Are Fish Oils on the Hook for AF Risk?

Home / Clinical Practice / Are Fish Oils on the Hook for AF Risk?

Are Fish Oils on the Hook for AF Risk?

John M. Mandrola, MD

August 10, 2023

Questions about omega-3 fatty acid supplements come up often in the atrial fibrillation (AF) clinic.

The story begins with the simple observation that populations who eat lots of oily fish have fewer coronary events. This correlation provoked great interest in concentrating fish oils in pill form and studying their use of to promote health.

A small post hoc study from a group in Norway stimulated me to review what we do and don’t know about fish pills, as I call them in clinic.

OMENI Secondary Analysis

Peder Myhre, MD, and colleagues recently published a secondary analysis of the OMENI trial looking at both the risk and possible causes of AF in the omega-3 group.

The OMENI trial randomly assigned slightly more than 1000 older patients (mean age, 75 years) post–myocardial infarction (MI) to either 1.8 g/d of fish oil supplements vs placebo for 2 years. The supplements comprised 930 mg of eicosapentaenoic acid (EPA) and 660 mg of docosahexaenoic acid (DHA). The main trial reported no difference in a composite primary endpoint of MI, revascularization, stroke, death, or hospitalization for heart failure.

The secondary analysis explored the 75% of patients in the main trial who had no history of AF. It looked at how many in each group developed either true clinical AF or what the authors called micro-AF, defined as short bursts of irregular atrial activity lasting seconds.

The sub-analysis had three main findings: Patients in the supplement arm had a 90% higher rate of AF or micro-AF compared with patients on placebo, EPA had the strongest effect on the association, and there was a graded risk for AF with increasing serum EPA levels.  

The authors raised the possibility that more micro-AF might be a possible mediator of AF risk.

Trials of Low-Dose EPA and DHA

First, the low-dose trials. In the ASCEND trial from 2018, more than 15,000 patients with diabetes were randomly assigned to either 1 g of omega-3 fatty acids (460-mg EPA and 380-mg DHA) or mineral oil.

The trial was neutral. After 7.4 years, the primary endpoint of MI, stroke, transient ischemic attack, or cardiovascular (CV) death occurred in 8.9% of the supplement group vs 9.2% of the placebo arm. The incidence of AF was higher in the omega-3 group but did not reach statistical significance. (2.1% vs 1.7% for placebo; hazard ratio [HR], 1.23; 95% CI, 0.98-1.54).

Another neutral CV trial, VITAL, specifically studied the effects of marine omega-3 pills (460-mg EPA and 380-mg DHA) in older adults without heart disease, cancer, or AF. After slightly more than 5 years, AF occurred at a similar rate in the active arm and placebo arms (3.7% vs 3.4% for placebo; HR, 1.09; 95% CI, 0.96-1.24; P = .19)

Trials of Very High-Dose Marine Omega-3s

Next came trials of higher doses in higher-risk populations.

In 2020, JAMA published the STRENGTH trial, which compared 4 g/d of a carboxylic acid formulation of EPA and DHA with a corn oil placebo in more than 13,000 patients who either had established atherosclerotic CV disease (ASCVD) or were at high risk for ASCVD.

The trial was terminated early because of futility and a signal of increased AF risk in the supplement arm.

Nearly the same number of patients in the supplement vs placebo arm experienced a primary composite endpoint of major adverse cardiac events: 12.0% vs 12.2%, respectively.

AF was a tertiary endpoint in this trial. An increase in investigator-reported new-onset AF was observed in the omega-3 group: 2.2% vs 1.3% for corn oil (HR, 1.69; 95% CI, 1.29-2.21; nominalP < .001).

The REDUCE-IT trial randomly assigned more than 8000 patients who had ASCVD or diabetes and high ASCVD risk and elevated triglyceride levels to either 4 g of icosapent ethyl daily, a concentrated form of EPA, or a mineral oil placebo.

After nearly 5 years, there was a 4.8% absolute risk reduction in the primary endpoint of CV death, MI, stroke, revascularization, or unstable angina with icosapent ethyl. An increase in atherogenic biomarkers in the mineral oil placebo complicated interpretation of this trial.

Hospitalization for AF or flutter occurred in 3.1% of the active arm vs 2.1% of the mineral oil group (P = .004).

Meta-Analysis of Marine Omega-3 Supplement Trials

In 2021, Baris Gencer and colleagues performed a meta-analysis of these five trials plus 2 more (GISSI-HF and RP) looking specifically at risk for AF. Their final analysis included more than 81,000 patients followed for nearly 5 years.

Omega 3 fatty acid supplements associated with a 25% increase in the risk for AF (HR, 1.25; 95% CI, 1.07-1.46=.013). Exploring further, they noted a dose-dependent relationship. Most of the increased risk occurred in trials that tested > 1 g/d.

Summary

When faced with surprise findings, I like to think things through.

First about plausibility. Omega-3 fatty acids clearly exert electrophysiologic effects on cardiac cells, an increase in AF risk is plausible. The exact underlying mechanism may be unknown, but exact mechanisms are less important than actual clinical effects (see sodium-glucose cotransporter 2 inhibitors).

What about causality? Factors supporting causality include plausibility, consistency of increased AF risk in multiple studies, and a dose-response relationship.

I see multiple clinical implications of this observation.

The first is the power of the randomized trial to inform practice. If we relied only on observational evidence, we might have assumed that since high fish consumption in populations associated with lower rates of cardiac events, fish oil supplementation would also reduce cardiac events. Other than the outlier trial, REDUCE-IT, with its mineral oil placebo, the preponderance of the randomized controlled trial evidence does not support fish oils for the reduction of CV events.

Randomized controlled trials also exposed the AF risk. This would have been difficult to sort out in nonrandom observational studies.

Another underappreciated lesson is the notion that drugs, including supplements, can have off-target effects.

Consider the case of statin drugs. It is widely assumed that statins reduce cardiac events by lowering low-density lipoprotein cholesterol (LDL-C). Yet, statins became a mainstay not because of LDL-C lowering but because multiple trials found that this class of drugs reduced cardiac events without increasing adverse effects.

Omega-3 fatty acids reduce triglyceride levels, but this is not enough to adopt use of these pills. The lack of consistent reduction in CV events and the off-target signal of AF risk argue against routine use of fish-oil pills.

I will close with uncertainty. Though there is plausibility and multiple reasons to infer causality of marine omega-3s in increasing AF risk, the effect size remains unknown.

In an editorial accompanying the recent meta-analysis, epidemiologist Michelle Samuel, MPH, PhD, and electrophysiologist Stanley Nattel, MD, cautioned readers on a technical but important point. It concerns the matter of competing risks, such as death, in the analysis of AF risk, meaning that patients who died may have developed AF had they lived. They provide a detailed explanation in the open access article, but the take-home is that the exact effect size is difficult to quantify without patient-level original data.

No matter, I find the signal of increased AF risk an important one to use at the bedside.

Intermittent AF has an unpredictable natural history. It often resolves as mysteriously as it arises. When patients take fish-oil supplements, I cite these studies, note the lack of CV protection, then I recommend stopping the pills.

This allows for one of the most important interventions in AF care: time.

John Mandrola practices cardiac electrophysiology in Louisville Kentucky and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence.

Leave a Reply

Your email address will not be published.