent, but not the claim date of the second patent. This does not get us very far, because the claim date of the two patents were the real contenders all along.

In concluding that “a sound argument could be made” for using the claim date of the second patent, Gleason J began by noting that there are two ways in which a subsequent patent could be void for obviousness-type double patenting [130].

[131] On one hand, a patentee could under-claim in its initial patent, divide the invention and seek to extend the monopoly period by filing a subsequent patent before the first one is published. In such circumstance, the relevant inquiry involves assessment of what was known at the point in time for assessing obviousness in respect of the first patent.

That is not what happened in this case, but as an aside, as I understand it, the law does not prevent this type of double-patenting. It is well-established, both in the US and Canada, that the double patenting analysis looks to the claims, and the disclosure is ignored except for the purpose of interpreting the claims: Whirlpool [63]; Mylan / tadalafil [125]. The reason for this was explained by in Lilly v Teva 2011-1561, -1562, 689 F.3d 1368, 1378-79 (Fed Cir 2012):

As a general rule, obviousness-type double patenting determinations turn on a comparison between a patentee's earlier and later claims, with the earlier patent's*1379 written description considered only to the extent necessary to construe its claims. This is so because the nonclaim portion of the earlier patent ordinarily does not qualify as prior art against the patentee and because obviousness-type double patenting is concerned with the improper extension of exclusive rights—rights conferred and defined by the claims. The focus of the obviousness-type double patenting doctrine thus rests on preventing a patentee from claiming an obvious variant of what it has previously claimed, not what it has previously disclosed.

Thus in US law, at least, the double-patenting doctrine does not prevent a patentee from underclaiming in the first patent and then filing a later patent for the unclaimed subject matter.

Gleason J continued:

[132] Conversely, a second type of impermissible evergreening concerns the addition of obvious amendments. This is akin to the “non-inventive bells and whistles” rationale exemplified in Farbwerke [Com’r of Patents v Hoechst [1964] SCR 49]. Particularly in the context of pharmaceutical patents involving a new use for an existing compound or class of compounds, there could be a situation where the common general knowledge advances after the claim date of the first patent that would render the new use claimed in the second patent obvious as of the claim date in the second patent, resulting in the later patent being an impermissible evergreening through extension based on obvious mendments to the initial patent. In such circumstance, I believe a sound argument may be made for the selection of the priority date of the second patent as being the date in respect of which the assessment of obviousness-type double patenting should be undertaken.

[133] The present case falls into the second of the above two possible scenarios. . .

I don’t find this persuasive. In Com’r of Patents v Hoechst [1964] SCR 49 the “bells and whistles” were known at the claim date of the first patent. (The first patent was for a process for making a compound useful for treating diabetes, and the second patent was for the compound mixed with a standard excipient. It seems that the second patent was primarily an attempt to avoid the product-by-process restrictions of then s 41, rather than an attempt at evergreening.) Hoechst was not a case in which the common general knowledge advances after the claim date of the first patent; on the facts of Hoechst, the second patent would have been obvious over the first on either claim date. This does not mean that the claim date of the second patent is inappropriate when the state of the art has advanced between the two claim dates; it is only to say that the authority of Hoechst does not help us answer this question.

So, I don’t find the arguments in favour of using the claim date of the second patent to be compelling – though neither did Gleason J, as she held only that “a sound argument exists” for using that date.

Turning to the argument that the claim date of the first patent should be used, in Mylan / tadalafil de Montigny J noted that:

If. . . the relevant date was to be the priority date of the second patent (in this case, July 14, 1995), the obviousness-type double patenting analysis would morph into a pure obviousness analysis, with the added benefit that the timing requirements of section 28.3 of the Patent Act would be circumvented [134].

In my blog post on his decision I said that “This strikes me as a very strong point. The judicially created obviousness type double-patenting cannot be used to do an end-run around a clear statutory provision.” Gleason J was good enough to cite my blog post on this point [127], even though she implicitly disagreed with the conclusion that the claim date of the first patent should be used. And on further reflection I think Gleason J was right to doubt this conclusion: after all, the whole reason that the courts created obviousness-type double-patenting is to do an end-run around the statute by precluding patent for evergreening inventions that would not be prohibited by the obviousness provisions of the statute itself.

de Montigny J also reasoned as follows:

Since the rationale behind this ground of invalidity is the prohibition against an improper extension of the monopoly granted by the first patent, the Court must ascertain whether the invention claimed in the second patent could or should have been included in the first patent [133]

This is a strong point, but it is not unassailable. de Montigny J is right to say that in a case where the common general knowledge has advanced after the claim date of the first patent, the monopoly granted by the first patent is not being extended. After the expiry of the first patent, anyone would be free to use tadalafil for any of the purposes listed in the first patent, such as treating hypertension. It is only its use in treating ED that would continue to be monopolized, beyond the expiry of the first patent, and that use was not obvious at the claim date of the first patent.

The counter-argument is that the purpose of the Patent Act to provide a lure to bring forth new inventions. What Lilly invented is tadalafil, and for that it has received its reward in the form of the first patent. Lilly did not discover that tadalafil is effective in treating ED, and to extend its monopoly for that use would reward Lilly unnecessarily for an invention that it did not make.

I can see the force of the arguments on both side of this question. Without writing an article, instead of a blog post, all I can say is that I think Gleason J was right both to doubt de Montigny J’s conclusion, but also to refrain from coming to any firm conclusion herself.

1 comment:

  1. Thoughtful analysis--but doesn't it illustrate that we've gone astray with double-patenting analysis? If the analysis is truly claims-bound, should not the claims of *each* patent receive a construction based on their own publication date (Whirlpool, para. 55)? Should the construction not be agnostic to later developments in the art (including future patent claims)? Some of my (aging) musings... Hastie, Miles. "Sanofi-Aventis and Double Patenting: The Wake of the Case; A Wake for the Doctrine or the Doctrine Awoken?." Intellectual Property Journal 22.2 (2010): 155.

    ReplyDelete