Amarin: Did the district court commit error by misinterpreting Kurabayashi?
Zachary Silbersher
Update April 8, 2020: This update is in response to some of the interesting comments below. If Amarin chose not challenge Dr. Heinecke’s interpretation of Table 3 from Kurabayashi, I would not presume that was necessarily a mistake or oversight on the part of Amarin or its counsel. Within the comments below, there appears to be reasonable disagreement over how to interpret Table 3. I am not a statistician, and for the purpose of this post, I take no position on whether Judge Du’s interpretation of Table 3 from Kurabayashi is correct or not. For one thing, the generics relied upon additional prior art references for a teaching or suggestion of a reduction of Apo-B from EPA, namely, the references Grimsgaard and Nozaki. As pointed out by one commenter, Amarin’s pre-trial proposed findings of fact suggested the data in Table 3 from Kurabayashi did not show a statistically-significant reduction in Apo-B between the two groups. Similarly, Dr. Toth’s expert report—which was exchanged long before trial—also included a paragraph opining that Kurabayashi did not show a statistical reduction in Apo B. (See Docket 248 at ¶ 108). Amarin appears to have deliberately decided not to make these arguments a centerpiece of its proposed post-trial findings. That said, on appeal, Amarin can technically rely upon anything in the record, including testimony that was solicited at trial but did not show up in the post-trial findings of fact or post-trial briefs. If Amarin had solicited a knock-out admission from Dr. Heinecke on the proper interpretation of Table 3 from Kurabayashi, then a fair presumption would be that it would wish to highlight it in the post-trial briefing. On the other hand, if it is not in the record, and if Amarin did not solicit expert opinion on the interpretation of Table 3 from Kurabayashi, or on the nuances and import of the legend to the Table, then it will be very difficult for Amarin to introduce that evidence during the appeal.
****
April 7, 2020
The post mortem of the district court’s decision invalidating Amarin’s ($AMRN) Marine patents continues. We received questions regarding a finding of fact related to the prior art reference, Kurabayashi. The Honorable Miranda M. Du from the District of Nevada found that Kurabayashi was not considered during prosecution of the patents and that Kurabayashi suggested that pure EPA would reduce Apo-B. Are these findings accurate? If not, are they a basis to reverse Judge Du’s decision on appeal?
Did the Examiner consider Kurabayashi?
In the course of invalidating the asserted patents, Judge Du considered Amarin’s argument that the inventions are not obvious because the claimed dosing regime reduced Apo-B and that was an unexpected benefit. Indeed, this is one of the reasons the USPTO granted these patents in the first instance—based upon the Examiner’s finding that it was unexpected that the claimed dosing regime would yield a reduction in Apo-B. Yet, Judge Du found that (i) Kurabayashi taught that EPA reduced Apo-B levels (Bench Order at 29-30), and (ii) Kurabayashi was not considered by the Examiner during prosecution of the patents. (Bench Order at 66). Based on this, Judge Du rejected a finding that it was unexpected that the claimed inventions reduce Apo-B.
So why did Judge Du find that Kurabayashi was not considered by the Examiner during prosecution of the patents? Judge Du cited to the generics proposed findings of fact and conclusions of law at pages 246 and 247. (Docket 373). Yet, those pages do not speak to this issue at all. While page 247 refers to Kurabayashi, it does not state that Kurabayashi was not considered by the Patent Office. Rather, it discusses the generics’ argument that Kurabayashi allegedly taught that a reduction in Apo-B from administration of pure EPA.
Yet, on page 85 of the same document (the generics’ proposed findings of fact,) the generics do state that the Examiner found that reduction of Apo-B to be an unexpected benefit, and yet the Examiner did not cite to Kurabayashi. (See Docket 373 ¶ 369). Further, the generics invalidity expert, Dr. Peter Toth, also stated in his expert report that Kurabayashi was not considered by the Examiner during prosecution of the patents. (See p. 45, ¶ 105).
What do the patents themselves say about this? Prior art references considered by the Examiner during prosecution of a particular patent are listed on the face of each patent. Amarin asserted six different patents in this case against the generics. Those patents include: U.S. Patent Nos. 8,293,728; 8,318,715; 8,357,677; 8,367,652; 8,431,560; and 8,518,929. On the first page of each patent, there is a section (numbered (56)) called “References Cited.” Under that heading, each of the foregoing six patents indicates that Kurabayashi was, in fact, considered by the Examiner during prosecution of each patent. As an example, on page 7 of the ‘728 patent, the following citation to Kurabayashi is listed: Kurabayashi, T., et al., “Eicosapentaenoic acid effect on hyperlipidemia in menopausal Japanese women.” ObstetGynecol 96:521-8 (2000).
The file history for the patents appears to back this up. For instance, in the file history for the ‘728 patent, Kurabayashi was disclosed to the Patent Office by Amarin on page 16 of an Information Disclosure Statement filed on February 9, 2012. The Examiner confirmed in a separate filing made on April 4, 2012, at page 16, that Kurabayashi was considered during prosecution of the ‘728 patent. So the file histories for the patents appear to confirm that Kurabayashi was, in fact, considered by the Examiner.
The six patents asserted by Amarin in this litigation are members of a larger family of patents. Each of the patents is a child (otherwise known as a “continuation”) of an earlier parent patent, U.S. Patent No. 8,293,727. The ‘727 patent was not one of the six patents asserted by Amarin in this case. Yet, because it was the parent to each of the asserted patents, both Amarin and the generics treated the prosecution history of the ‘727 patent as essentially a stand-in for that of the asserted patents. For instance, both Amarin’s and the generics’ description of the prosecution history of the asserted patents actually refers to the prosecution history for the ‘727 patent. (See Docket 374, Amarin’s proposed findings of fact, p. 32; see Docket 373, generics proposed findings of fact, p. 82-83).
Was Kurabayashi disclosed during prosecution of the ‘727 patent? Yes. Kurabayashi was disclosed on page 11 of an Information Disclosure Statement filed within the application for the ‘727 patent on June 3, 2011. On June 20, 2011, the Examiner filed a document within the file history for the application for the ‘727 patent indicating that Kurabayashi was considered (see page 11). Indeed, Kurabayashi was identified as a “Reference Cited” on page 6 of the ‘727 patent.
So, given all this, why did Judge Du find that Kurabayashi was not considered by the Examiner during prosecution of the patents? The answer appears to be that, although Kurabayashi was technically “considered” by the Examiner during prosecution of the patents, it was not one of the primary prior art references discussed in any of the Office Actions.
When a patent application is filed with the USPTO, the Examiner conducts a search for prior art. If the Examiner finds any prior art that allegedly teaches the elements of the invention as claimed, the Examiner will reject the patent application. The rejection will come in the form of document known as an “Office Action.” The Office Action will identify the prior art and explain how each element from the pending patent claims is taught within the cited prior art.
Separate from that, an applicant typically unilaterally submits prior art to the Patent Office. Certain statutory and regulatory provisions require that a patent applicant submit prior art believed to be material to the patentability of the application. Failure to do so can lead to loss of an otherwise non-obvious patent. Thus, it is not uncommon for patent applicants to submit hundreds of prior art references. Citations to the prior art references are disclosed in Information Disclosure Statements (IDS). The Examiner then typically files a corresponding document wherein the Examiner indicates whether each of the references filed on the applicant’s IDS was considered.
Accordingly, what happened in this case is not unusual. Amarin submitted Information Disclosure Statements, which cited to hundreds of prior art references. This is evident from the face of the patents, which include numerous cited references. (The references without an asterick are those submitted by Amarin.) Yet, in the course of preparing the Office Actions, the Examiner only discussed a handful of prior art references. Those references included, at the very least, Katayama, Saito, Mori 2000, Hayashi, Grimsgaard, Yokoyama, Mori 2006.
The key distinction here is that, while the Examiner may “consider” hundreds of prior art references cited by the applicant, the Examiner typically only “cites” to and discusses a handful of references within the Office Actions. The Examiner is not obligated to discuss, within an Office Action, whether or not each reference submitted by the applicant does or does not teach any elements of the claimed invention.
Going back to the generics’ proposed findings of fact, the generics actually stated, “[t]he examiner did not cite Kurabayashi, which showed that EPA reduced Apo-B.” (Docket 373 at 85) (emphasis added). Technically, that is correct. The Examiner did not “cite” or discuss Kurabayashi within the Office Actions. As explained above, that is different from suggesting that the Examiner did not “consider” Kurabyashi.
So what does all of this mean? On the one hand, it means that Judge Du may have committed clear error when finding that Kurabyashi was not “considered” by the Examiner. On the contrary, the patents and the file histories themselves confirm that Kurabyashi was considered by the Examiner during prosecution of the patents.
Did the district court misinterpret Table 3 from Kurabayashi?
On the other hand, it is not clear this error will vindicate Amarin. A district court is not hindered from invalidating a patent based upon a prior art reference that was technically considered by the Examiner. Judge Du’s Bench Order did affirmatively find that Kurabyashi taught that EPA would reduce Apo-B. (Bench Order at 58). That finding will be subject to deference by the Federal Circuit. Even if Judge Du was mistaken as to whether Kurabayashi was considered by the Examiner, if Judge Du found that the reference adequately taught a reduction of Apo-B for pure EPA, that may technically be enough to avoid reversal of her decision, at least with respect to Kurabayashi.
But what about Table 3 from Kurabayashi? On page 29 of the Bench Order, Judge Du found that the Kurabayashi study showed a 6.9% reduction of Apo-B for the EPA group, and because this had a p value of < 0.001, Judge Du deemed it to be “highly significant.” On page 30 of the Bench Order, Judge Du pasted in Table 3 from Kurabayashi to illustrate this finding. Yet, the p value for week 48 of the EPA group indicates “NS”, which we can assume to mean “not significant.” Further, the actual Kurabayashi reference includes a legend that is missing from the snapshot of Table 3 included in Judge Du’s Bench Order. Does that legend indicate the difference between the differently-designated p-values in the chart shows a misreading by Judge Du?
The first and most important question is whether Amarin made any of these arguments at trial. If not, then they are essentially waived. You cannot raise new arguments on appeal. Amarin’s principle argument at trial for distinguishing Kurabayashi was that Kurabayashi focused on the wrong patient population, namely, those with TG levels below 500 mg/dL. (See e.g., Amarin’s post-trial brief at 27). Amarin’s proposed findings of fact do not appear to have focused on Table 3 or the argument that the “NS” notation under week 48 for the EPA group is more significant than the p* value of < 0.001 for the same group. (See Docket 374 at 161-62). (In addition, Amarin cannot argue that, since the snapshot of Table 3 in her opinion is cropped, that means Judge Du did not have the benefit of the legend for Table 3. Rather, the full Kurabayashi article was separately submitted into evidence by the generics during trial (DX1534)).
Focusing on the merits of whether Judge Du’s opinion misinterpreted Table 3 of Kurabayashi, this is a relatively technical statistical question on which expert opinion is likely warranted. The Federal Circuit Judges do not pretend to be experts in biostatistical analysis. They will not speculate on the proper way to read Table 3 from Kurabayashi. Rather, the standard for factual findings on appeal in an obviousness analysis is one of clear error. In practice, that means that if Judge Du found one expert’s opinion to be more credible over another, that finding is afforded considerable deference by the Federal Circuit. Amarin cannot simply approach this issue at the Federal Circuit by arguing that no reasonable scientist would have interpreted Table 3 from Kurabayashi in the manner done by Judge Du. For the purposes of the appeal, at least one such scientist has, in fact, interpreted it that way.
That scientist is Dr. Heinecke, who was the generics’ invalidity expert at trial. The generics cite Dr. Heinecke’s trial testimony in their proposed findings of fact on the interpretation of Table 3 from Kurabayashi that finds 6.9% reduction of Apo-B in the EPA group to be significant. Judge Du appears to have credited Dr. Heinecke’s testimony on this point in the Bench Order. Amarin’s proposed findings of fact do not appear to identify any testimony from the trial—either on the cross-examination of Dr. Heinecke or the direct examination of Dr. Toth (Amarin’s invalidity expert)—that would discredit Dr. Heinecke’s reading of Table 3. This is the lens through which the Federal Circuit will likely review Judge Du’s findings for clear error with respect to Kurabayashi.
The theme that Amarin pursued at trial is that there is a material distinction between the patient populations, i.e., those with TG levels below 500 and those above. That theme permeates a number of factual findings, from whether Kurabayashi taught an expected reduction in Apo-B for patients above 500 to whether there was a reasonable expectation from Mori that pure EPA would reduce LDL-C in patients with severe hypertriglyceridemia. As we discussed in our earlier blog post, there is traction to Amarin’s chances of reversing Judge Du’s decision based upon this theme.
Amarin spent considerable time developing this theme at trial, it created a substantial record on this theme, and it is likely to return to this theme on appeal. One of my former mentors, a successful patent trial attorney, frequently reminded me that when you go to trial, you try your case to the Federal Circuit. It remains to be seen whether Amarin successfully did that in this instance.