Wednesday, November 23, 2011

Proving the Promise of a Selection Patent

Eli Lilly Canada Inc. v. Novopharm Ltd / olanzapine (No 2) 2011 FC 1288 O'Reilly J
2,041,113 – olanzapine – ZYPREXA

To repeat the introduction from yesterday’s post, in Eli Lilly Canada Inc. v. Novopharm Ltd. / olanzapine (No 1) 2009 FC 1018 O’Reilly J held Lilly’s olanzapine patent, 2,041,113, to be an invalid selection patent over the genus patent 1,075,687. On the appeal, the FCA held that invalid selection is not an independent basis upon which to attack the validity of a patent (2010 FCA 197, [27], [33], [90]), and that O’Reilly J’s finding of invalidity was fatally tainted by this error. The FCA held that the patent was neither anticipated nor obvious, nor was it invalid for double patenting, but it remitted the case to O’Reilly J on the issues of utility and sufficiency. O’Reilly J at [124], following the lead of the FCA at [99], construed the patent as promising superiority to (1) prior known antipsychotics; and (2) other '687 compounds, in particular flumezapine. O’Reilly J held that patent invalid for failure to fulfil these promises. Yesterday’s post discussed the first promise. This post discusses the second promise.

The second promise is directly related to the problem of selection patents. As Hughes J put it the NOC litigation over olanzapine between the same parties, “a valid selection patent is one which claims an advantage for a compound that is within a previously disclosed class of compounds which advantage has not been disclosed in the prior art” (Eli Lilly Canada Inc v Novopharm Ltd / olanzapine (NOC) 2007 FC 596, [134] appeal dismissed for mootness 2007 FCA 359). In IG Farbenindustrie Maugham J set out criteria for a selection patents which were accepted in Sanofi 2008 SCC 61, [10]-[11]. These criteria require that the selected compound must secured a “substantial advantage” over the genus which is “peculiar” to the selected compound. O’Reilly J held the 113 patent invalid for failure to establish a special advantage over the '687 compounds:

[260] More particularly, the evidence does not support a prediction that the alleged advantages of olanzapine over two ‘687 compounds, flumezapine and ethyl olanzapine, are substantial. To the extent they existed at all, their magnitude was insignificant. In addition, there is no evidence that olanzapine was superior to any other compounds in the ‘687 class in respect of the characteristics described in the ‘113 patent. The comparisons did not relate to the class as a whole and I have no evidence that any advantage was peculiar to olanzapine.

(Note that in its decision remitting this case to O'Reilly J the FCA had based the requirement of a special advantage on the promise of the patent, while Hughes J in his NOC decision treated it as a matter of sufficiency. How this requirement should be classified is an interesting question in its own right, but in the end, Hughes J held the patent invalid for substantially the same reason as O'Reilly J: 2007 FC 596, [162].)

O’Reilly J’s finding reveals a fundamental problem with the IG Farbenindustrie criteria. In Sanofi the Supreme Court said that Maugham J.’s analysis in IG Farbenindustrie “is consistently referred to and is well accepted” [11]. While this was true enough at the time, those rules have now been rejected by the English Court of Appeal. As it happens, this took place in the English olanzapine litigation, Dr Reddy’s Laboratories v Eli Lilly / olanzapine [2009] EWCA Civ 1362, [2010] RPC 9 Jacob LJ, Lord Neuberger MR, Richards LJ aff’g [2008] EWHC 2345 (Pat) Floyd J, where Jacob LJ described Maugham J’s rules “as part of legal history, not as part of the living law” [37]. In part this was simply because Maugham J’s rules have never been used in European law, and they were therefore superceded by the Patents Act, 1977, which implemented the European Patent Convention. But Jacob LJ also criticized the rules substantively at [39]:

I would only add this about the IG rules. If one thinks about them, it is difficult to see how, realistically, they could be complied with unless the patentee carried out an enormous range of experiments. How can you show your class (or compound) has a “substantial advantage” over the prior class without experimenting with at least quite a lot of the class – enough to make a sound prediction at the very least? Or how can you show that the quality which makes your selected class is peculiar to that class? If you put in your thumb and pull out a plum, how are you to say that there are no other plums in the pudding?

This criticism goes exactly to the point on which O’Reilly J held that the olanzapine patent to be invalid. To repeat O’Reilly J’s finding, “ [t]he comparisons did not relate to the class as a whole and I have no evidence that any advantage was peculiar to olanzapine.” The reasons the comparison did not relate to the class as a whole is that Lilly did not experiment with “quite a lot of the class,” presumably for the practical reasons adverted to by Jacob LJ.

On the facts, a dozen '687 compounds were synthesized and subjected to preliminary in vitro tests [19]. Three specific compounds were subject to some degree of additional study: these were flumezapine, ethyl flumezapine and ethyl olanzapine. The most extensive studies were carried out on flumezapine, which had even been the subject of clinical trials. After a detailed comparison of each of the studied characteristics [162]-[183], O’Reilly J concluded that “the evidence does not support a prediction that the alleged advantages of olanzapine over two ‘687 compounds, flumezapine and ethyl olanzapine, are substantial” [260]. In other words, the evidence was that flumezapine was just as promising as olanzapine.

Why then did Lilly abandon flumezapine, at a time when olanzapine had not even been synthesized? The answer is that “Lilly halted its clinical trials on flumezapine at the request of the FDA after receiving reports of elevated liver enzymes and a muscle enzyme called creatine phosphokinase [CPK] in some patients” [21]. However, when O’Reilly J considered the evidence related to CPK, he determined that “the flumezapine data on CPK were poor” [178]. The numbers were small, and the reported CPK spikes might have been “isolated laboratory abnormalities” [172], or due to hepatitis infections, or vigorous exercise by the patients in question. The important point here is that flumezapine was insufficiently tested to establish that it had poor CPK results, and therefore it could not serve as basis for establishing the superiority of olanzapine. O’Reilly J stated that “more testing was needed to determine whether flumezapine really did have an effect on CPK and, if so, whether that effect showed up at therapeutic doses and was clinically significant” [178]. However, as a practical matter this testing could not be undertaken, because the FDA was of the view that it was too dangerous to continue the trials. While O’Reilly J concluded that the CPK results for flumezapine were not clinically significant, the FDA was of the view that it was of sufficient concern that patient safety should not be put at risk. No doubt the criteria used by the FDA used for this determination are different than those which O’Reilly J applied, but the point remains that Lilly could not, as a practical matter, obtain the flumezapine data which O’Reilly J viewed as necessary to support the validity of the patent.

Lilly anticipated this problem, and therefore investigated ethyl olanzapine as being the closest analogue, purely for the purposes of patenting olanzapine [185]. This study, using dogs, did not show substantial advantages for olanzapine over ethyl olanzapine, except in respect of effects of cholesterol, for which dogs are a poor model for humans [194]. Lilly might presumably have decided to continue studying ethyl olanzapine as well, and perhaps claim it as well, but then it would have lost the benefit of a comparator in establishing the advantages of olanzapine.

The evidence did established that olanzapine was superior to ethyl flumezapine, which caused a reduction in white blood cells; this was precisely the problem with the previous generation of antipsychotics [20]. “Ethyl flumezapine had been an abject failure” [25]. However, this was not enough to establish that olanzapine had special advantages: while it is not necessary to test all members of the genus class “there must be ‘sufficient representative testing that a person skilled in the art could soundly predict that the surprising characteristic would not be expected to be found in a large number of the other members of the genus’” [227, quoting 2008 FC 593, [70]). (No mention was made of the 45 compounds rejected in the preliminary screen, presumably because these were not directly referenced in the specification.)

In summary, after Lilly had found promise with olanzapine, in order to establish its patentability Lilly would have had to investigate a some significant number of other '687 compounds sufficiently to establish that they were inferior to olanzapine. This is exactly the requirement which Jacob LJ found to be so unrealistic as to warrant the rejection of the IG Farbenindustrie rules. Jacob LJ’s point makes eminent good sense. The purpose of the patent system is to provide an incentive to undertake research that will provide a social benefit. It is expensive enough to test promising compounds; it is perverse to require the inventor to test a wide variety of unpromising compounds from the genus simply in order to prove that they are indeed useless.

No comments:

Post a Comment