Amarin: Response to comments on Kurabayashi.

Zachary Silbersher

We have received several follow-up questions and comments, both online and offline, regarding some of our earlier posts – especially on the topics of Kurabayashi and prospective settlement.  This post will attempt to collect further thoughts on Kurabayashi.  This post admittedly gets deeper into the weeds than most of our earlier posts, but it is in response to thoughtful questions and comments from those looking for areas of weakness in the district court’s opinion.  We will follow up shortly on another post discussing dynamics of a potential settlement.   

 More on Kurabayashi . . .

The Honorable Miranda M. Du’s Bench Order found that it was not unexpected that pure EPA would reduce Apo-B because Kurabayashi showed a “statistically significant reduction in Apo B levels in the EPA group of 6.9%.  With a p-value of < .001, EPA’s effects on Apo B were highly significant.  In contrast, Kurabayashi reports a non-statistically significant 1.5% reduction in Apo B levels in the control group.”  (Bench Order at 29, citations omitted). 

The argument that Judge Du committed a clear factual error goes like this:  Table 3 shows that the EPA group and the control group both experienced reductions in Apo B (6.9% and 1.5%, respectively).  However, the p+ values for weeks 12, 24 and 48 are labeled “NS” (not significant.)  Given that, Judge Du committed clear error when finding that Kurabayashi showed a statistically significant reduction in Apo B that was attributable to EPA.  Since the reduction in Apo B between the EPA group and the control group (pure estriol) was not statistically significant, Kurabayashi could not have created an expectation that pure EPA would reduce Apo B.

As we discussed in our earlier post, one question is whether Amarin has preserved the right to make this argument on appeal.  Several commenters have pointed out that Amarin’s pre-trial proposed findings of fact included argument to this effect.  That may be so, and Amarin may have indeed preserved the right to challenge Judge Du’s interpretation of Kurabayashi on appeal.  But there are a few caveats to consider.

First, to make that challenge, Amarin is still limited to the evidence in the record.  Judge Du essentially adopted the generic’s argument regarding how to interpret Table 3 from Kurabayashi.  That argument was included in the generics pre-trial proposed findings of fact.  (See Docket 333 at paras. 418 and 422).  Thus, Amarin was aware that the generics might make this argument going into the trial.  Despite that, neither Amarin’s pre-trial nor post-trial proposed findings of fact appear to include expert testimony on why p+ values are the statistically-significant values rather than the p* values.  

Amarin’s post-trial proposed findings of fact do not appear to raise this argument, and the proposed pre-trial findings of fact raise the argument, but do not appear to delve into the nuanced distinction between the p+ and p* values.  For Amarin to delve into the p-values on appeal, it will theoretically need to point to some expert opinion.  If it did not solicit expert opinion on this at trial—either in the form of direct examination testimony from Amarin’s expert (Dr. Toth) or cross-examination testimony from the generics’ expert (Dr. Heinecke)—then it may be limited in what it can rely upon.  (You cannot introduce new expert testimony during an appeal.)  Amarin may try to make that attorney argument on appeal, and perhaps the Federal Circuit panel will find Judge Du’s misinterpretation of the p-values in Table 3 as clearly erroneous as many of those following this case.  That is not impossible.  But without concrete expert admissions from the testimony at the trial, then it remains just that—i.e., something of a gamble.

There is another caveat to consider, which we discussed in the update to our earlier post.  If we compare Amarin’s pre-trial proposed findings of fact with its post-trial findings of fact, Amarin appears to have dropped the argument challenging the generics’ interpretation of the statistical significance of the 6.9% reduction from Table 3 from Kurabayashi.  If Amarin made a deliberate decision not to challenge Kurabayashi at trial on the basis that the generics were misinterpreting Table 3, it is questionable whether Amarin will choose to return to that argument on appeal.  More to the point, if Amarin refrained from pressing that challenge at trial—most importantly, by refraining from cross-examining Dr. Heinecke on this point—then Amarin may not have sufficient testimonial evidence to mount this challenge on appeal.  (This is another way of saying, if Amarin didn’t solicit and put this testimony in front of Judge Du before issuance of the Bench Order, it can’t really argue on appeal that Judge Du sorted through that evidence incorrectly.)

But if the generics interpretation of Table 3 from Kurabayashi was so clearly erroneous, then why did Amarin refrain from challenging it at trial?  Presumably, the reason is because Kurabayashi was not the only prior art relied upon by the generics to show a reduction of Apo B from pure EPA.  Dr. Heinecke also relied upon Grimsgaard and Nozaki.  The tactical calculation may have been that it is not worth burning considerable time at trial challenging the hairy statistical conclusions of Kurabayashi when other prior art references taught or suggested the same thing.  (The parties did not have unlimited time at trial to present their respective cases.) 

Moreover, Amarin clearly chose to distinguish Kuarabayashi on different grounds—namely, the study focused on the wrong patient population and the EPA group was also administered estriol.  The former ground dovetails with Amarin’s distinctions of other primary prior art references, such as Mori and Hayashi.  Thus, Amarin may have chosen to develop a larger theme at trial to distinguish much of the prior art at once.  Having spent the time to develop the record of those distinctions during trial, that is likely the best record Amarin is equipped to return to on appeal.  This strategy has traction and it may still save Amarin on appeal, as previously discussed.

 There are other issues with Kurabayashi other than Table 3.  One of the problems faced by the generics’ reliance on Kurabayashi was that the EPA group was not administered pure EPA, but rather EPA and estriol.  Thus, Kurabayashi, on its face, does not disclose a reduction in Apo B from administration of pure EPA, which is what the patent claims require.  That is why, to overcome this problem, the generics also argued in their post-trial findings of fact that the results of Kurabayashi show no “interaction or synergy” between EPA and estriol.  (Docket 373 at para. 281).  They were essentially trying to paper over the question of whether Kurabayashi would have anything relevant to say about pure EPA to a person of skill.  Amarin clearly challenged the generics on this point at trial, as shown by citation to testimony solicited at trial within their post-trial findings of fact.  (See Docket 374 at para. 580).  Judge Du did not address Amarin’s evidence on this point within the Bench Order.  Thus, this is a ground on which Amarin may choose to challenge Kurabayashi on appeal based upon testimony clearly developed at trial.

If Amarin does challenge Judge Du’s interpretation of Table 3 at trial, the generics are likely to double-down on the argument that the court’s focus on the 6.9% reduction was, in fact, the correct data point.  The generics will argue that the claims do not require any comparison to placebo (more on that in the next section,) and to show prior art disclosed, taught or suggested a certain benefit (reduction of Apo B) does not require absolute certainty, only reasonable certainty.  Put another way, the generics are likely to argue that even if Judge Du misread the significance (or non-significance) of the comparison between the EPA and control groups, it is irrelevant because the comparison to baseline is sufficient to defeat the showing that reduction of Apo B was truly unexpected.  Indeed, Amarin itself repeatedly argued that the Marine study showed a reduction of Apo B based upon a comparison to both baseline and placebo.  (See e.g., Docket 327 at 7).  So Amarin itself relied upon a comparison to baseline to show the benefit.

This is a hairy issue.  Even given that Judge Du may have misconstrued the p-values in Table 3, Amarin will have to weigh whether it is worth burning pages in its appeal brief to focus on this issue, especially given the other prior art in play and whether testimony was solicited on this issue at trial. 

            What about claim 8 from the ‘677 patent?

We received another comment about an alleged factual error by Judge Du that may be apparent on the face of the patents.  Dr. Heinecke’s reply expert report, dated June 7, 2019, responds to Dr. Toth’s report (Amarin’s expert).  In Dr. Toth’s report, he argues “that the difference between the EPA group [in Kurabayashi] and the estriol-only control group was not itself statistically significant.”  (See Docket 234-12 at 201).  In his reply report, Dr. Heinecke states, “none of the asserted claims as construed requires an actual comparison to a second patient in a control group.”  (Id.).  Yet, one of the asserted claims does appear to require exactly that comparison.  Indeed, claim 8 from the ‘677 patent recites as follows: “The method of claim 1, comprising administering to the subject about 4 g of the pharmaceutical composition daily for the period of at least about 12 weeks to effect a reduction in apolipoprotein B compared to placebo control.”  (emphasis added).

 So, given this, the alleged error on the part of Judge Du appears to be as follows:  Judge Du relied upon Dr. Heinecke’s interpretation of Table 3 from Kurabayashi, and in doing so, she erroneously focused on the reduction in the EPA group compared to baseline rather than the lack of statistical significance between the EPA and control groups.  This error, the argument goes, is compounded by the fact that Dr. Heinecke incorrectly stated that the asserted claims do not require comparison to a second patient control group.  In so doing, did Dr. Heinecke mislead Judge Du about the significance of the comparison between the EPA group and the control group?

The short answer is, it doesn’t seem so.  The key phrase from Dr. Heinecke’s statement in his reply expert report is “as construed.”  That phrase refers the Judge Du’s prior claim interpretations (otherwise known as “constructions”) that issued in the claim construction order.  (See Docket 135).  That order included Judge Du’s interpretation of the phrase “compared to”.  (That phrase appears in many of the asserted claims, not just claim 8 from the ‘677 patent, but we will focus on the language of claim 8.)  Judge Du found that though claim 8 recites “compared to a placebo control,” that comparison is not actually required by the claim.  Put another way, Judge Du interpreted claim 8 from the ‘677 patent in manner that does not require a comparison to a placebo control group. 

While Judge Du’s interpretation may seem counter-intuitive, it was an interpretation that was actually pressed and put forth by Amarin, not by the generics.  Indeed, Judge Du’s construction of the “compared to” language was, at the time, technically a win for Amarin and a loss for the generics.  By contrast, the generics argued that to infringe claim 8 from the ‘677 patent (and other claims with similar “compared to” language) would require a doctor to essentially conduct a clinical trial each time he or she treats a patient to therefore satisfy the “compared to placebo control” limitation.  Judge Du found that to be an absurd result.  

In the big picture, this was a good result for Amarin.  Had Judge Du agreed with the generics on this particular dispute, rather than agreeing with Amarin, then Amarin would never have been able to prove infringement of the patents.  The generics’ proposed labels do not instruct physicians to assess a reduction in Apo B through comparison to a placebo group.  On this basis alone, the generics would have been able to show no infringement on summary judgment, and there never would have been a trial. 

This is not an uncommon dilemma that arises within patent litigation.  If you are the patent-holder, and you stake out a construction of the patent claim that is too broad, it may make it easier for you to show infringement, but it may simultaneously make it easier for the defendant to show invalidity.  By contrast, if the defendant stakes out a construction that is too broad, it may sweep in more prior art and make it easier to show invalidity, but simultaneously open up the door for the patent-holder to more easily show infringement.  Threading this needle, and choosing your constructions carefully, is where cases are won and lost.

And so, Dr. Heinecke is not necessarily incorrect when he stated in his reply report that the asserted claims “as construed” do not require comparison to a placebo group.  Indeed, he references Judge Du’s claim construction order in his report at Docket 234, paragraph 197.