Future Clinical Research Directions on Omega-3 Fatty Acids and Cardiovascular Disease

I. Purpose

For more than a quarter century, clinical studies have provided information on possible health benefits of consuming polyunsaturated omega-3 fatty acids from either fish or plant sources. Results of these studies suggest that omega-3 fatty acid intake is associated with reduced risk of numerous diseases, including cardiovascular disease (CVD), stroke, cancer, immune disorders, asthma, and neurological disorders, although causality has not been determined. Omega-3 fatty acids are a group of bioactive food components found in fish and fish oils (eicosapentaenoic acid, EPA, and docosahexaenoic acid, DHA) and plant sources such as canola oil, walnuts, soybeans, and flaxseeds (alpha-linolenic acid, ALA). Recommendations for consuming fish and omega-3 fatty acids have come from the American Heart Association (AHA) Nutrition Committee (Kris-Etherton, 2001, 2003) and the National Cholesterol Education Program Adult Treatment Panel III (ATP III) (NHLBI, 2002). The AHA recommended that the general public consume fish at least twice per week and that individuals with coronary heart disease (CHD) consume EPA and DHA from fatty fish or supplements. ATP III stated that higher intake of omega-3 in the form of fatty fish or vegetable oil is an ?option? because the strength of evidence was only ?moderate? for coronary benefits. The 2000 Dietary Guidelines Advisory Committee did not make a recommendation for the consumption of omega-3 fatty acid sources.

In FY 2001 and FY 2002 appropriations hearings for the National Institutes of Health (NIH), the U.S. Congress encouraged NIH?s Office of Dietary Supplements (ODS) to ?begin preliminary work on a major assessment of the health benefits of omega-3 fatty acid consumption.? In 2002, the ODS commissioned the U.S. Agency for Healthcare Research and Quality (AHRQ) to provide evidence-based reports on omega-3 fatty acids in various health conditions and diseases using all available scientific literature. AHRQ subsequently contracted with three of its Evidence-based Practice Centers (EPCs)?Tufts-New England Medical Center (Tufts-NEMC), the University of Ottawa EPC, and the RAND Southern California EPC?to provide 11 evidence-based reports. The ODS and the NIH?s National Heart, Lung, and Blood Institute (NHLBI) collaborated to support the evidence-based reports on CVD and stroke.

This report summarizes a subsequent Working Group meeting on Future Clinical Research Directions on Omega-3 Fatty Acids and Cardiovascular Disease convened by ODS in collaboration with NHLBI. The Working Group was chaired by Dr. Scott Grundy, Professor at the University of Texas Southwestern Medical Center at Dallas, and consisted of members from public and private academic and medical institutions and from federal agencies. The charge to the Working Group was to review the evidence-based reports on omega-3 fatty acids and in relation to CVD, to CVD risk factors, and to markers of CVD, and to advise the NHLBI and ODS on future directions for research to fill gaps in knowledge on the effects of omega-3 fatty acids on CVD and its risk factors.

Specifically, the Working Group was charged to provide advice to the ODS and NHLBI on the following:

• Is there a need and sufficient evidence for a clinical trial of omega-3 fatty acid interventions and CVD outcomes?
• If a trial is warranted, what is the research question (population, intervention, outcome)?
• What design and methodological approaches should be used for a trial?
• What other research is needed?

II. Recommendations

A. Recommendations on Omega-3 Fatty Acid Trial

The Working Group concluded that the body of evidence is consistent with the hypothesis that intake of omega-3 fatty acids reduces CVD but that a definitive trial is needed. The potential magnitude of effect is 20% to 45% reduction in CVD events. Evidence comes from cohort studies conducted in a variety of population samples, including U.S. samples, and consists mostly of individuals without CVD. There have been few clinical trials conducted, which were generally in men who had experienced a myocardial infarction, and conducted in Europe and elsewhere, where background diets and medical care differ from the U.S. Much of the trial evidence is based on one large trial that had an open-label control group rather than a placebo control group. The benefit is not well explained through effects on CVD risk factors such as LDL-cholesterol and may work through an independent mechanism, probably through arrhythmias and sudden cardiac death. U.S. diets are low in EPA + DHA (approximately 0.1 g/d) and ALA (approximately 1.3 g/d) compared to doses used in trials (1g/d EPA + DHA or 3 g/d ALA).

The Working Group recommended that a clinical trial be conducted to test the effects of omega-3 fatty acid intake on CVD mortality with other CVD outcomes considered secondarily. A summary of the Working Group's recommendations regarding the research question and design approaches for a trial is as follows:

• The trial should include both primary and secondary prevention populations, although the exact mix of participants can be determined by conducting power calculations.
• Calculations should be performed to generate the sample size needed to obtain high statistical power for the primary outcome of CVD mortality.
• The intervention should consist of well-measured and highly controlled supplements rather than diet and have a placebo control.
• The trial should have three randomized arms: fish oil, ALA, and placebo.
• The trial should assess potential adverse effects, such as prostate cancer.
• Blood levels of omega-3 fatty acids should be measured to assess adherence, to determine the extent to which ALA can raise EPA and DHA levels, and to conduct secondary analyses, e.g., to assess the relationship between changes in blood levels of omega-3 fatty acids and CVD.
• Ancillary studies should be encouraged to evaluate mechanisms by which omega-3 fatty acids may reduce CVD and, if possible, to examine dose-response relationships.
• The trial should be a large simple trial.

B. Recommendations on Other Research

Because of research conducted thus far that yielded potentially conflicting or unexpected results, several areas of research were identified to complement a large trial and fill gaps in knowledge to better understand the effects of omega-3 fatty acids on CVD, CVD risk factors, and markers of CVD. Data may be obtained from observational or intervention studies previously conducted or to be conducted in the future. Recommendations on other research are as follows:

• Studies testing the effects of omega-3 fatty acids on CVD risk factors for which there are insufficient data:
  • Glucose tolerance overall and in subgroups
  • Lipoprotein particle types associated with CVD
  • Composite of all risk factors
• Studies testing the effects on CVD risk factors, markers of CVD, and/or CVD (overall and in subgroups) of:
  • Omega-3 fatty acids
  • ALA
  • Fish oil versus ALA and their potential interaction
  • Ratio of omega-6 to omega-3 fatty acids
• Physiological measurements to identify and enhance knowledge on potential mechanisms, including:
  • Measures of arrhythmias to potentially explain effects on sudden death
  • Imaging studies to determine whether there are anti-atherogenic effects
  • Genetic studies to help determine predictors of response or modifications of effects
• Observational or methodological studies to examine:
  • Relationships between measurements of biomarkers of omega-3 fatty acids intake (e.g., blood levels and red blood cell membrane levels) and dietary intake of fish and omega-3 fatty acids
  • Best methods to estimate fish consumption, including type, amount, and method of preparation
  • Effects of single versus multiple doses of omega-3 fatty acids
  • Dose-response relationships of omega-3 fatty acids
  • Duration of intake required for effectiveness of omega-3 fatty acids
• In future studies of omega-3 fatty acids, reports on or measurements of:
  • Dietary omega-6 to omega-3 fatty acid ratio
  • Type, amount, and method of preparation of fish intake
  • Effects of higher fish consumption on intake of other foods and nutrients
  • Effects after cessation of omega-3 fatty acid intervention
• Other nutrition studies for consideration include effects of potassium, calcium, flavonoids, and obesity on various health outcomes.

III. Review Of AHRQ Evidence-Based Reports

The Working Group reviewed two evidence reports; ?Effects of Omega-3 Fatty Acids on Cardiovascular Risk Factors and Intermediate Markers of Cardiovascular Disease? (summary is at http://www.ahrq.gov/clinic/epcsums/o3cardrisksum.pdf  and ?Effects of Omega-3 Fatty Acids on Cardiovascular Disease? (summary is at http://www.ahrq.gov/clinic/epcsums/o3cardsum.pdf . To view the full reports, visit http://www.ahrq.gov/clinic/epcindex.htm#cardiovascular, scroll down to the reports (under the section on ?Dietary Supplements?), and click on the ?File Download? link for each report, which will open a Zip file.

The evidence-based reports were produced by reviewing six databases of scientific literature dated from 1966 through April 2003, focusing on observational epidemiologic studies and clinical trials on specific CVD outcomes and CVD risk factors and markers. CVD outcomes reviewed included CVD death (including fatal myocardial infarction and fatal stroke, cardiac death, sudden death), nonfatal myocardial infarction, and nonfatal stroke. All-cause mortality was also included. Specific CVD risk factors examined included blood levels of lipids (total cholesterol, low-density lipoprotein [LDL] cholesterol, high-density lipoprotein [HDL] cholesterol, triglycerides, lipoprotein (a), and apolipoproteins); glucose, insulin, and hemoglobin A1C; C-reactive protein; hemostasis measures (Factor VII, Factor VIII, von Willebrand Factor, platelet aggregation); and blood pressure. Intermediate markers of CVD were also examined, including coronary artery restenosis after coronary intervention, carotid intimal-media thickness, exercise tolerance, and heart rate variability. A total of 123 studies was included in the CVD risk factor and markers evidence report, and 39 studies were included in the CVD outcomes evidence report.

A. CVD Risk Factors and Intermediate Markers of CVD

Because of the large number of studies available, the EPC focused on randomized controlled trials for the review of the effects of omega-3 fatty acids on CVD risk factors and measures and markers of CVD. Most of the trials tested fish oil supplements. The evidence report noted that there was a great deal of heterogeneity of treatment effects across the studies. Very few studies assessed whether the effects on risk factors levels of omega-3 fatty acids were sustained after treatment ceased. It was also noted that the ratio of omega-6 to omega-3 fatty acids was not assessed in most trials.

A summary of the effects on risk factors and markers of CVD is as follows:

• A strong, consistent, dose-dependent beneficial effect on triglycerides (10%-30% lowering) that was generally significant across studies.
• A small overall significant beneficial effect on systolic and diastolic blood pressure (about 2 mm Hg decrease).
• Possible beneficial effects on:
  • coronary artery restenosis after angioplasty
  • exercise capacity in patients with coronary atherosclerosis
  • heart rate variability, particularly in patients with recent myocardial infarction
• Small, non-significant increases in total, LDL-, and HDL-cholesterol (the non-significant increases in LDL-cholesterol tended to be similar across studies, at about 10%).
• No consistent effects for carotid intimal-media thickness or for blood levels of apolipoproteins, lipoprotein (a), hemoglobin A1C, glucose, insulin, C-reactive protein, and all hemostasis measures.

A key observation from these studies is that if cardiovascular benefits of omega-3 fatty acid intake exist, they are not well-explained by their effects on the CVD risk factors examined. It is possible that any overall cardiovascular benefit may be due to the accumulation of small effects on triglycerides, blood pressure, coronary atherosclerosis, heart rate variability, and other unspecified factors. It is also possible that omega-3 fatty acids exert a benefit by preventing arrhythmias. It was noted that no study reported associations linking the effects of omega-3 fatty acids on CVD risk factors with their effects on CVD outcomes.

B. Cardiovascular Disease

Thirty-nine studies reported on CVD outcomes, 27 of which were observational epidemiologic studies (22 cohort studies), and 12 were randomized controlled trials. The study populations and interventions were heterogeneous as were the outcomes reported. Almost all of the observational studies (26 out of 27) were in individuals without pre-existing CVD (primary prevention population), whereas almost all of the trials (11 out of 12) were in individuals with pre-existing cardiovascular disease (secondary prevention population). Of the observational studies, mostly fish-based sources of omega-3 fatty acids (fish, EPA, DHA) were examined (26 analyses), compared with 5 analyses on plant-based sources (ALA). Among the trials, 7 analyses were on fish-based sources (Burr, 1989; Burr, 2003; Sacks, 1995; Singh, 1997; Leng, 1998; Nilsen, 2001; Marchioli, 2002) compared with 6 analyses on plant-based sources (Leren, 1966; Natvig, 1968; Singh, 1997; Singh, 2002; de Lorgeril, 1999; Bemelmans, 2002). Five of the 11 secondary prevention trials used supplements (Sacks, 1995; Singh, 1997; Leng, 1998; Nilsen, 2001; Marchioli, 2002), one of which included ALA (Singh, 1997), whereas 6 used dietary advice as the intervention (Leren, 1966; Burr, 1989; Burr, 2003; de Lorgeril, 1999; Singh, 2002; Bemelmans, 2002). The one primary prevention trial used linseed oil as the ALA supplement (Natvig, 1968); however, this trial had a follow-up period of only one year, which is too short to see an effect on primary prevention. Two of the 12 trials had very small sample sizes, ranging from 31 to 60 (Leng, 1998; Sacks, 1995), while one had a very low event rate (Bemelmans, 2002), precluding meaningful results. The results of these 4 trials are not included in this report (Natvig, 1968; Leng, 1998; Sacks, 1995; Bemelmans, 2002).

1. EPA and DHA Trials. The trial evidence on EPA + DHA was heavily weighted by the Gruppo Italiano per lo Studio della Sopravvivenza nell?Infarto miocardico (GISSI)-Prevenzione study, which had a total sample size of over 11,000 participants (Marchioli, 2002). The GISSI trial was conducted in post-myocardial infarction patients, consisting mostly of men, whose consumption of fish was moderate. GISSI reported a 21% reduction in all-cause mortality, 30% reduction in CVD death, 45% reduction in sudden death, 32% reduction in fatal myocardial infarction, and 20% reduction in all CVD events. There was no significant effect on stroke (relative risk 1.2, 95% CI = .81, 1.9) or nonfatal myocardial infarction (relative risk 0.9, 95% CI = 70, 1.2). GISSI had an open-label control group, whereas other trials had a placebo control group. The lack of placebo control in GISSI is a weakness in the study design because participants and their health care providers may alter their behavior based on knowledge of their treatment assignment, potentially leading to biased results. One supplements trial consistently showed no effects among various CVD outcomes (Nilsen, 2001), owing perhaps to the very high background fish intake in this population, whereas another one showed trends for reduction in sudden death, all CVD events, cardiac death, and nonfatal myocardial infarction, the latter two significant (Singh, 1997). One dietary trial advising increased fish intake showed significant reductions in all-cause mortality and cardiac death (Burr, 1989), while another one did not (Burr, 2003).

2. ALA Trials. Only one trial tested ALA supplements, using 2.9 g/d (Singh, 1997), and three trials gave dietary advice and estimated ALA intake (Leren, 1966; de Lorgeril, 1999; Singh, 2002). The supplements trial had a placebo control group and reported non-significant trends toward reductions in all CVD outcomes reported. The three dietary advice trials generally showed reductions in all-cause mortality as well as CVD outcomes, with about half of the various outcomes? effects being significant. It was noted that with dietary advice interventions may result in changes in other nutrients and dietary components, making it difficult to attribute the results specifically to ALA.

3. Observational Studies. The evidence from observational studies applies mainly to primary prevention populations, based on 27 studies, 22 of which were cohort studies. ALA was examined in only 3 prospective studies and one cross-sectional study. The studies were conducted in many parts of the world, including, the U.S., and included women. The follow-up period in the prospective studies ranged from 4 to 30 years, and the number of participants ranged from 272 to more than 223,000. Omega-3 fatty acid intakes were mostly estimated from food frequency questionnaires, although some studies reported only on the frequency of fish consumption or whether fish was consumed. Some studies used blood levels of omega-3 fatty acids.

In general, the prospective studies reported significant 20-60% lower risks of CVD outcomes such as CVD death, cardiac death, sudden death, and myocardial infarction with greater omega-3 fatty acids intakes, although several studies reported either positive or no significant associations with CVD outcomes. Associations with stroke were inconsistent.

A summary of the studies and their CVD results is as follows:

• Most primary prevention observational studies examined fish consumption and generally found inverse associations with all-cause mortality, CVD death, and CVD events. For stroke there were few studies, and findings were inconsistent and generally not significant.
• Trials in which the population had high baseline fish consumption tended to find no benefit of omega-3 fatty acids.
• None of the trials was conducted in the U.S., thus leaving open the question of their applicability to the U.S. population, where fish consumption is low.
• A recent meta-analysis of trials, which was not included in the evidence report but contained most of the the same studies (Bucher, 2002), reported overall 20-30% reductions in all-cause mortality, sudden death, and fatal myocardial infarction. Reduction in nonfatal myocardial infarction was not significant.
• The body of evidence for ALA is smaller than that for EPA and DHA; there are few observational studies, only one supplements trial, and three diet trials for ALA. Only one supplements trial tested the effects of both ALA and fish oil supplements on CVD outcomes.
• The influence of the ratio of omega-6 andto omega-63 fatty acids on outcomes has not been well-studied.
• Most studies of fish intake do not report the type of fish, method of preparation, or amounts of EPA and DHA in the fish.

C. Discussion of AHRQ Evidence Reports

The following observations on the AHRQ evidence reports were noted during the course of discussions:

• Previous observational studies reporting associations between nutritional factors and CVD endpoints were not subsequently supported by trial evidence. Also, in general, estimates of effects from observational studies tend to be larger than effects found in randomized controlled trials.
• Omega-3 fatty acids have been shown to be effective in reducing CVD in randomized controlled trials in secondary prevention populations only.
• The number and types of endpoints used in both the observational studies and trials varied.
• Because the AHRQ evidence report included 22 prospective cohort studies with up to 223,000 participants, there may be diminished returns in conducting additional observational studies.

IV. Considerations For Clinical Trials

A. Potential Interventions Using Omega-3 Fatty Acide

1. Omega-3 Fatty Acid Supplements. To test the effects on CVD of omega-3 fatty acid supplements in a clinical trial, the type, source, and dose of the supplement and of the placebo need to be defined. EPA and DHA are found primarily in fish, generally at a ratio of approximately 20:50 EPA to DHA. Fish oil supplements of different ratios have been used in trials. It was recommended that the ratio of EPA:DHA in supplements be somewhere between 3:2 to 2:3, as this range includes that which is typical for oily fish (2:3) and the ratio in Omacor (2.4:2), which was used in the GISSI study. ALA is primarily found in plant-based oils, such as flaxseed oil, which is a particularly rich source and contains 53% ALA. Interventions using supplements rather than foods or dietary advice are of known composition and can be better controlled, more feasibly delivered, more easily monitored for adherence, and can provide results less subject to potential confounding.

The Omega-3 Working Group of the Council for Responsible Nutrition (CRN)(http://www.crnusa.org/pdfs/O3FINALMONOGRAPHdoc.pdf)(link is external)  has produced information on a standard for omega-3 fatty acid formulation. Omega-3 fatty acid supplements should be pharmaceutical grade that meet CRN standards for high quality control. Capsules that contain the highest concentration of omega-3 fatty acids, such as those containing ethyl esters, reduce the required number of capsules by approximately two-thirds, compared to capsules providing omega-3 fatty acids as triglycerides. Based on clinical trials of supplements, a dose of approximately 1 g/d of EPA + DHA or of approximately 2-3 g/d of ALA is likely to be effective for secondary prevention. An ideal placebo oil would represent the usual fatty acid intake in the U.S. population and be low in omega-3 fatty acids. There is some concern that oils rich in omega-6 fatty acids may interfere with effectiveness of omega-3 fatty acids. Fish oil capsules should be indistinguishable from placebo capsules, e.g., through appearance, taste, and smell.

2. The Alpha-Omega Trial: Use of Food. The Alpha Omega Trial, a placebo-controlled 2 x 2 factorial randomized controlled trial, currently ongoing in the Netherlands, will determine the effect of low-dose supplementation with omega-3 fatty acids (EPA /+ DHA with or without ALA, or ALA with or without EPA + DHA) on CHD mortality in patients who had a myocardial infarction within the past 10 years. EPA + DHA (400 mg) and ALA (2 g) are delivered through specially prepared margarine which is given to participants. Placebo margarine contains oleic acid. Low-dose EPA + DHA is used because observational epidemiologic studies suggest that only 1-2 servings per week of fish is associated with lower risk, and 1g EPA + DHA was shown to be effective in the GISSI trial. The research question of interest is whether a lower dose, equivalent to about 1 serving of salmon per week, is effective. The Netherlands population has a low background diet of fish intake, and trial participants are excluded if their habitual fish intake exceeds 150 g/d (about 1 serving per day). Margarine intake is high in the Netherlands, making this mode of delivery of ALA omega-3 fatty acids feasible.

B. Considerations for Trial Designs

The Working Group discussed the types of trials that may be considered in investigating the effects of omega-3 fatty acids intake on CVD outcomes. The discussions focused on best approaches to conduct a trial, and revolved around the research question, which must define the target population, intervention, and outcome. Elements of trial design that were discussed included eligibility criteria, expected event rates, expected effect size, trial setting, number of arms, intervention type, dose, and mode of delivery, intervention duration, endpoints, and rough estimates of sample size.

1. Acute Coronary Syndrome Patients. A trial could be conducted in patients with Acute Coronary Syndrome (ACS), i.e., acute myocardial infarction or unstable angina. Participants would follow their normal drug treatment regimen that could include anti-atherosclerotic or anti-thrombotic medications.

Hypothetical trials approaches were explored for several types of ACS patients: stablized ACS patients (after early treatment is completed); acute ACS patients (upon hospital or emergency department presentation); or a combination of the two (acute treatment followed by treatment after stabilized). The doses would be higher and intervention duration shorter during the acute ACS phase. Estimated event rates would vary, with higher rates for acute compared with stabilized ACS patients. Specific primary or composite endpoints, such as CVD death, cardiac death, coronary death, and/or sudden death, would also vary depending on the patient population. Follow-up need be only for 30 days for the acute patients and 1-2 years for the combined patient population. With the combined approach, results would be applicable to most ACS patients and would determine whether omega-3 fatty acids have both a short-term and a long-term benefit. Rough sample size calculations suggest a population sample of 4,000 to 14,000 with a cost of $40M-$50M. If resources were available it would be useful to add measurements of CVD risk factors and subclinical disease markers to determine their validity as surrogate endpoints and to learn about the mechanistic pathway of any observed benefits.

2. Secondary Prevention Patients. Secondary prevention patients with chronic pre-existing disease could be defined simply as having had a myocardial infarction or other CVD events in the past. An older population would allow for a smaller sample size because of higher event rates; however, expanding the age eligibility would allow greater generalizability of results. A placebo control should be used to ensure design rigor and avoid potential bias problems. Having three arms (fish oil, ALA, placebo) would move the field forward because ALA has significant public health applicability even though most of the evidence on CVD benefit is for fish oil. Testing both fish oil and ALA would maximize options for public health recommendations. Although it would be desirable to test more than one dose, a test of multiple doses would increase trial complexity and sample size. The primary endpoint could simply be CVD death, which would have an expected rate of 1-1½% per year. However, provided the plausibility of a benefit is accepted, a composite endpoint would allow accrual of more events and thus keep the sample size to an affordable level. Other endpoints of interest include nonfatal or fatal stroke, nonfatal or fatal myocardial infarction, sudden death, cardiac death (CHD or heart failure), and all nonfatal or fatal CVD events. In GISSI, CVD deaths were significantly reduced by omega-3 fatty acids, but nonfatal myocardial infarction and nonfatal stroke were not. Therefore, CVD death may be an appropriate primary outcome. Based on an expected CVD death rate of approximately 1-1½% per year, a 2-arm trial may need 6,000-10,000 patients.

The setting could be clinic-based or through mail and/or Internet. Most information is obtained through self-report with verification through hospital records. Mail trials have been successful with good adherence and high-quality self-reports in highly educated populations such as health professionals, but a trial by mail may not be successful in a more general population, which is likely to require direct contact. A large simple trial would help keep costs down.

3. Primary Prevention for High-Risk Populations. NHLBI estimated that in 2002, more than 800,000 people in the U.S. were hospitalized for myocardial infarction as a primary diagnosis. The Centers for Disease Control and Prevention estimated that 400,000 ? 460,000 people die of heart disease in an emergency department or before reaching a hospital, which accounts for over 60% of all cardiac deaths. A substantial number of lives could be saved each year if increased intake of omega-3 fatty acids can prevent 30 percent of sudden deaths. Thus, the primary prevention of CHD, especially prevention of myocardial infarction, would have great public health benefit. A primary prevention trial that includes people without clinical CVD would be broadly generalizable to the U.S. population and allow testing the omega-3 fatty acids hypothesis sooner than if the hypothesis is first tested in only patients with disease, and if supportive, subsequently tested in a population without disease.

Sample size for a primary prevention study, even if targeting high-risk people by virtue of age or risk factor level, is likely to be large and thus may need to be a simple trial. To increase the risk level, the target population might be older, e.g., age 55 years and above, with at least one risk factor, or a high Framingham risk score.

To further increase the risk level of the population sample, one might consider a ?mixed? population sample where a certain percentage has experienced a CVD event. The percentage of each type needs to be considered, perhaps 70% primary prevention and 30% secondary prevention. With a combined population sample, study results would apply to both types of participants. An estimated sample size for a combined primary and secondary prevention population is 20,000 or 30,000 for a 2- or 3-arm trial. To maximize generalizability, there should be few other exclusion criteria, and participants should be able to continue their medication use. Use of mail, telephone, or Internet could be considered for intervention delivery or for communication. Regional clinical centers could be used to recruit large numbers of participants.

Several trial designs were explored, each placebo-controlled: (1) a 3-arm trial of EPA + DHA (1 g/d), ALA (2-3 g/d), and placebo; (2) a 3 x 2 factorial design with another unrelated intervention as the other factor; and (3) either of the previous designs with additional outcomes, such as blood pressure, triglycerides, diabetes, depression, or arthritis. Design assumptions could include lag time of 3-6 months and be powered for an effect rate size of 15-20%. The primary outcome could be death from CVD, including sudden death and fatal myocardial infarction. Secondary outcomes could be all-cause mortality, sudden death, and nonfatal myocardial infarction and stroke.

A primary prevention trial would fill a major gap in knowledge because virtually none of the trials to date have been conducted in a primary prevention population and thus, although they are hypothesized to do so, it is unknown whether omega-3 fatty acids can prevent first events. On the other hand, a trial of individuals both with and without disease could potentially address the secondary prevention question as well and may be needed to obtain a sufficient event rate to be feasible.

4. Summary of Discussions on Study Designs.The working group concluded the following:

• There was little enthusiasm for a 2 x 2 factorial design (i.e., fish oil and ALA) because the two interventions are related and there is risk of interaction or sub-additivity. If there is an interaction, the efficiency in asking two research questions is lost.
• Any trial needs clinical CVD endpoints rather than surrogate endpoints because the mechanism underlying the effects of omega-3 fatty acids is not known.
  • Using a surrogate endpoint could lead to the wrong conclusions if the wrong surrogate is used.
  • If preventing arrhythmias is the main mechanism through which omega-3 fatty acids work, there are no good surrogate endpoints. The presence or prevention of arrthythmias may not be an indicator of occurrence or prevention of clinical events, as exemplified by the CAST trial (CAST, 1989), which showed that certain drugs that suppress arrthythmias increase mortality.
• Definitions of CVD outcomes should be precisely described.
• Nonfatal myocardial infarction should not be included as part of a composite outcome because nonfatal myocardial infarction was not reduced in GISSI, and its inclusion may reduce any observed effect size. However, it could be a secondary outcome.
• The effects on stroke have been mixed in trials and observational studies, although recent observational studies tend to show a protective association. Because of an unclear direction of effect, nonfatal stroke should not be included in the primary outcome; however, it should be measured as a secondary outcome.
• Use of concomitant medication is important to assess, especially aspirin, and would be expected to be distributed equally across treatment arms. (There is a question based on observational data as to whether omega-3 fatty acids are less effective when combined with aspirin, though the GISSI trial sample had a high proportion of participants taking aspirin.)
• Potential effect modifiers should be measured, including demographic characteristics, background medication use, and background diet, particularly intake of fish (type, amount, and method of preparation) and of omega-3 and omega-6 fatty acids.

References Cited

Bemelmans WJ, Broer J, Feskens EJM, Smit J, Muskiet FAJ, Lefrandt JD, Bom VJJ, May JF, Meyboom-de Jong B. Effect of an increased intake of alpha-linolenic acid and group nutritional education on cardiovascular risk factors: the Mediterranean Alpha-linolenic Enriched Groningen Dietary Intervention (MARGARIN) study. Am J Clin Nutr 2002;75:221-227.

Bucher HC, Hengstler P, Schindler C, Meier G. N-3 polyunsaturated fatty acids in coronary heart disease: A meta-analysis of randomized controlled trials. Am J Med 2002;112:298-304.

Burr ML, Ashfield-Watt PAL, Dunstan FDJ, Fehily AM, Breay P, Aston T, Zotos PC, Haboubi NAA, Elwood PC. Lack of benefit of dietary advice to men with angina: results of a controlled trial. Eur J Clin Nutr 2003;57:193-200.

Burr ML, Gilbert JF, Holliday RM, Elwood PC, Fehily AM, Rogers S, Sweetnam PM, Deadman NM. Effects of changes in fat, fish, and fibre intakes on death and myocardial reinfarction: Diet and Reinfarction Trial (DART). Lancet 1989;2:757-761.

Cardiac Arrhythmia Suppression Trial (CAST) Investigators. Preliminary report: effect of encainide and flecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infaction. The Cardiac Arrhythmia Suppression Trial (CAST). N Engl J Med 1989; 321:406-412.

De Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: Final report of the Lyon Diet Heart Study. Circulation 1999;99:779-785.

Kris-Etherton P, Eckel RH, Howard BV, St. Jeor S, Bazzarre TL, for the Nutrition Committee. American Heart Associaton Science Advisory. Benefits of a Mediterranean-style, National Cholesterol Education Program/American Heart Association Step I dietary pattern on cardiovascular disease. Circulation 2001; 103:1823-1825.

Kris-Etherton P, Harris WS, Appel LJ, for the AHA Nutrition Committee. Omega-3 fatty acids and cardiovascular disease. New recommendations from the American Heart Association. Arterioscler Thromb Vasc Biol 2003;23:151-152.

Leng GC, Lee AJ, Fowkes FG, Jepson RG, Lowe GD, Skinner ER, Mowat BF. Randomized controlled trial of gamma-linolenic acid and eicosapentaenoic acid in peripheral arterial disease. Clin Nutr 1998;17:265-271.

Leren P. The effect of plasma cholesterol lowering diet in male survivors of myocardial infarction. A controlled clinical trial. Acta Medica Scandinavica ? Supplementum 1966;466;1-92.

Marchioli R, Barzi F, Bomba E, Chieffo C, Gregorio DD, Di Mascio R, Franzosi MG, Geraci E, Levantesi G, Maggioni P, Mantini L, Marfisi RM, Mastrogiuseppe G, Mininni N, Nicolosi GL, Santini M, Schweiger C, Tavazzi L, Tognoni G, Tucci C, Valagussa F. Early protection against sudden death by n-3 polyunsturated fatty acids after myocardial infarction: time-course analysis of the results of the Gruppo Italiano per lo Studio della Sopravvivenza nell?Infarto Miocardico (GISSI)-Prevenzione. Circulation 2002; 105:1897-1903.

National Heart, Lung, and Blood Institute, National Institutes of Health. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Final Report. NIH Publication No. 02-5215, September, 2002, pp. V14-V15. Available at: http://www.nhlbi.nih.gov/guidelines/cholesterol/index.htm

Natvig H, Borchgrevink CF, Dedichen J, Owren PA, Schiotz EH, Westlund K. A controlled trial of the effect of linolenic acid on incidence of coronary heart disease. The Norwegian vegetable oil experiment of 1965-66. Scand J Clin Lab Invest-Supplement 1968;105:1-20.

Nilsen DW, Alberektsen G, Landmark K, Moen S, Asrsland T, Woise L. Effects of a high-dose concentrate of n-3 fatty acids or corn oil introduced early after an acute myocardial infarction on serum triacylglycerol and HDL cholesterol. Am J Clin Nutr 2001;74:50-56.

Sacks FM, Stone PH, Gibson CM, Silverman DI, Rosner B, Pasternak RC. Controlled trial of fish oil for regression of human coronary atherosclerosis. HARP Research Group. J Am Coll Cardiolog 1995;25:1492-1498.

Singh RB, Niaz MA, Sharma JP, Kumar R, Rastogi V, Moshiri M. Randomed, double-blind, placebo-controlled trial of fish oil and mustard oil in patients with suspected acute myocardial infarction: the Indian Experiment of Infarct Survival?4. Cardiovasc Drugs Ther 1997;11:485-491.

Singh RB, Dubnov G, Niaz MA, Ghosh S, Singh R, Rastogi SS, Manor O, Pella D, Berry E. Effect of an Indo-Mediterranean diet on progression of coronary artery disease in high risk patients (Indo-Mediterranean Diet Heart Study): a randomized single-blind trial. Lancet 2002;360:1455-1461.