Time for Re-Formulation of the Patent Sound Prediction Test From Apotex Inc. v Wellcome Foundation Ltd., 2002 SCC 77 (“AZT”)

Part 3 of the three part test for sound prediction from AZT (“The Test”) should be added to part 1. This would better reflect the practical application of the sound prediction test and avoid unnecessary judicial scrutiny into ultimately irrelevant factual and evidential areas.

It has been 10 years since the Supreme Court of Canada (“SCC”) established The Test (in Apotex Inc. v. Wellcome Foundation Ltd., 2002 SCC 77). The Test has invalidated many pharmaceutical patents (and more recently, non-pharmaceutical patents). As we await another SCC decision on utility and sufficiency, this may be an appropriate time to evaluate its sound prediction test, and rationalize the need for a change/clarification of The Test.

The SCC set out the simply phrased sound prediction analysis as follows:

1. there must be a factual basis;
2. the inventor must have …an articulable and sound line of reasoning from which the desired result can be inferred from the factual basis; and
3. there must be proper disclosure. “Normally, it is sufficient if the specification provides a full, clear and exact description of the nature of the invention and the manner in which it can be practised”. (per H. G. Fox 1969 Text).

It is noteworthy that disclosure (part 3) was not in issue in AZT.

Three issues arise from The Test as described by the SCC:

1. the SCC appears to leave open whether the facts and line of reasoning must always be in the patent to meet the disclosure requirement;
2. although The Test is understandably directed at whether the inventors had a sound prediction, practically once a court is reviewing the soundness of the prediction, the question is really whether the person skilled in the art understands that the inventors had a sound prediction – arguably it is the person skilled in the art who assists the court by evaluating the factual basis and sound line of reasoning; and
3. where the invention is not a new use of an old compound (as it was in AZT), is there any special requirement for disclosure – or is the requirement merely the standard sufficiency requirements of the Patent Act which in fact the SCC was citing in the Fox 1969 text.

These questions have been discussed in several decisions. There have also been varying interpretations of the AZT Decision.

In some cases, the Federal Court (“FC”) has relied on the part 3 disclosure requirement to find a patent invalid for no sound prediction. Most notably, Mr. Justice Hughes cautioned that the public should not have to “scour the world’s publications” to satisfy the disclosure requirement for a sound prediction. The FC has also explicitly recognized that it is the person skilled in the art who needed to be able to “make the leap to predict use of the compound in humans”. Although finding a compound patent valid based on actual utility after construing a low Promise, Mr. Justice Rennie recognized (in obiter) there would be problems with a sound prediction of any higher Promise, and referred to the “common view” of the disclosure requirement of AZT as follows:

[188] The disclosure requirements for sound prediction are more onerous than for demonstrated utility. …For the purpose of sound prediction a patentee cannot rely on a document that has not been included or referred to in the patent.

In other cases, the FC has somewhat ignored the sound prediction disclosure requirement as long as the standard sufficiency test was met. In an action involving a new chemical class for enzyme inhibition in the area of hypertension (“Apotex perindopril”), Madam Justice Snider considered the three part test and found that prior art documents (not any tests in Servier’s patent) satisfied parts 1 and 2 of The Test, which were “reinforced” by Servier (secretly) making and testing perindopril in particular:

[368] Apotex did not argue …that the ‘196 Patent does not satisfy the disclosure element of the three part test for sound prediction. In any event, even if it had, I am satisfied that the specification provides a full, clear and exact description of the nature of the invention and the manner in which it can be practised [AZT].

Both Madam Justice Snider and the Court of Appeal may have considered the “prior art” documents to reflect well known knowledge, that did not require the ill-fated “scouring” prohibited by Mr. Justice Hughes. Three months earlier, in Eli Lilly raloxifene, [2009 FCA 97] the FCA stated that “when a patent is based on a sound prediction, the disclosure must include the prediction. As the prediction was made sound by the Hong Kong study, this study had to be disclosed” (para. 15). Another difference may be that the raloxifene patent was for a new use of an old compound (i.e. the inventiveness was the new use), as opposed to perindopril where the inventiveness was a new class of compounds. This latter difference was certainly not clearly laid out in Apotex perindopril because no one seems to have argued disclosure.

Recent decisions have indicated that although there may have been some uncertainty regarding the disclosure requirement, the FCA has spoken definitively on the topic and upheld its earlier finding in Eli Lilly raloxifene.

One need only review a few cases to start to see the problem with the three part test – for example, ramipril action, latanoprost NOC, olanzapine action#2. Mr. Justice Boivin’s recent application of the sound prediction test to the selection of 1 enantiomer for preventing blood clots aptly illustrates the problem with the sound prediction test. Following a detailed review of the historical work of the inventors, the inventors’ prima facie sound line of reasoning, including their “track record” with similar compounds, the court concludes in a matter of a few paragraphs that there were problems with the disclosure of Sanofi’s patent which had only rat studies. There were no instructions in Sanofi’s patent that these studies “could be extrapolated to the prediction that the compound – clopidogrel – had a use in humans”.

Apotex clopidogrel highlights two problems: (1) the confusing order of The Test; and (2) the trap that patentees can sometimes fall into.

1. The court (and experts) needlessly review facts relating to parts 1 and 2 of The Test, only to find there is no disclosure under part 3.
2. There is a risk of inutility where a patentee provides extensive detail to the court regarding the actual work it has conducted, to show the patentees in fact had a “secret” sound prediction. There is no “secret sound prediction” available to a patentee; the only “secret” aspect of utility that a patentee can rely on is actual utility. Once the court has set the critical bar/Promise for utility higher than what the patentees actually secretly did – the patentee must revert to sound prediction. Just as Eli Lilly unfortunately found in atomoxetine, it is very difficult to switch from an unsuccessful argument for actual utility to sound prediction – because one is left with only the disclosure of the patent (and not the extensive tests relied on for actual secret utility).

Part 3 of The Test should be added to part 1 going forward to streamline/focus the sound prediction test and avoid these problems.