Discussion

In the present article we reviewed the evidence from randomized controlled trials and clinical trials involving the use of omega-3 fatty acids for at least six months until January 20th, 2012. In total, of 452 initial potential items, 21 studies matched the inclusion and exclusion criteria and ranked 3/5 or higher on the Jadad Scale⁽Reference Dehmer, Popma and van den Berg^8–Reference Nodari, Triggiani and Campia²⁸⁾ (Table 1). As a primary outcome, we observed that the use of omega-3 results in a fall of approximately 10 % in cardiovascular events (OR 0·90; [0·85–0·96]); Z = 3·28, p = 0·001). We also observed lower frequency of cardiac death (OR 0·91; [0·83–0·99]; Z for overall effect 2·13; p = 0·03), and coronary events (OR of 0·82 (0·75–0·90), Z = 4·17; p < 1 × 10^− 4). Although we did not find differences in total mortality, there was a trend towards a lower mortality in omega-3 users (OR of 0·95 [0·89–1·02], Z = 1·45, p = 0·15). We did not stratify the populations according to risk, and included all participants in the same equation, which implies that, in fact, a greater effect of omega-3 could have been obtained if the population is reduced to secondary prevention only. However, we wanted to include the as much information available from the current data as possible.

Table 1 Summary of the articles included in this review

Table1: Main Outcomes, doses, follow-up and main results of the articles included in this review.

* Approx content in EPA/DHA is 0·80 g for each 1 g of capsule content. Where there are notable differences to this fact in the present list, the number of capsules and/or total omega-3 amount is shown.

† Total months, or average mean follow-up when available. Drugs appearing in last column are grouped as they appear in the original reports.

‡ All patients were advised to follow a National Cholesterol Education Program Step I type diet.

§ Diet counseling was aimed at increasing the use of vegetable oils and margarines, vegetables, fruit and fish, and to decrease the use of meat and fat from animal sources. Special oil and margarines containing rapeseed were provided for these participants. 1) To eat at least two portions of oily fish each week, or to take up to 3 g of fish oil (‘Maxepa’) as a partial or total substitute. 2) To eat four to five portions of fruit and vegetables (apart from potatoes) and drink at least one glass of natural orange juice daily, and also increase the intake of oats, so as to obtain a higher intake of vitamin C and at least 8 g of soluble fibre from all sources every day. 3) A combination of both these forms of advice. 4) ‘Sensible eating’- non-specific advice that did not include either of the above interventions. AAS: acetylsalicylic acid. ACE-I: Angiotensin-converting enzyme inhibitor. ARB: Angiotensin II receptor blocker. ALD: aldosterone receptor blocker. AMI: Acute myocardial infarction. BB: beta-blocker. CCB: Calcium channel blockers ICD: implantable cardioverter-defibrillator. OAC: Oral Anticoagulants. T1AA : Type 1 antiarrythmic agents; VT: ventricular tachycardia, VF: ventricular fibrillation.

We excluded from the present study a recent study by Kromhout et al, which evaluated the effect of marine omega-3, ALA or placebo on the incidence of cardiovascular events after AMI⁽Reference Kromhout, Giltay and Geleijnse²⁹⁾. The study was well designed and conducted, but used ALA as an active arm, which we excluded from this study. Furthermore, the methodology of the statistical approximation made it impossible to compare directly the effects of marine omega-3 versus placebo, when ALA is excluded from the analysis.

Most of the studies analyzed in the present work showed an intriguing uniformity in the direction of their results. The divergent results of one study (Burr et al, 2003)⁽Reference Burr, Ashfield-Watt and Dunstan¹⁹⁾ are particularly interesting, as they served as a mirror image for nearly all the outcomes. In fact, all the outcomes of all the studies showed a low rate of inter-assay heterogeneity I² (below 25 %), when this study was excluded. Actually, we performed an alternative examination of the results excluding this study, and as well as increasing all the significant findings on cardiovascular events, cardiac death and coronary events, we found a significant change in the total mortality outcome, by lowering the OR to 0·93, 0·87–0·99, Z = 2·18; p = 0·03). It is not easy to work out the explanation for this fact. Some authors have recently reported methodological issues during this study⁽Reference Lavie, Milani and Mehra^3, Reference von Schacky and Harris^30–Reference Saravanan, Davidson and Schmidt³³⁾, such as a transient cessation of the study for a year due to lack of funds, or the fact that participants were not uniformly randomized to omega-3 supplements, and that these were only given to the participants who entered the study after a one-year break, which could have created a bias. Some of the authors who recently conducted meta-analyses even excluded the Burr study⁽Reference Lavie, Milani and Mehra^3, Reference von Schacky and Harris^30, Reference Marik and Varon³²⁾. We agree that such issues place limitations on the article, but do not go as far as to discredit it: hence, we have included it. We excluded just one study on methodological grounds, on the light of previous reports⁽Reference Singh, Niaz and Sharma^34–Reference White³⁶⁾.

All the results coming from reviews and meta-analyses (and so the ones coming from this work) have to be accepted with caution, due to the fact that the number of studies included are hardly ever designed with the same type of patients, under the same conditions, or with the same intervention. The present study offers results from studies coming from different sets of patients (primary, secondary prevention, spanning over a large range of ages, some of them only accepting men, others only involving persons about to undergo revascularization…), and with remarkable differences in the treatment doses (ranging from 0·3 to 6·9 g) (Table 1). Thus, pooling all the participants together may make it difficult to discriminate effects in subgroups. Furthermore, the size of the study is a critical factor when evaluating the results of a systematic review or meta-analysis. The fact that the two GISSI studies are the largest ever conducted (comprising about 20 000 patients) mean that the present article is clearly influenced by the great difference in size, as are most of the other studies included. Another fact to be noted is that the differences in the medication given to the patients in the different groups may have conditioned the results of the meta-analysis. For example, the GISSI prevenzione study was performed in 1999, when some of the current protocols for the treatment of coronary disease had not been initiated. The Rauch study, on the other hand, used state-of-the-art treatment (in 2010) alongside the omega-3 allocation, and failed to find differences in the outcomes depending on omega-3. This may suggest that the improvements in the clinical outcomes provided by current medication could minimize the positive effects of omega-3 (Table 1). Nevertheless, the other recent studies (from 2007 on) showed a similar trend to the GISSI-prevenzione study, despite including most of the medication currently used.

Another important hypothetical point to be considered when dealing with the effects of a certain nutrient (or a substance that may be ingested, such as omega-3 fatty acids) in health or disease is the possibility that the effects of the ‘controlled’ intake of the nutrient (in this case, omega-3) may be influenced by the background diet, and the ‘uncontrolled’ additional intake made by the participants. Moreover, the average intake of the given nutrient in the different populations studied may also influence the results. The effects of omega-3 in populations where its mean consumption is low may be higher than in those populations where the dietary habits include the use of blue fish as a habitual food. For example, even ‘control’ subjects, randomized to not receive supplements of omega-3 in populations like Japan (where usual diet is up to 15 times higher than in western countries⁽Reference Iso, Kobayashi and Ishihara³⁷⁾), probably have good average omega-3 intake, when compared to populations with western dietary models, and, hence, there may be an underestimation of the effects of omega-3 due to the fact that even ‘control’ subjects are receiving a good dose of omega-3 via diet. In this sense, omega-3 supplements for people who have a high dietary omega-3 intake fail to reduce triglycerides or modify other plasma lipids⁽Reference Watanabe, Watanabe and Kumagai³⁸⁾. In contrast, populations with low average omega-3 consumption would be theoretically better able to study the effects of the ‘regulated’ intake of omega-3, given that the consumption in normal subjects would be more sporadic. However, these other models could, on the other hand, influence the results of the quantification of the effects of omega-3. In populations with a very low omega-3 intake, it could occur that even with the supplementation of omega-3 (either by dietary or pharmacological means), omega-3 concentration in plasma would not reach a certain level, or, furthermore, that the effects of a simple intervention with a certain amount of omega-3 would not be enough to counteract the deleterious effects of the global diet. Although the information provided in many of the studies included in the present review do not allow us to estimate the omega-3 intake, we included the dietary advice given in these studies in Table 1. The dietary advice given (when available) does not seem to influence the results of the intervention with omega-3.

Another possible concern when evaluating this particular topic is the fact that the generalized opinion that omega-3 is good for health (and in particular cardiovascular disease) may be influencing the intake of omega-3 (in the form of fish) in the general population, and, more accurately, in those persons that suffer from coronary or other cardiovascular disease. This way, even ‘control’ subjects in randomized control studies may have a higher fish consumption than matched healthy persons in their respective populations, and subsequently, a lower difference between the ‘active’ and the ‘placebo’ group would show that even the ‘placebo’ group are receiving (at least partially) the beneficial effects of their intake of fish.

Adherence may be an important factor when analyzing the results, and, in modern studies, persons at secondary prevention receive a large number of medications, especially in the immediate period following the cardiac event. For example, in the OMEGA trial⁽Reference Rauch, Schiele and Schneider²⁷⁾, which failed to show any effect of 1 g/d of omega-3 in clinical outcomes, more than the 80 % of the patients were taking over 5 different drugs a day, apart from the omega capsules. In a recent sub-study from the JELIS study, adherence was an important determinant of outcomes⁽Reference Origasa, Yokoyama and Matsuzaki³⁹⁾. The rate of adherence was in itself a determinant of the clinical outcomes, even within the treatment arm, and the poor responders had an 80 % possibility of receiving the omega-3 pills. Nevertheless, the OMEGA investigators reported a compliance of 70 % or more in 93 % of the omega-3 group. Regarding this last study, it is also worth noting also that data from the randomization table shown in the article allow us to infer an OR for the combined risk of stroke, previous AMI, previous revascularization, previous by-pass, or moderate (35 %–44 %) and severe ( < 35 %) reduction of the ejection fraction of 1·12 (1·04–1·20) for the active treatment (omega-3) versus control. In other words, the patients allocated to omega-3 were persons at higher risk (12 %) for the combined items of these conditions⁽Reference Rauch, Schiele and Schneider²⁷⁾, which may have influenced the ‘negative’ results.

When evaluating the effects of omega-3 fatty acids on clinical outcomes, we are very limited by the time of exposure, the statistical power and the dosage used in the studies. In fact, most of the studies ‘designed’ to discriminate clinical endpoints in a short time (less than six months) fail to find differences - probably because the effects would not become evident until much later. In addition, some of the studies used doses lower than 0·5 g EPA/DHA, which may not allow the trigger plasma concentration required for omega-3 effects to be reached. In other words, some of the beneficial effects of omega-3 may require a prolonged period (up to years) at a certain dose. It has been suggested that although antithrombosis effects may appear with low doses and a low time of exposure (weeks), others, like blood pressure regulation, lowering of triglycerides or lowering of heart rate, could take anything up to a few years⁽Reference Mozaffarian⁴⁰⁾. Another important factor to be pointed is that some of recent studies calculated their sample size according to expected event rates that were not achieved. For example, in the recent OMEGA trial investigators had a statistical power of 19 % to detect differences of risk reduction of 25 %. As the risk reduction shown in the omega-3 intake for the clinical outcomes covered in the present study ranges from 5–18 %, the probability of this study discriminating such differences was extremely low⁽Reference Rauch, Schiele and Schneider²⁷⁾. As an example of the possible infra-treatment and/or low follow-up, in the GISSI-HF trial, allocation to marine omega-3 (1 g/d) required a follow-up of at least 3·9 years to find differences in total mortality or cardiovascular events⁽Reference Gissi, Tavazzi and Maggioni²⁴⁾. In our opinion, only the evaluation of clinical events from long-term exposure to omega-3 fatty acids (up to ten years) with amounts of at least 1 g EPA/DHA would answer this question.

Another issue to take into account when evaluating the apparent disparity of results between studies is population genetics. It is true that genetic background plays an important role in many cardiovascular risk factors, like lipid metabolism, both in fasting and postprandial states⁽Reference Perez-Martinez, Delgado-Lista and Perez-Jimenez^41, Reference Teslovich, Musunuru and Smith⁴²⁾. It is precisely in the modulation of the lipid profile caused by the intake of omega-3 that leads some authors to justify the favorable effects of these products. Recently, Pishva et al showed how a mutation in the FABP gene may account for up to 70 % of the effect of the omega-3 fatty acids on triglyceride levels⁽Reference Pishva, Mahboob and Mehdipour⁴³⁾. Furthermore, Olano-Martin found that the carriers of another gene mutation at the ApoE locus show an increase in LDL levels when exposed to a high DHA intake⁽Reference Olano-Martin, Anil and Caslake⁴⁴⁾. This gene-diet interaction was also previously published by Caslake⁽Reference Caslake, Miles and Kofler⁴⁵⁾. Thus, a different prevalence of gene variations in the populations submitted to each study may lead to different results when evaluating outcomes related to the lipid effects of omega-3.

Although this review is limited to cardiovascular events in randomized controlled trials and clinical trials, we also wanted to remind the reader that many reviews and meta-analyses have included evidence on the efficacy of omega-3 coming from observational and cohort studies, and most of them report a much closer inverse relationship between omega-3 consumption (whether from dietary intake or in the form of supplements) and cardiovascular disease⁽Reference Mente, de Koning and Shannon^4, Reference Saravanan, Davidson and Schmidt^33, Reference Mozaffarian and Rimm⁴⁶⁾.

The exact mechanisms by which omega-3 performs its functions are still into debate, but the main mechanisms proposed are plaque stabilization⁽Reference Thies, Garry and Yaqoob⁴⁷⁾, lipid profile⁽Reference Harris, Miller and Tighe⁴⁸⁾, anti-inflammatory⁽Reference Serhan, Chiang and Van Dyke⁴⁹⁾, blood pressure⁽Reference Morris, Sacks and Rosner⁵⁰⁾, heart failure⁽Reference Gissi, Tavazzi and Maggioni^24, Reference Yamagishi, Nettleton and Folsom⁵¹⁾, or anti- arrhythmic properties⁽Reference Leaf, Albert and Josephson^20, Reference Mozaffarian, Psaty and Rimm^52, Reference Gillet, Roger and Bougnoux⁵³⁾. It is precisely over this last effect where major controversy exists today. Initial data from the GISSI-Prevenzione found that most of the decrease in mortality associated with the intake of omega-3 could be due to a decrease in the frequency of sudden death, which leads to the theory that these fatty acids could have anti-arrhythmic properties⁽¹⁵⁾. Nevertheless, the evolution of the study on this topic has shed more controversy than light on the issue. Although it seems, in the view of some observational and cohort studies, that omega-3 may have the effect of lowering the appearance of atrial fibrillation, especially in the type linked to atrial remodelling associated with age⁽Reference Mozaffarian, Psaty and Rimm⁵²⁾ (but not in the type associated with cardiac surgery)⁽Reference Saravanan, Bridgewater and West⁵⁴⁾, the impact of omega-3 in ventricular arrhythmias is much more debatable. Both pro- and anti-arrhytmic effects of these fatty acids have been reported⁽Reference Leaf, Albert and Josephson^20, Reference Raitt, Connor and Morris^21, Reference Brouwer, Zock and Camm^22, Reference Brouwer, Raitt and Dullemeijer⁵⁵⁾ and, perhaps, both really exist. From the different in vitro and in vivo studies, it has been established that omega-3 shortens the action's potential duration and slows down impulse conduction⁽Reference Verkerk, van Ginneken and Berecki^56, Reference Coronel, Wilms-Schopman and Den Ruijter⁵⁷⁾. These actions may reduce the appearance of a triggered activity and subsequent ventricular fibrillation to those usually present in the recent post-infarct state, and, on the other hand, facilitate the appearance of re-entry induced ventricular fibrillation, such as those provoked by other clinical or anatomical causes⁽Reference Saravanan, Davidson and Schmidt³³⁾.

Regarding the safety of using omega-3, a recent work revised the incidence of side effects. The only adverse effect that reached statistical significance were gastrointestinal side effects (mainly nausea)⁽Reference Filion, El Khoury and Bielinski⁶⁾, which occur more frequently at 4 g/d or higher doses, in up to 20 % of the patients (or 4 % of the patients below 3 g/d)⁽Reference Wang, Harris and Chung⁵⁸⁾, which may be accompanied by a fishy taste when belching and may be minimized by taking the supplements with meals or freezing the capsule⁽Reference Chalupka⁵⁹⁾. The most severe potential adverse effect studied regarding omega-3 is the potential risk of bleeding, by their effects on platelet metabolism⁽Reference Knapp, Reilly and Alessandrini^60, Reference Kristensen, Iversen and Schmidt⁶¹⁾. However, studies which evaluated the safety of using omega-3 up to 4 g/d in takers of anti-platelets and anticoagulants, reported no increased risk on minor or major bleeding events⁽Reference Eritsland, Arnesen and Seljeflot^62, Reference Watson, Joy and Nkonde⁶³⁾, which has been also confirmed in a recent meta-analysis⁽Reference Filion, El Khoury and Bielinski⁶⁾. Another potential side effect of the consumption of omega-3 from a dietary source (fish) is the mercury intake that may be involved, especially in big fishes (blue fin tuna, shark, tilefish, swordfish, or king mackerel). Although mercury intake has been related to cardiovascular disease, this relationship is still not yet clearly proven, especially in the amounts that can be derived from eating fish, and to date, the net health benefits of overall fish consumption in adults are clear (for a recent review, please see⁽Reference Park and Mozaffarian⁶⁴⁾).

In vitro and animal studies suggest that PUFA are more prone to become oxidized than other sources of fatty acids, such as MUFA⁽Reference Fremont, Gozzelino and Franchi⁶⁵⁾. Furthermore, it has been shown that diets rich in omega-3 fatty acids cause an increased susceptibility of lipids to oxidation⁽Reference Hau, Smelt and Bindels⁶⁶⁾. Nevertheless, other studies have not found such results⁽Reference Bonanome, Biasia and De Luca^68–Reference Higdon, Du and Lee⁷⁰⁾, and a recent study did not find any differences in the different markers of oxidative stress in over 400 patients with metabolic syndrome who consumed four dietary models for 12 weeks, one of which enriched in 1·24 g/d of long chain omega-3 fatty acids⁽Reference Petersson, Riserus and McMonagle⁷¹⁾. In conclusion, although omega-3 fatty acids have been shown to increase the susceptibility to oxidation of circulating lipids, studies have not proved that they are linked to an increase in markers of oxidative stress. Further studies are needed to clarify this question.

In conclusion, we have reviewed the effects of marine omega-3 fatty acids on cardiovascular events from randomized controlled trials and clinical trials. The accumulated evidence, as of January, 2012, indicates that marine omega-3, when administered as food or in supplements for at least 6 months, reduces cardiovascular events by 10 %, cardiac death by 9 % and coronary events by 18 %, while showing a trend for a lower total mortality (5 %, p = 0·13). These results are based in the evaluation of studies that included mainly persons with high cardiovascular risk, and in studies which are highly heterogenic in the dose administered, although there is no evidence of dose-dependent protection. Our results, along with the existing evidence on the myriad of physiological effects of omega-3 on human health (coagulation, heart rate, heart rhythm, blood lipids, etc), reinforce the AHA recommendations for the intake of omega-3 in the prevention of cardiovascular disease, especially in persons with high cardiovascular risk and secondary prevention.