chatgpt standard disclaimer: ChatGPT can make mistakes. Check important info.
Analysis of Judicial Error in Amarin v. Hikma: Invalidity of Vascepa® Patents
Context: Judge Du’s Holding
Judge Miranda Du issued a 70-page bench opinion invalidating six of Amarin’s patents under 35 U.S.C. § 103 (obviousness). The court concluded that the use of pure EPA (eicosapentaenoic acid) to treat severe hypertriglyceridemia was obvious in light of:
Existing commercial products (Lovaza, which combines EPA + DHA),
Scientific literature (notably Mori (2000) and Kurabayashi (2000)), and
Expert testimony, particularly from the defendants’ witness, Dr. Jay Heinecke.
The decision hinged on two claims:
That EPA does not raise LDL-C, unlike DHA.
That prior art (specifically Kurabayashi) demonstrates statistically significant benefit from EPA alone.
These were used to justify the finding that switching from a mixed EPA/DHA formulation (Lovaza) to pure EPA (Vascepa) would be an obvious step to a person of ordinary skill in the art (POSITA).
Error 1: Misinterpretation of Statistical Significance in Mori (2000)
Court’s Reasoning:
“Serum LDL cholesterol increased significantly with DHA (by 8%; P = 0.019), but not with EPA (by 3.5%; NS).” – Judge Du, citing Mori (2000)
This distinction was central to the opinion’s finding that EPA avoids DHA’s drawback (raising LDL), and thus provided a motivation to pursue EPA alone.
Judicial Error:
This is a well-documented statistical fallacy—mistaking a lack of significance in one group (EPA) versus a significant result in another (DHA) as proof of a difference between the two groups.
Dr. Gavin Jarvis, a biostatistician engaged by EPADI II, explained:
“The presence of a statistically significant difference in one arm and not in the other does not imply a statistically significant difference between the arms.” – Jarvis Expert Report
Further backed by a peer-reviewed Nature Biotechnology article (Curfman, Bhatt, Pencina, 2020), which critiqued the court’s interpretation as:
“A common statistical mistake… the court invalidated patents based on an improper inference of differential effect without appropriate comparative testing.”
Conclusion: The court’s core factual finding (EPA ? DHA in LDL-C effect) was based on improper statistical reasoning, which undermines the motivation-to-combine logic central to its obviousness ruling.
Error 2: Improper Reliance on Cropped Data from Kurabayashi (2000)
Court’s Finding:
“Kurabayashi showed a 6.9% reduction of Apo-B for the EPA group… with a P-value of <0.001, EPA’s effects were highly significant.”
Judicial Error:
The court relied on a version of Table 3 from Kurabayashi that had been cropped by defendants in their post-trial filings. This version:
Omitted the “NS” (Not Significant) legend for the control-vs-EPA comparison.
Presented EPA’s results in isolation without context.
EPADI II demonstrated that:
The original table contained clarifying statistical annotations that contradicted the claimed significance.
The cropped version was reproduced verbatim in the court’s opinion, showing clear judicial reliance on altered evidence.
Consequences:
This omission led Judge Du to conclude that EPA had an independent, statistically significant effect on Apo-B and LDL-C. However, the full data shows that the differences were not statistically significant when compared to control.
This is not a minor oversight—it’s a material misrepresentation of primary trial evidence, creating the illusion of prior art superiority for EPA where none existed.
Error 3: Deference to Expert Testimony Built on Flawed Premises
Judge Du credited the testimony of Dr. Heinecke (defendants’ expert) over Amarin’s Dr. Toth. Dr. Heinecke testified that Mori and Kurabayashi supported the view that EPA does not raise LDL-C, and that EPA alone had proven benefits in lipid modulation.
However:
His interpretation relied heavily on the statistical errors in Mori.
His conclusions on Kurabayashi were based on incomplete data, which EPADI later proved was truncated.
Despite this, the court found Dr. Heinecke more persuasive—creating a credibility-based factual ruling insulated on appeal unless “clearly erroneous.”
But where credibility is derived from flawed or falsified data, the finding cannot stand, especially when the evidence presented to the judge was itself incomplete or misleading.
Error 4: Misapplication of Secondary Considerations
Court’s Acknowledgment:
Vascepa’s commercial success was “substantial.”
REDUCE-IT trial showed cardiovascular outcome benefits.
Amarin met a long-felt, unmet medical need (LDL-neutral triglyceride-lowering agent).
Court’s Dismissal:
Judge Du held these were “insufficient to overcome the strength of the prior art.”
Judicial Error:
Secondary considerations—such as unexpected results, long-felt need, and commercial success—must be considered as part of the totality of the obviousness inquiry. However, courts must not discount these factors unless there is:
A strong nexus argument failure, or
Overwhelming prior art.
Given that the prior art (Mori/Kurabayashi) was misinterpreted or misleadingly presented, discounting these secondary indicia was premature and legally unsound.
Moreover, the REDUCE-IT trial was cited as demonstrating Vascepa’s unanticipated benefits beyond triglyceride reduction—suggesting the invention’s true impact was greater than assumed.
Summary Table: Core Judicial Errors
Area: Statistical Analysis
Error: Treated “significance vs. non-significance” as proof of difference
Impact on Judgment: Inflated EPA’s distinctiveness over DHA
Area: Evidentiary Integrity
Error: Relied on cropped data from Kurabayashi (Table 3)
Impact on Judgment: Misrepresented Apo-B/LDL benefit
Area: Expert Testimony
Error: Favored defense testimony built on flawed data
Impact on Judgment: Gave weight to incorrect conclusions
Area: Secondary Factors
Error: Undervalued commercial success and clinical impact
Impact on Judgment: Reduced balancing weight in §103 analysis
Final Evaluation:
The invalidation of Amarin’s Vascepa patents rested on a fragile framework of misapplied statistics, misrepresented data, and misweighted evidence.
None of these errors were minor. They materially influenced the legal determination of obviousness, a highly fact-dependent standard. The judgment therefore exhibits clear factual and procedural error, warranting scrutiny under Rule 60(b)(1) (mistake), 60(b)(3) (fraud), and 60(d)(3) (fraud on the court).
The court was not presented with complete and properly interpreted data, and relied on a falsified evidentiary record to reach its conclusion. These failures—factual, procedural, and analytic—represent a breakdown in judicial function and call into question the integrity of the resulting verdict.
It's the ratio, baby ... the EPA/AA ratio!
AI
