A debate at ESC Congress: Benefit from omega-3 fatty acids?
Chairpersons: Prof. Magnus Back (Stockholm, Sweden) and prof. Ulrich Laufs (Leipzig, Germany)
Positioning of the topic
Magnus Back (Stockholm, Sweden)
This discussion was focused on the most recent studies on omega-3 supplementation with contradictory results; the REDUCE-IT trial and the STRENGTH trial. Questions have been raised on optimal dosing, formulation of omega-3 fatty acids (EPA, DHA or combination), and choice of placebo. Furthermore, which patients may benefit from omega-3 FAs, is consumption of fish sufficient, and does the response vary based on genetic profile? And the key question: What is the mechanism of action? Two protagonists tried to answer these questions in this debate.
Patients benefit from omega-3 FAs– PRO
Prof. Christie Ballantyne (Houston, TX, USA)
Ballantyne started by explaining the process of science. First, a specific hypothesis is generated. Then, a trial is conduced to test the hypothesis. Third, the data are analyzed, and strengths and weaknesses of the initial study are reviewed. Finally, a second trial is designed and conducted based on the strength and weaknesses of the first study.
In this case, the initial study was the JELIS trial, which was designed to test the hypothesis that treatment with highly purified EPA ethyl ester together with lipid lowering therapies is more effective than treatment without EPA in reducing major coronary events. JELIS was a prospective randomized open-label blinded endpoint evaluation (PROBE) trial. Patients in this trial received low-dose statin and the trial was conducted in Japan. These three points raised some concern with regard to the design of the study.
In the JELIS trial, addition of EPA (2 gram per day) resulted in significant reduction of major CHD events by 19% when compared to placebo, without impressive changes in lipids. Benefit on major CHD events was larger in those with high triglycerides and low HDL-c and appeared to be related to changes in EPA. The trial did not receive much attention at the time, because a lot was happening in the area of LDL-c lowering therapies.
After EPA was evaluated in another small trial (ANCHOR trial), a larger outcome trial was designed. This trial was designed to address weaknesses in the JELIS trial, so the REDUCE-IT was a multinational, randomized, double-blind, placebo controlled trial, with high-efficacy statin therapy as background therapy. Also, baseline EPA of included patients was lower, and this was a population of mostly secondary prevention patients, as this was the population that benefited most in the JELIS trial. Furthermore, enrolled patients had elevated triglycerides at baseline. In REDUCE-IT, treatment with EPA (4 gram per day) reduced the primary endpoint by 25% compared to placebo. To achieve power, it had a longer duration of follow-up (median of 4.9 years). In both REDUCE-IT and JELIS, the greatest reduction was observed in those with the highest levels of EPA.
The STRENGTH trial tested another hypothesis: omega-3 carboxylic acids (a mixture of DHA and EPA) was evaluated in a similar patient population (high risk with elevated triglycerides), but with more primary prevention patients. The trial was stopped early for futility. There was no difference in the primary endpoint between the omega-3 CA group and the placebo group.
Differences between REDUCE-IT and STRENGTH included: a higher proportion of secondary prevention patients in the REDUCE-IT trial, a higher proportion of patients on high-intensity statin in STRENGTH, higher median of triglycerides in STRENGTH, and larger reductions in apoB and hs-CRP in REDUCE-IT. Other differences between the trials included the active treatment, placebo (mineral oil in REDUCE-IT vs. corn oil in STRENGTH), follow-up (4.9 years in REDUCE-IT vs. 3.5 years in STRENGTH), endpoint (slightly different), event rates (higher in REDUCE-IT), EPA levels (higher in the REDUCE-IT).
So, Ballantyne then asked, how can the differences in outcomes be explained? Not by the reduction in triglycerides, that is clear. Calculations (during evaluation by the FDA) of the effect of mineral oil showed that this had a minimal effect on risk reduction. Ballantyne continued and noted that icosapent ethyl and omega-3 CA are different drugs with possible different antithrombotic and anti-inflammatory effects.
He showed the molecular structures of EPA, DHA and DPA. Higher dose of EPA results in higher levels of DPA than with DHA. Also, EPA and DHA have different biological effects. EPA results in E-Resolvins and DHA in D-Resolvins, with differences in oxidations. Furthermore, EPA and DHA have different effects on membrane elasticity, particularly in the presence of high cholesterol.
He made a comparison with steroid sex hormones. There are small difference in the molecular structure of testosterone, estradiol and cholesterol. But would a trial that examines the effects of a mixture of testosterone plus estradiol test the same hypothesis as a trial of only estradiol? ‘No’, he said. So, does a trial that examines the effects of mixture of DHA plus EPA test the same hypothesis as a study testing pure EPA? The answer here is also ‘no’.
Ballantyne concluded that the mechanisms of EPA is most likely related to biological effect on vascular biology including resolution of inflammation.
Patients benefit from omega-3 FAs – CON
Prof. Michael Lincoff (Cleveland, OH, USA)
Lincoff, in his rebuttal, showed much of the same data, but with a different point of view. He said, we are not just talking about two trials, we are talking about the entire field of fish oil consumption and outcomes.
The initial suggestion of benefit came from observational studies with natural fish oil consumption. A subsequent randomized trial of dietary fish or supplemental fish oil did not show benefit of either (DART-2 trial). After that, large trials have been conducted with different formulations of fish oil in different populations (ORIGIN, VITAL, ASCEND), which all showed no benefit. A meta-analysis of >10 trials in >70.000 patients showed no effect on CV outcomes. This was before the REDUCE-IT, which did show CV benefit with purified EPA.
In the STRENGTH trial, free fatty acids rather than ethyl esters were used. According to Lincoff, it was anticipated that this novel formulation allowed for distinct bioavailability advantages, because omega-3 carboxylic acid (CA) is freely absorbed and does not require hydrolysis by pancreatic lipase. It can therefore be taken one-daily independent of meals.
Subsequently, Lincoff presented the design and results of the STRENGTH trial. There was no effect of omega-3 CA on the primary endpoint. Although the trial was shorter than REDUCE-IT, there was no hint of divergence of the curves at any point, said Lincoff. And there was no relationship between achieved EPA levels and outcomes.
So, why was STRENGTH neutral but REDUCE-IT favorable, he asked. It is not likely that this is caused by high EPA levels, because EPA levels were also substantially increased in STRENGTH. It is also not likely that DHA is toxic, because there are no data to support this and in biological models it even appears to be beneficial. It is also not likely that this can be explained by the trial population. Although REDUCE-IT enrolled more patients with established CVD, there was no difference in subgroup analyses. Then, Lincoff mentioned the use of mineral oil in REDUCE-IT, which has become a controversial point and according to Lincoff this topic has been minimized. The corn oil used in STRENGHT has no substantial effects, but the mineral oil used in REDUCE-IT resulted in a 10% increase in LDL-c, 8% increase in apoB and 32% increase in hsCRP. When data of STRENGHT and REDUCE-IT were combined using the same endpoint (Lincoff recognized the limitations of this analysis), the event rate was similar in the omega-3CA group, the corn oil group and the icosapent ethyl group. The mineral oil group in the REDUCE-IT trial had a higher event rate than the other groups, which might explain much of the benefit of icosapent ethyl.
The adverse effects of fish oil are not trivial, Lincoff said, with the most important one being atrial fibrillation (AF); a 70% increase was observed in STRENGTH. Also, REDUCE-IT and OMEMI showed an increased in AF.
There are two other supportive studies for REDUCE-IT; the imaging study EVAPORATE and JELIS.
EVAPORATE showed a benefit with icosapent ethyl. But that benefit seems odd, according to Lincoff. The volume of low attenuation plaque at follow-up in the two treatment arms was the same. It was the volume at baseline that was different and drove the benefit. Moreover, a 109% increase in volume of low attenuation plaque was seen in the placebo arm over a period of 18 months, which has never been observed in other imaging studies. In IVUS studies, this is on average 1-2%. These results suggest adverse effects of the mineral oil.
Then, the JELIS trial. Benefit was observed with EPA, but there were many limitations in this study and this trial does not reflect contemporary practice. Besides the open-label, there was a baseline LDL-c of 180 mg/dL and patients were only treated with low dose statins.
Prof. Chris Packard (Glasgow, UK) provided commentary on this discussion. According to him, the question really is, what is the mechanism of action that is pertaining to these two trials, particularly EPA vs. DHA?
He summarized the data in one slide and presented the options. The first is lipoprotein synthesis in the liver. Both EPA and DHA decrease VLDL production in the liver by increasing FA oxidation, reducing triglycerides. Triglyceride reduction was similar in REDUCE-IT and STRENGTH. And although some benefit may be due to TG lowering in REDUCE-IT, total risk reduction was not related to TG or changes in TG.
Second, fish oil has modest effect on other lipoproteins, such as remnants, LDL-c and HDL-c. But given the changes observed in the trials, it is difficult to assume that changes in lipoproteins are related to risk reduction in CV outcomes.
Third, processes as oxidation, inflammation, cell signaling and thrombosis may be involved. After many decades of research, nothing has come out that is a strong signal to explain the differences in findings, Packard said. There are, however, newer studies that provide insights. EPA may stabilize cholesterol membrane rafts and DHA does not. The transition from macrophage foam cells to necrotic inflammatory cells is membrane cholesterol-load dependent and may be altered in the presence of stabilized rafts by EPA, whereas this does not happen by DHA.
He then proposed to assume that it is impossible that both trials were conducted so badly that one gave the exactly wrong answer and the other exactly the right answer. The mineral oil may contribute a little bit to the benefit, but does not fully explain it. He then quoted Sherlock Holmes: “When you have eliminated the impossible, whatever remains, however improbable, must be the truth”.
Finally, he noted that there was large uniformity of risk reduction in all the subgroups of REDUCE-IT, which suggests the mechanism is not explained by inflammation of TG-lowering, but by a modification of the core atherogenic process.
During the discussion, Ulrich Laufs asked the discussants which trials they would like to design. Michael Lincoff answered he would like to see a trial in a high-risk population (primarily patients with established CVD), well-treated on maximally tolerated LDL-c-lowering therapies randomized to a neutral placebo or a purified EPA formulation. EPA plus DHA has been studied enough, there is little likelihood that a future trial with EPA plus DHA formulation would show benefit.
Ulrich Laufs then asked: “What are the most important experiments to be done?” Chris Packard said macrophage biology in individuals treated with EPA or DHA needs to be studied, in addition to thrombotic and anti-inflammatory effects. “We need to get to the bottom of this in order to understand how EPA works”, concluded Packard.