Friday, November 8, 2019

Application of Principle of Arbitrary Selection Affirmed by FCA

Millennium Pharmaceuticals Inc v Teva Canada Limited 2019 FCA 273 Stratas JA: Webb, de Montigny JJA aff’g 2018 FC 754 Locke J
            2,203,936 / 2,435,146 / 2,738,706 / bortezomib / VELCADE

This FCA decision has an interesting point on arbitrary selection hidden in what is otherwise a routine application of the deferential standard of appellate review. In the decision under appeal, Locke J had granted Teva compensation under s 8 of the NOC Regulations for losses suffered during the time its version of bortezomib was kept off the market. Millennium and Janssen had defended on the basis, inter alia, that sales by Teva would have infringed the 936 and 146 patents. Locke J found both those patents to be invalid for obviousness [1], [FC 344]. Millennium and Janssen appealed this finding [2]. The appeal was dismissed on the basis that “the appellant is trying to transform adverse findings of fact and mixed fact and law into errors of legal principle to avoid the difficult standard of palpable and overriding error” [17]. While there is nothing new here, the FCA decision has a good brief general discussion of the standard of appellate review, noting that an appellate court must read the reasons of the court below “as a whole” in context, while keeping in mind the rebuttable presumption that the first-instance court reviewed and considered all of the evidence [11].

With that said, one novel point regarding arbitrary selection was touched on tangentially. The general problem is this. The prior art discloses a genus compound exemplified by a few specific species, but with many substitutions from a class possible at one location (or several). If the inventor selects a new component from that class that exhibits surprising advantages over the previously disclosed species, that is an inventive selection. But what if the inventor randomly selects a new component that has no advantages over those previously disclosed? In one sense, the component is not obvious, because there are many in the class that might have been selected, and a skilled person looking to develop a new member of the family of compounds would not have picked out that particular compound in advance. But on the other hand, there is no invention in picking out one compound at random that gives exactly the results one would have expected from any random selection from the class. On its face, the arbitrary selection is not obvious, but neither is it inventive. This is the problem of arbitrary selection.

The problem of arbitrary selection was raised in this case. Locke J had held that the compound claimed in the 936 patent was not anticipated because it differed from prior art compounds by the selection of five different components [178]. Moving to non-obviousness, he first considered whether the selection of any of the individual components was inventive. For three components, he held that the selection was obvious or obvious to try: [185 re boronic acid warhead],[189 re dipeptide], [193 re Leu at P1]. But two of the selections raised the problem of arbitrary selection. In particular, as I discussed in my post on Locke J’s decision, one of the points distinguishing bortezomib from that prior art was the selection of the amino acid phenylalanine (Phe) at P2 position of the peptide. It was agreed that several amino acids would work at P2, and the evidence was that a hydrophobic amino acid, such as Phe, would be preferred. The question was whether it was inventive to select Phe from the fewer than 10 hydrophobic amino acids. 

Locke J framed the question not as whether the selection was “obvious,” but “whether it was inventive to select Phe from those options” [195], and he cited (at [198]) with apparent approval the decision of Jacob LJ in Actavis UK Ltd v Novartis AG, [2010] EWCA Civ 82 [36]-[37], discussing the “5¼ inch plate paradox”:

[198] Jacob L.J. referred to a plate of a size that is novel (in that no plate of that size has ever been made) and arguably non-obvious since nothing in the prior art suggests that size. Such a plate cannot be patentable because its size is arbitrary and non-technical. Similarly, the evidence in the present case indicates that the choice of Phe at P2 was similarly arbitrary and non-technical. There appears to be no advantage in selecting Phe over other known options.

Locke J concluded that “I have heard no adequate reason to conclude that the selection of Phe at P2 was inventive,” which implies the holding was based on arbitrary selection, though he also went on to hold that “it would have been obvious to try Phe at P2" [199].

The arbitrary selection principle was reinforced with the fifth selection of pyrazine carbonyl as the N-terminal blocking group. Again Locke J characterized the question not as being one of whether the selection was obvious, but “whether there was anything inventive in the selection of pyrazine carbonyl from among the available options” [200]. This selection raised the problem of arbitrary selection even more directly than the selection of Phe at P2, because “ Contrary to the situation with the selection of Phe at P2, there were many more N terminal blocking group options to choose from that would have been expected to yield good results” [201]. The selection was not obvious in the sense that a skilled person ex ante would not have preferred pyrazine carbonyl over the alternatives, but nonetheless arbitrary because “the work of the inventors did not reveal pyrazine carbonyl to offer better results than other N-terminal blocking groups” [201]. In a sense, the selection was not obvious precisely because it was not inventive: because there was nothing special about pyrazine carbonyl, it was not obvious to select it. Locke J concluded that:

[202] As with the selection of Phe at P2, I have heard no adequate reason to conclude that the selection of pyrazine carbonyl as an N-terminal blocking group was inventive. Despite the greater number of options, it remains the case that there were a finite number of identified predictable solutions known to a skilled person. The 936 Patent does not provide any unexpected results.

This is evidently an application of the principle of arbitrary selection. The reference to a finite number of identified predictable solutions echoes the obvious-to-try test from Sanofi 2008 SCC 61, ¶ 69, but in contrast with the first three components, Locke J did not otherwise hold or imply that this selection was obvious to try.

Locke J then went on to hold that it was not inventive to select the five components in combination, saying “There are a finite number of possible practical combinations to try, and any of them would have been expected to offer some potency” [203]. This is consistent with the arbitrary selection principle, though Locke J did not put it in quite those terms: if we select five components arbitrarily, the combination is still arbitrary, even though the precise combination could not have been predicted in advance.

On appeal, the appellants attacked this holding, arguing that “the Federal Court improperly employed hindsight in its analysis when it broke down bortezomib into component parts for the purpose of analysis” [14]. In particular, the appellants argued that:

If bortezomib was not disclosed prior to the ‘936 Patent, its structure was not known. Thus, it could not be known that the claimed compound had a boronic acid warhead, was a dipeptide, had a leucine side chain at P1, phenylalanine at P2, and a pyrazine carbonyl as the N-Terminal Blocking Group. (Reasons, para 178) However, despite these features not being known as of the time at which obviousness was assessed, these are the headings by which the Trial Judge assessed obviousness of the bortezomib molecule.

The Trial Judge wrongly concluded, on the basis of this improper hindsight analysis and despite a prior art genus of trillions of possibility compounds, that the combination of all these elements in the bortezomib molecule was also obvious.
Appellants' Memorandum of Fact and Law ¶ 42, 43

I read this as arguing, in effect, that the combination must have been non-obvious because the exact combination could not have been predicted in advance by a skilled person.

The appellants’ position is based on the principle that the claimed subject-matter must not have been obvious to a skilled person on the claim date, without the benefit of hindsight. If the skilled person is presented with the problem and asked for a solution, the invention is not obvious unless the skilled person would say “Obviously the solution is X” where X is exactly what is claimed. The principle of arbitrary selection asks instead whether there was any invention involved in arriving at the claimed subject matter; when the selection is arbitrary, the answer is no, even though the specific answer might not have been predicted in advance. If I plunge my hand into a bowl of Halloween jelly beans and pull out a green one, there is nothing inventive about my selection, even though no one could have guessed in advance that it would be green.

The FCA rejected the appellant’s submission on this point, saying that “The Federal Court had ample evidence upon which to find, without employing hindsight, that the selection of components of bortezomib lack inventive ingenuity” [14, my emphasis]. Again, the focus is on inventive ingenuity, not obviousness directly. In affirming Locke J’s decision, the FCA was at the same time affirming the principle of arbitrary selection, at least indirectly.

I acknowledge that I am reading a lot into a paragraph of the FCA decision, which turned primarily on the standard of review. But when faced with this problem, the FC has very consistently held that arbitrary selection is not invention, even though it is not normally couched in those terms, and every further decision to the same effect adds to the weight of precedent. Significantly, Locke J’s decision at trial is perhaps the first to refer to the “5¼ inch plate paradox” and the issue of arbitrariness, and in affirming Locke J on this point, the FCA has added further support to the arbitrary selection principle. This is welcome: in my view, the arbitrary selection principle is entirely sound, for the reasons given by Jacob LJ and as illustrated by the facts in this case.

No comments:

Post a Comment