Pure Fish Oil Lowers CVD Risk Even If We Don’t Understand How

John M. Mandrola, MD


November 11, 2018

No doubt about it: icosapent ethyl—a purified fish oil supplement—reduced cardiac events. The stunningly positive primary endpoint slide from the REDUCE-IT trial[1] earned applause from a packed audience here at the American Heart Association (AHA) Scientific Sessions 2018 in Chicago.

In a large multicenter randomized controlled trial, patients with high cardiovascular risk and elevated triglycerides who took 4 g of a purified eicosapentaenoic acid ethyl ester (EPA) per day  vs a mineral oil placebo had a 4.5% lower risk of dying from cardiac causes or experiencing myocardial infarction, stroke, coronary revascularization, or unstable angina necessitating hospital admission.

Here’s how strong the result was: If you assume no difference in the drug relative to placebo, the chance these results would have been observed (eg, the P value) was 1 × 10 8.

Strengths of REDUCE-IT include that randomization resulted in well-matched groups; the 25% relative risk reduction is robust; the active treatment reduced all components of the composite endpoint (not just softer ones like coronary revascularization); and the findings showed a consistent effect in important subgroups (eg, sex, race, high vs low levels of triglycerides, statin intensity). Finally, the 25% relative risk reduction in REDUCE-IT comports with the JELIS trial,[2] a non–placebo-controlled trial of Japanese patients that found a 19% reduction in cardiac events with a lower daily dose of EPA.

The high-dose EPA did show a trend toward higher rates of overall bleeding, but the rate of life-threatening bleeding events was similar. An unexpected but definite 1.4% increased risk for atrial fibrillation occurred in the EPA group. 


By email, Ethan Weiss, MD, from the University of California San Francisco, highlighted two major concerns with REDUCE-IT. The first is whether the mineral oil in the placebo group caused harm, thereby making the active arm look better in comparison. The evidence for this comes from Table 4 of the supplement accompanying the publication in the New England Journal of Medicine, which showed significant increases in levels of non-high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), apolipoprotein B, and high-sensitivity C-reactive protein in the placebo arm.

His second concern was that active treatment resulted in only modest reductions in triglycerides. The authors discussed this in the paper, noting that the benefits of EPA were observed irrespective of the attained triglyceride level at 1 year (≥150 or <150 mg/dL). They proposed that EPA might work via other mechanisms.

Weiss concluded that if there are unclear mechanisms of benefit and the placebo is actually harmful, then how can we assess how “beneficial” the drug is?” He added: "I’m bummed."

Sanjay Kaul, MD, from Cedars Sinai Medical Center in Los Angeles, California, downplayed these concerns. He explained via email that the 7- to 10-mg/dL increase in LDL-C in the placebo arm will likely explain about 5% of the relative risk reduction. That leaves a 20% relative risk reduction, which is still a large treatment effect.

In a tweet following the presentation, Sekar Kathiresan, MD, from Harvard Medical School in Boston, Massachusetts, agreed with Kaul, saying that the differences in non-HDL-C and LDL-C between the active and placebo arm "do not compromise the integrity of the trial result."


Doctors begin their training by learning the biology of disease. This creates a frame in which we are drawn to treatments we can explain. If high levels of a substance are associated with a disease, a drug that lowers that substance should improve outcomes. Think statins.

Our embrace and reliance on such a physiology-based frame has resulted in tragic failures—antiarrhythmic drugs for premature ventricular contractions after myocardial infarction, for example.

This is why evidence-based medicine is so important. It doesn’t care how something works; it just measures if it works. And it applies here. REDUCE-IT convincingly shows that if you use 4 g of EPA in patients like those enrolled in the trial, you will lower the risk for adverse cardiac outcomes. Even with the caveats on the inertness of the placebo, a major treatment benefit remains.

You can dislike the first line of the data disclosure statement in which the authors answer, ‘No’ to the question, "Will the data collected for your study be made available to others?" You can be concerned that Amarin, the makers of icosapent ethyl, was responsible for data collection and analysis. And patients put on this drug will need to know there is a higher risk for atrial fibrillation.

But if you advocate for the use of evidence in the care of patients, and I do, the results of REDUCE-IT should be incorporated into practice.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.