Diagnostic Methods at CIPO

Something strange is going on at CIPO in respect of diagnostic method patents. As discussed in Richard Owens’ op-ed in the National Post and Kathleen Marsman’s article of 2 November in the Lawyer’s Daily, CIPO is systematically refusing to grant patents for some diagnostic methods. The holiday break gave me a chance to look into the issue in a bit more detail (and to read the fascinating IP Fly on the Wall materials that shed light on how those policies were developed), and to write a longer post than usual on the issue.

What is CIPO doing with patent applications related to diagnostic methods?
CIPO is systematically denying certain patents related to diagnostic methods, including, for example, patents related to personalized medicine technologies which tailor medical treatment to an individual patient. So, different kinds of cancer that superficially seem the same, may actually be very different metabolically, and consequently respond differently to different treatments. Before personalized medicine, all a doctor could do would be to try one treatment after another on their patient, and hope that something worked before it was too late. Personalized medicine allows the doctor to tailor the treatment to the disease, and the particular patient. For example, certain breast cancers produce large quantities of a compound known as HER2/neu, and the drug trastuzumab is effective in treating those types of cancer, but not others. Personalized medicine might involve testing the blood of a breast cancer patient and prescribing trastuzumab, if, and only if, elevated levels of HER2/neu were detected. Diagnostic methods generally rely on establishing a relationship or correlation between what CIPO calls the “analyte” (eg HER2/neu), and the existence of a particular disorder, or response to a certain drug (eg trastuzumab).

That CIPO is systematically denying certain patents related to diagnostic methods may not be well known to the public, but the fact itself is not disputed or secret. On the contrary, it is set out explicitly in the Practice Notice Respecting Medical Diagnostic Methods - PN 2015-02, and in the November 2017 update to CIPO’s Manual of Patent Office Practice (MOPOP), Section 17.03.04 [now 23.033.04a] that subsumed and elaborated this examining practice.

To be patentable, any invention has to be inventive, or non-obvious. The “inventive concept” is the idea or discovery that makes an invention inventive. Both PN 2015-02 and MOPOP say, in effect, that if the inventive concept behind an application for a diagnostic method is the discovery of a new analyte, the invention will normally be considered patentable subject matter, and the patent will be granted so long as the other requirements for patentability are met. (CIPO calls this a case of a "data acquisition problem".) On the other hand, if the inventive concept is the discovery of a new correlation between a condition and a known analyte, the invention will not be considered to be patentable subject matter, and a patent will be refused, even though the diagnostic method is new, useful and inventive. (CIPO calls this a case of a “data analysis problem”). That is, it is official CIPO policy that patents should be refused for otherwise patentable diagnostic methods where the inventive concept is the discovery of a diagnostic correlation.

There is no medical or scientific reason for distinguishing between diagnostic methods where the inventive concept is the discovery of a new analyte, and those where it is the discovery of a new correlation. The diagnostic method is just as useful for the patient in either case, and CIPO does not claim otherwise (or the patents would be rejected for lack of utility). The science behind a new diagnostic method may be just as difficult and pathbreaking, regardless of whether the inventive concept is the discovery of a new analyte or a new diagnostic correlation, and CIPO does not claim otherwise (or the patents would be rejected for obviousness). None of this is controversial.

CIPO’s legal reasoning
Why is CIPO refusing to grant these patents for this type of new, useful and inventive diagnostic methods? While CIPO does not assert that there is any scientific basis for the distinction it draws, CIPO is of the view that there is a legal basis for the distinction. This in itself is cause for concern. Patent law is often said to be technical, but that is because of the effort to fit the law to a complex technical reality. When CIPO is denying patents solely for technical legal reasons, divorced from any scientific reality, we must suspect that there is a problem either with the law itself, or with CIPO’s understanding of the law.

To start, it is important to emphasize what is not the reason. There are some kinds of invention which are considered not to be patentable subject matter, regardless of the degree of technical inventiveness or usefulness to society. The most famous example in Canadian law is the so-called “Harvard Mouse,” a mouse that was genetically engineered to be especially susceptible to cancer. This was very useful in cancer research, and a major scientific breakthrough at the time, yet the Supreme Court of Canada in the Harvard Mouse decision, 2002 SCC 76, held, essentially on ethical grounds, that higher life forms are not patentable subject matter, even though they might be new, useful and inventive. One might imagine that CIPO had decided that diagnostic methods are not patentable for some kind of similar policy reason. But that is not what is happening. CIPO is perfectly willing to grant patents for diagnostic methods – so long as the inventive concept lies in the discovery of an analyte, and not in the discovery of a correlation. If CIPO were of the view that diagnostic methods were not patentable subject matter, it would refuse patents for all diagnostic methods, regardless of what the inventive concept might be.

If that’s not CIPO’s legal reasoning, what is? I don’t want to get caught up in the details of the specific passages of MOPOP and the Practice Notices. As with the fact that CIPO is systematically refusing to grant patents for diagnostic methods, the substance of CIPO’s reasoning is not itself secret or controversial. The controversy is as to whether that reasoning is legally correct. So I will simply set out what I understand to be the core of CIPO’s reasoning.

CIPO’s basic position is that diagnostic methods where the essential element of the invention is a thing, such as a molecule, is patentable, while a diagnostic method where the essential element is information, such as a correlation between a known analyte and a disorder, is not patentable. Note that CIPO uses “essential element” to mean, in effect, the inventive concept, or what might be called the “point of novelty” in US law, or the “technical contribution” in European law. This use of the term is not correct, for reasons I will go into later. The precise terminology is not important. The point is that CIPO focuses on the aspect of the invention that makes it inventive, and enables it to overcome and obviousness objection. So, if the analyte was not known, and identifying it took inventive ingenuity, the method will be patentable. But if the analyte was known, and the disease was known, and all that is new is the correlation between the two, CIPO’s position is that the method will not be patentable.

An Invention is Information
Thus, CIPO’s position is that an otherwise new, useful and non-obvious invention is not patentable if the inventive concept is itself information. This position is wrong. As Lord Hoffmann pointed out in Merrell Dow [1995] UKHL 14, [28], “An invention is a piece of information.” In Shell Oil [1982] 2 SCR 536, 549, the Supreme Court of Canada stated that an invention is “the practical embodiment of the new knowledge.” And in Consolboard [1981] 1 SCR 504, 517, the SCC stated that it is the information in the disclosure which is “the quid pro quo for which the inventor is given a monopoly for a limited term of years on the invention,” not the thing which the inventor might have produced. In Merrell Dow, the point of Lord Hoffmann’s statement was that for prior art to anticipate, providing the thing itself is not enough; the prior art must provide information, or the invention has not been disclosed. In Shell Oil the invention, held patentable by the SCC, was a new use for a known compound. There was no new thing, and nothing inventive involved in putting the invention into effect. The inventive concept was as much pure information as in the case of the discovery of a new diagnostic correlation. Similarly, in Continental Soya [1942] SCR 187 the patentee had discovered that soy beans contain a natural bleaching agent and claimed soy bean meal as an agent for bleaching flour. Again, there was no new thing, only information as to a new use for an old compound. The SCC held this to be patentable.

That an invention is a piece of information is true of any invention. It is true for diagnostic methods inventions where the inventive concept is the discovery of a new correlation, but it is equally true for diagnostic method patents which CIPO is willing to grant. As we have seen, CIPO considers a diagnostic method to be patentable when the point of novelty is the discovery of a new analyte. But in that case, the analyte itself is not really new – it is a compound that has always existed in the human body. The inventor’s contribution in such a case is information, namely information that (a) the compound exists, and (b) it is correlated with a disease condition. The first point, that the analyte exists, is as much pure information as its correlation with a disease condition. Yet CIPO takes the view that information of type (a) will support a patent, but information of type (b) will not. It is difficult to see what basis there is for this distinction. Indeed, information regarding the correlation is arguably more important, because if an inventor only discovers a new compound, without any information as to its role, it is clear law that a patent will be refused for lack of utility. As another example, CIPO also considers that a diagnostic method will be patentable if the inventive concept lies in the application of a known method in a new way, such as within a specific time window (17.03.04a, n12). But in that case the method itself is not new, but only the information that it works best when applied within the specified time window.

Thus, the inventive concept underpinning an invention is always information, and the invention itself is “the practical embodiment” of that information. The Supreme Court of Canada in Shell Oil, 554, summarized the law as follows:

A disembodied idea is not per se patentable. But it will be patentable if it has a method of practical application.

CIPO does not dispute that the discovery of a previously unknown correlation is “new knowledge” and CIPO does not assert, or even suggest, that the diagnostic methods where the inventive concept is a correlation are any less practical in terms of diagnosing disease and saving lives than those in which the inventive concept is a new analyte. Thus new information regarding a correlation between a known analyte and a disease clearly has “a method of practical application” when claimed as a diagnostic method, and CIPO does not assert otherwise. Refusing to grant such patents is simply contrary to established law.

Where did CIPO go wrong?
If CIPO’s position is so unfounded in law, where did CIPO go wrong? From the IP Fly on the Wall materials it is clear that many, and indeed perhaps most, of the individuals at CIPO involved in the debate over diagnostic methods, understand perfectly well the law that I have set out above, and disagreed with the policy that is now in place. My sense is that the true root of the problem is a misunderstanding of the nature of an invention on the part of some individuals, whose opinion carried the day. Professor Rob Merges once remarked that for the US founding fathers, “if you put technology in a bag and shook it, it would make some noise” (14 Berkeley Tech LJ 577, 585 (1999)). This view was understandable in the early days of the industrial revolution, when many inventions were machines, and no doubt this notion that an invention is a thing retains considerable intuitive appeal. But every time this argument has been squarely raised it has been authoritatively rejected, in cases like Shell Oil (new use for an old product), Hickton’s Patent, (1909), 26 RPC 339, approved in Shell Oil 550 (a new method of operation of old machinery), NRDC [1959] HCA 67 (new use for known compound), and Calgon 2005 FCA 410 (new use for known method).

In any event, I don’t want to dwell on the psychology of decision making at CIPO, so I’ll turn to CIPO’s formal position, which turns on what it calls a purposive construction of the patent. The core of CIPO’s reasoning, as I understand it, is as follows: Claims define the invention for the purpose of assessing validity. The claims are construed through purposive construction. Purposive construction requires identification of the essential elements. Therefore the essential elements define the invention for the purpose of assessing validity.

The logic is superficially appealing, and all of the premises in this chain of reasoning, taken individually, have support in the caselaw. Yet the conclusion is clearly wrong. This is because “essential elements,” as implemented by CIPO, is more or less equivalent to the inventive concept, and so the approach boils down to assessing subject matter validity against the inventive concept, instead of against the claims. But this is contrary to the principle that the claims define the scope of the monopoly, and therefore all aspects of validity are assessed against the invention as claimed. In particular, assessing the subject matter requirement against the inventive concept, is known as the “contribution analysis” and is not permitted: Amazon.com 2010 FC 1011, [43].

The error lies in the step from saying that the inventive concept is relevant to construing the claims, to effectively saying that the inventive concept is the claims. As discussed above, the invention, in the sense of the inventive concept, is always information, yet information cannot be claimed as such; it can only be claimed as a method of practical application. So, when Newton discovered F=ma, he could not obtain a patent in which claim 1 was “F=ma”, but he would be able to claim a method of aiming a cannon in which the only point of novelty was the information that force equals mass times acceleration. The reason for this rule is not a concern over the breadth of the claims. It is permissible to claim all applications of the entire inventive concept. Whether it is possible to do so will depend on the nature of the inventive concept. If the inventive concept is very fundamental, such as Newton’s laws of motion, it will be very difficult to claim all applications, whereas if the contribution is narrower, it may well be possible to claim every practical application: see Kirin-Amgen [2004] UKHL 46 [77]; Burton Parsons [1976] 1 SCR 555, 565.

The key point is that the claims must be framed in terms of a practical application of the inventive concept. Section 27(8) of the Act provides that no patent shall be granted for any “abstract theorem.” Both words are important. It does not say that no patent may be granted for any theorem. As discussed above, a theorem, or idea, or discovery, or information, is always the contribution for which a patent is granted. The prohibition is against granting a patent for an abstract theorem; that is, a theorem as such, without being defined as a practical application. For a discussion of the policy behind this principle, see my article on The Rule Against Abstract Claims: History and Principles, (2011) 26 CIPR 205. Regardless of the policy, the principle itself is extremely well established.

With this in mind, the significance of shifting from assessing validity against the claimed invention to assessing validity against the inventive concept is apparent. The inventive concept is necessarily abstract. The inventive concept supporting an invention cannot be claimed as such. The point of the claims is to define the invention in terms of a practical application. If the exercise in “purposive construction” strips away the practical application and focuses solely on the inventive concept, of course the invention will be unpatentable.

Purposive construction does not turn the claims into the inventive concept
Purposive construction requires identification of the essential elements, and it does in effect require identification of the inventive concept, but it does not turn the claims into the inventive concept.

The leading Canadian cases on purposive construction are Free World Trust 2000 SCC 66, and Whirlpool 2000 SCC 67. Purposive construction was described by the SCC as a unitary inquiry into the intent of the inventor (see eg Whirlpool [45]), but as a practical matter, as Lord Neuberger recently pointed out in Actavis [2017] UKSC 48 [54], there are two distinct issues. One is the problem of “normal interpretation” [58], which typically involves interpreting specific words or phrases. Does “vertical” in a claim mean exactly 90°, or does it mean sufficiently upright to serve the function of supporting the upper beam. Does “anhydrous” mean a salt without any water of hydration at all, or are salts with some amount of water included in the claim, and if so, how much? What compounds are encompassed in a Markush claim? What does “dextro-rotatory isomer” mean? In the context of normal interpretation, “purposive construction” means that terms like “vertical” are not be understood mechanically, but rather they must be understood as they would be by a person skilled in the art, reading the claims in the context of the specification as a whole. Sometimes this normal interpretation is so trivial as to be almost invisible, and other times it is contentious and even determinative, but one way or the other, it happens in every case.

The second aspect of purposive construction “involves not merely identifying what the words of a claim would mean in their context to the notional addressee, but also considering the extent if any to which the scope of protection afforded by the claim should extend beyond that meaning” Actavis [56]. Should a claim to “pemetrexed disodium” extend to pemetrexed dipotassium (Actavis)? Should a claim to a farm equipment in which the “hindmost” wheels can be dismounted extend to equipment in which the foremost wheels can be dismounted (Van der Lely [1963] RPC 61)? Should a claim specifying that a coil should “hang freely” extend to a mechanism in which the coil is mounted to a fixed base (McPhar Engineering [1956-60] Ex CR 467, 35 CPR 105)? Should a claim to a boot with pieces attached “side-by-side” extend to a boot in which the pieces overlap (Bauer v Easton 2011 FCA 83 aff’g 2010 FC 361 post here)? Does a claim to a “helical spring” encompass a rubber rod (Improver [1990] FSR 181)? In all these cases, the problem arises when a defendant has taken advantage of the inventive concept of the patented invention, while avoiding infringement on a normal interpretation. The issue then, as the Protocol on the Interpretation of Article 69 EPC puts it, is to balance "a fair protection for the patent proprietor with a reasonable degree of legal certainty for third parties" (Actavis [53]).

The inventive concept is relevant to this inquiry, because the problem arises when the defendant has taken the information provided by the disclosure – the quid pro quo for which the monopoly is granted – while avoiding the monopoly defined by the claims as normally interpreted. A wide range of positions are defensible. One extreme is to say that normal interpretation is enough, and if the claim as normally interpreted fails to encompass an embodiment which takes the inventive concept, that is a self-inflicted wound. At the other extreme is the old “pith and marrow” approach. The essential elements analysis is somewhere in between. Broadly speaking, the UKSC in Actavis held that the inventor is rebuttably presumed to have intended to claim everything which a skilled person would understand to be obviously based on the inventive concept which was disclosed. The extent to which that might differ from the essential elements analysis in Canadian law is a question for another day. The important point here is that anything beyond the narrowest approach, which refuses to go beyond normal interpretation, requires consideration of the inventive concept. The intuition is that an inventor who has invented and disclosed a patentable invention is entitled to a claim with a scope commensurate with her technical contribution, and she should therefore be presumed to have intended her claim to extend that broadly. (I should say that I am not asserting that this is an established legal principle; I am simply trying to explain why, intuitively, identification of the inventive concept is relevant to the scope of the claims under an essential elements analysis.)

Normal interpretation is routine. Analysis of essential elements is rare. This is because, most of the time, not only did the inventor, or her agent, intend to draft a claim commensurate with the scope of her contribution, she will have actually succeeded. Most of the time, a well-drafted claim, normally interpreted, will have a scope commensurate with the inventive concept. There is therefore usually no need to go beyond normal interpretation. The essential elements analysis becomes important only when the claims as normally interpreted do not encompass as much material as the inventor was entitled to claim, either because of poor drafting (or, more accurately, the impossibility of perfect foresight in drafting), or, as in Actavis, because of objections by the examiner.

Two points arise from this. First, validity often turns on normal interpretation. For example, in B.V.D. Co. v. Canadian Celanese Ltd. [1937] S.C.R. 221 the specification disclosed a method of making stiffened shirt collars by weaving a thermoplastic fabric into cotton fabric and applying heat. However, the claim was to a thermoplastic fabric “associated” with cotton fabric, and that was already known, in the form of a sheet of thermoplastic laid on top of a cotton fabric and heated together. The claim was held invalid as anticipated, notwithstanding that the concept of weaving the thermoplastic into the cotton was novel. In contrast, the essential elements analysis, properly applied, never results in invalidity. If the claim at issue in BVD had said “woven,” so that the normal interpretation of the claim was confined to such materials, no court would ever say “Your claim as normally interpreted is valid, but we hold that ‘weaving’ is not essential to your claim, and so, as we construe it, the claim extends to fabrics which were previously known, and the claim is therefore invalid.” This argument is self-evidently ridiculous, but that is fundamentally what CIPO is doing in the case of diagnostic methods. The reason that no court would ever so hold is that the whole purpose of the essential elements analysis – the reason it was developed in the first place and refined and varied through multiple cases – is because of the desire to give the inventor protection for what she has in fact invented. Because of the countervailing need to provide legal certainty to third parties, the essential elements analysis may result in a construction that excludes some embodiments which may have drawn on the inventive concept disclosed in the patent. But uncertainty can only ever increase when the court goes beyond normal interpretation, and the court will therefore never go beyond normal interpretation in order to invalidate a patent. When going beyond normal interpretation would invalidate the patent, both factors, fair protection for the inventor and legal certainty for third parties, point against going beyond normal interpretation.

More specifically, while the inventive concept is relevant to the essential elements analysis, that analysis never holds that the claim should be construed as a claim to the inventive concept as such. The inventive concept in Actavis, for example, was the discovery that the toxicity of antifolates is mitigated by co-administration with vitamin B12. An analysis equivalent to that which CIPO applies to diagnostic methods would be to say “The essential element of the patent at issue is that the toxicity of antifolates is mitigated by co-administration with vitamin B12. The claims are construed to encompass only the essential element, and that is a discovery as such. Discoveries are not patentable as such, and therefore the patent is invalid.” No court applying an essential elements analysis has ever said anything remotely like that. The essential elements analysis is always conducted in the context of construing the claims with the ultimate objective of determining whether the defendant’s specific embodiment infringes. The question is not whether a claim to “pemetrexed disodium” should be construed to mean “I claim the discovery that the toxicity of antifolates is mitigated by co-administration with vitamin B12.” The question is whether it should be construed to include pemetrexed dipotassium. To return to the basic principle set out in Shell Oil, an idea “will be patentable if it has a method of practical application.” The question in the essential elements analysis is as to the scope of practical application encompassed by the claim. To say that the inventive concept is relevant to the scope of the claim, is not the same as saying that the inventive concept is the claim. The “purposive construction” which CIPO offers as a justification bears only the most superficial resemblance to any version of purposive construction developed and applied by the courts.

Genencor
One response to this might be to point out that CIPO does not consider itself strictly bound by Free World Trust and Whirlpool, in light of the Federal Court decision in Genencor 2008 FC 608. In Genencor Gibson J pointed out that “Whirlpool was, of course, an impeachment proceeding,” and he held that Binnie J’s endorsement of purposive construction in Whirlpool “was directed to trial judges and to judges of courts of appeal and not to patent examiners in the course of examinations” [62]: see MOPOP 13.05, n 3; and Refresher Training on Purposive Construction (PCon) – March 2017. This may explain why the “purposive construction” applied by CIPO to diagnostic methods is so different from that applied by the courts.

Now, the suggestion that the Whirlpool principles of purposive construction do not apply in examination may be doubted. While it is true that Whirlpool was an infringement proceeding, there is nothing in the SCC’s decision suggesting that the principles were confined to that context. And subsequent to Genencor, in Amazon 2011 FCA 328 the FCA held explicitly that “the jurisprudence of the Supreme Court of Canada, in particular Free World Trust and Whirlpool, requires the Commissioner’s identification of the actual invention to be grounded in a purposive construction of the patent claims” [43]. Moreover, despite the sentence I quoted from Genencor, I’m not convinced the decision actually stands for the proposition that principles of purposive construction do not apply in examination. Paragraph [52] of Whirlpool, which was quoted and distinguished by Gibson J, emphasized the need to construe the claims “with reference to the entire specifications,” and that the terms “must be read in context.” The Re-Examination Board in Genencor did indeed read the claims in context, and with reference to the entire specification, as is evident both from the summary provided by Gibson J at paras [64]-[65], and from the Board’s decision as a whole, which Gibson J appended to his reasons. The Re-Examination Board did carry out a purposive construction, and counsel for Genencor was evidently trying to argue that the Board erred on the facts in its conclusion, while dressing up this argument as an error of law. Gibson J’s conclusion at [70] may be taken to suggest that all he meant is that the Board was entitled to rely on its own expertise rather than that of expert witnesses in undertaking its purposive construction.

In any event, what Gibson J actually meant, and whether he was right or wrong in his holding, is all very interesting – but for our present purposes, it is entirely beside the point. If CIPO really does believe, rightly or wrongly, that it is not bound to apply purposive construction as set out by Free World Trust and Whirlpool, this makes things worse, not better.

The basic problem is that CIPO is not following Shell Oil. The story I told above was to the effect that CIPO departed from Shell Oil because it was trying to apply Free World and Whirlpool, but erred in its understanding of those cases. In that case, we might understand how CIPO could have departed from Shell Oil, if it believed, in good faith, albeit erroneously, that Free World and Whirlpool mandated a different result. But if CIPO truly believes that Free World and Whirlpool do not apply in examination, that story doesn’t hold up.

Instead, the true story appears to be that CIPO chose to defy Shell Oil, based on its own independent policy decision, without regard to the law set out in the Act or by the courts. This is completely impermissible. A patent is a statutory right; the Commissioner of Patents has “no discretion to refuse a patent . . . if the statutory criteria are met.” Harvard Mouse decision, 2002 SCC 76, [11, original emphasis]; and see ibid [119], Monsanto [1979] 2 SCR 1108, 1119.

Conclusion
CIPO is systematically refusing to grant patents for diagnostic methods where the inventive concept is the discovery of a diagnostic correlation, even though such methods are every bit as new, useful and non-obvious as diagnostic methods where the inventive concept is the discovery of a new analyte. CIPO does not assert that there is any scientific basis for the distinction it draws. To put the best face on it, CIPO’s position is based on a misunderstanding of the law related to purposive construction. A more troubling view is that CIPO has not misunderstood the law, but ignored it, in furtherance of some hidden policy agenda. Either way, CIPO’s position on diagnostic method patents is contrary to established law as summarized in Shell Oil.

I must say that I am mystified as to why CIPO is singling out some types of diagnostic methods in this way. As I have explained above, the same reasoning used to refuse patents for diagnostic methods where the inventive concept is a new correlation, could equally be used to refuse patents for diagnostic methods where the inventive concept is a new analyte. Indeed, it could be used to refuse patents for any type of invention on which CIPO’s unique brand of “purposive construction” is brought to bear. Diagnostic methods are a crucial part of effective medical treatment. Why CIPO is singling them out for differential treatment is baffling. As I noted above, it is clear from the IP Fly on the Wall materials that many people at CIPO pointed out the defects in the current policy as it was being developed, so I expect that answer must have something to do with internal personal politics. That is no basis for an effective and legal patent system.

So far as patent principles and patent law are concerned, diagnostic methods where the inventive concept is a new correlation are not different in any relevant way from those where the inventive concept is a new analyte. A data acquisition problem is not patentably different from a data analysis problem. Diagnostic method patents are no different from method patents generally. The obvious solution to the whole problem is simply to rescind PN 2015-02 and MOPOP 17.03.04.