ESC 2019 Two experts took opposite sides of the statement ‘fish oil: it’s all about triglycerides’ and presented evidence to back up their statements, including data of the REDUCE-IT trial.

Chairpersons: Marja-Riitta Taskinen (Helsinki, Finland) and Eva Prescott (Copenhagen, Denmark)

Fish oil: it's all about triglycerides – PRO

Professor Philippe Gabriel Steg (Paris, France)

Prof. Steg shared three lines of evidence for the statement ‘Fish oil: it’s all about the triglycerides’ and started with the epidemiological evidence. Long-term follow-up data of the BIP trial showed that high triglyceride (TG) levels are associated with all-cause mortality in patients with established coronary heart disease (CHD). Data from trials enrolling acute coronary syndrome (ACS) patients show similar findings; in the MIRACL with short-term follow-up and DAL OUTCOMES trial with longer follow-up, higher TG levels were associated with increased risk of CV events.

He continued by briefly presenting genetic evidence to back up his statement. Danish studies showed associations between genetic variants of genes encoding possible TG-lowering targets and reduced risk of ischemic vascular disease. Several study results are consistent with this finding. He highlighted one study by Ference, Kastelein and others that described that TG levels, reflected by a genetic LPL score, have an impact on outcomes. Also, this LPL score is additive to a LDLR score in predicting CV outcomes. Indeed, genetically determined lower levels of TG and LDL-c are both associated with lower risk of CV outcomes.

Third, the REDUCE-IT trial with icosapent ethyl, a highly purified form of the omega-3 fatty acid (FA) eicosapentaenoic acid (EPA), provides interventional trial evidence that lower TG levels result in decreased risk of CV outcomes. Steg said he expected that prof. Armitage, in defending the CON position, would list the many trials that have evaluated the effect of omega-3 FAs and did not show CV benefit. Also, a meta-analysis in 2018 (Aung T et al. in JAMA Cardiology) showed that omega-3 FAs failed to reduce CV risk. Steg said, however, that these trials examined low-doses of omega-3 FAs and did not select patients with high TG levels. Moreover, type of fish oils may impact outcomes.

In the REDUCE-IT trial, presented last year (2018), patients with elevated TG levels were selected, which is an important feature of this trial, according to Steg. Median TG levels were 216 mg/dL and 60% of patients had TG levels ≥200 mg/dL. Furthermore, patients had LDL-c levels between 40-100 mg/dL, were on statins, and had established CVD or had diabetes with at least one risk factor. After one year, icosapent ethyl lowered TG levels by ~20% compared to placebo and was associated with 25% reduction in the primary end point of CV death, myocardial infarction (MI), stroke, coronary revascularization and unstable angina; an impressive finding said Steg. The key secondary outcome of CV death, MI and stroke was reduced by 26% in those taking icosapent ethyl compared to placebo and prespecified hierarchical testing showed that all secondary endpoints were statistically significantly reduced in the icosapent ethyl group compared to the placebo group.

Steg proposed to apply the principle of Occam’s razor: the fewer hypotheses to explain an observation, the more likely a hypothesis is. In this case, icosapent ethyl lowers TG levels, in patients with elevated TG levels, and improves outcomes: thus, this drug must have something to do with TGs! Compared to other trials evaluating fibrates, niacin, and omega-3 FAs, patients in the REDUCE-IT trial had the highest TGs levels, biggest reduction in TGs and this was associated with greatest clinical benefit.

He continued by saying that prof. Armitage would probably present evidence that this benefit has nothing to do with TGs. Subgroups in the REDUCE-IT trial based on baseline TG levels (≥150 mg/dL or<150 mg/dL) showed no difference in outcomes in the icosapent ethyl group. Looking at subgroups based on achieved TG levels after one year (≥150 mg/dL or <150 mg/dL), also did not result in different outcomes. Steg had asked an expert in this field, prof. Nordestgaard, how to interpret these data. Nordestgaard had answered that it is very simple; all these patients have achieved TG levels that are still elevated, that is why you do not see a difference in treatment effect. Nevertheless, said Steg, when you look at tertiles of baseline TG levels a difference in treatment effect might become visible. Those with the highest baseline TG levels have a numerically lower HR with icosapent ethyl than those in the middle and lowest tertiles (HR: 0.68 vs. HR 0.80 and 0.79, respectively), but no significant interaction was found. The lower HR for those with higher TG baseline levels was both observed for first events and recurrent events. Therefore, Steg said, these data do not disprove the statement that it is about TGs. These patients all had elevated levels, namely ~150 mg/dL, while physiological levels are ~50 mg/dL.

Steg said that we do not know whether there are other potential mechanisms resulting in benefit of icosapent ethyl. ‘This will not be the first time we have a drug that gives benefit and we do not know the mechanism of action, such as CETP inhibitors and SGLT2 inhibitors’. But he would prefer to have it this way, rather than the other way around.

Side effects in the REDUCE-IT trial were increase in bleeding with icosapent ethyl. Steg thinks this is not accidental, as icosapent ethyl may affect platelet function. Therefore, this may be one of the mechanisms via which icosapent ethyl affects outcomes. He showed a slide with potential benefits of icosapent ethyl, which was quite extensive and benefits ranged from processes related to endothelial function, oxidative stress, inflammation, plaque growth and stability of plaques.

He pointed out an important observation with EPA in the MARINE and ANCHOR trials: as fasting TG levels lowered, EPA levels in plasma and erythrocytes increased. This observation suggests that presence of EPA in membranes of erythrocytes and in plasma may mediate some of the benefits.

Finally, Steg asked what the dyslipidemia guidelines recommended on this topic. New recommendations in the 2019 ESC/EAS dyslipidemia guidelines include that for high risk patients with TG levels between 135-500 mg/dL, EPA should be considered, in combination with statins.

More data on fish oil and CV outcomes in those with elevated TG levels are awaited; the STRENGTH and PROMINENT trial are currently ongoing.

Fish oil: it's all about triglycerides – CON

Professor Jane Armitage (Oxford, UK)

For many years, said prof. Armitage, observational studies have very consistently shown that higher fish consumption is associated with lower risk of CHD, and more particularly CV death. But as mentioned by Steg, a meta-analysis published in 2018 showed that low dose omega-3 FAs did not show CV benefit, nor did the ASCEND and VITAL trials (with 1 gram omega-3 FAs, a combination of EPA and docosahexaenoic acid (DHA). But, results of the REDUCE-IT trial have opened up the field again with substantial effects of 4 gram purified EPA. What then, could the mechanisms of benefit with icosapent ethyl be? Armitage stated that it is not all about TGs.

Looking at lipid levels in the REDUCE-IT trial with purified EPA treatment, reveals that there was a 20% reduction in TG levels, 13% reduction in non-HDL-c and 10% in apolipoprotein B. Using plots from the Emerging Risk Factors Collaboration (ERFC), she showed that the reduction in TG levels, observed in the REDUCE-IT trial, only lowered the HR for CHD outcomes a little bit, possibly accounting for 9% in risk reduction. She said that when comparing effect size observed in observational and interventional studies, typically only half of the effect observed in observational studies is achieved in interventional studies.

Another way to explore possible mechanisms of benefit observed in the REDUCE-IT trial, is by plotting the achieved level of non-HDL-c in REDUCE-IT and in statin-trials (using data of the CTT meta-analysis) against CV risk reduction. This graph suggests that about a third of the risk reduction in the REDUCE-IT trial can be explained by the reduction in non-HDL-c. So, what about the remaining two-third?

Armitage first looked into the chemistry of omega FAs. Omega-3 FAs, including EPA and DHA, are long-chain FAs and have a first double-bond at the third carbon, calculated from the methyl-end of the molecule. Omega-6 FAs are shorter and have a first double-bond at the sixth carbon. Omega-3 FAs and omega-6 FAs stimulate different metabolic pathways. Omega-6 FAs result in an increase in pro-inflammatory cytokines and pro-aggregatory effects, whereas omega-3 FAs result in anti-inflammatory effects and inhibition of platelet function and aggregation. These pathways compete with each other to some extent; when more omega-3 FAs are present, fewer omega-6 FAs are metabolized.

Other potential mechanisms by which omega-3 FAs may result in benefits are through their anti-thrombotic, anti-inflammatory, anti-arrhythmic and hypotensive effects. As Steg reminded us, there was a bleeding signal in REDUCE-IT (albeit non-significant); ~30% increase in bleeding in those who took icosapent ethyl. This bleeding risk is similar to that observed with aspirin, which has anti-thrombotic benefits. Therefore, risk reduction with icosapent ethyl may be partly mediated by anti-thrombotic effects.

It is also possible that risk reduction is mediated by anti-inflammatory effects, but it is difficult to demonstrate this in human studies. There is a large body of evidence in animal models, demonstrating that EPA reduces the level of anti-inflammatory cytokines, such as TNF- and IL-6. In addition, resolvins and protectins, which derive from EPA and DHA, have potent anti-inflammatory effects.

Anti-arrhythmic effects with use of omega-3 FAs have already been described a long time ago and might explain the observed reduction in cardiac death in observational studies. Early studies in dogs and rats quite convincingly showed that there are anti-arrhythmic effects. It is much harder to demonstrate this in RCTs, although some trials suggest there is a reduction in sudden cardiac death. REDUCE-IT did, however, show an increase in atrial fibrillation (AF), going against the evidence of an anti-arrhythmic effect. The ASCEND trial with low-dose omega-3 FA did not show a signal for AF. Nevertheless, a trend was seen for reduced coronary death and any cardiac death with omega-3 FA, but little was thought of this because of multiple comparisons.

Armitage then addressed the question whether fish oils can reduce blood pressure (BP). A meta-analysis of 110 trials suggests that there might be a small effect on BP, but there was evidence of publication bias, so it is difficult to assess the true effect. REDUCE-IT showed a decrease of 1.3 mmHg in SBP and 0.5 mmHg in DBP with icosapent ethyl, indicating that ~5% of benefits could be attributed to BP benefits.

She showed a graph that was a summary of trials, expressed as reduction in outcomes per gram of fish oil. Most of low-dose omega-3FA trials used a 1 gram capsule (containing 0.85 g of omega-3 FAs) and the ones that used a somewhat higher dose suggested greater benefit. She therefore argued that fish oils may have an effect , but the low-dose trials were underpowered to detect the small effects. She concluded that higher dose of fish oils might work, even in the absence of hypertriglyceridemia.

Fish oil: it's all about triglycerides – DISCUSSION

During the discussion, Steg said that he agreed with Armitage regarding that TGs are part of the explanation of benefits seen with omega-3 FAs; it is very likely that there are other mechanisms. He also agreed with the content on all slides of Armitage, except one: the slide with data of the ERFC that showed that, when adjusting for risk factors, TGs levels are no longer associated with CV outcomes. Steg said that this has been disproven: we now recognize that overadjustment takes place when adjusting for HDL-c in the association between TG levels and CV outcomes, as there is a direct relationship between TGs and HDL-c. We now know that HDL-c is not a causal factor for CVD and now we have data of an intervention trial that shows benefit with purified EPA. Armitage responded that she agreed and said that this is why she did not use the adjusted line in the graph. She said the epidemiologists had it wrong, that we always thought it was HDL-c that mattered. It remains an interesting conundrum that HDL-c is a good predictor of risk and yet does not appear to be causally related.

A question from the audience was how much fish is needed to produce one pill and how to deal with the huge environmental problem that is created by producing (purified) omega-3FAs. Steg answered that one way to get greater benefit and be environmentally respectful is to avoid manufacturing useless over-the-counter supplements that a lot people use on their own initiative and that result in no benefit. We should focus on manufacturing drugs that have shown to be efficacious in clinical trials. That would be a better use of the fish resource on the planet. Armitage said that last year when she presented the ASCEND trial, which was a negative study, it was one of the things she was relieved about. REDUCE-IT, however, has changed the picture and she said the need for fish is indeed a concern.

Read our summary of the REDUCE-IT trial