Chapter  93:  Effects of Omega-3 Fatty Acids on Cardiovascular Risk Factors and Intermediate Markers of Cardiovascular Disease

Metabolism and Biological Effects of Essential Fatty Acids

Dietary fat is an important source of energy for biological activities in human beings. Dietary fat encompasses saturated fatty acids, which are usually solid at room temperature, and unsaturated fatty acids, which are liquid at room temperature. Unsaturated fatty acids can be further divided into monounsaturated and polyunsaturated fatty acids. Polyunsaturated fatty acids can be classified on the basis of their chemical structure into two groups: omega-3 (n-3) fatty acids and omega-6 (n-6) fatty acids. The omega-3 or n-3 notation means that the first double bond from the methyl end of the molecule is in the third. The same principle applies to the omega-6 or n-6 notation. Despite their differences in structure, all fats contain the same amount of energy (9 kcal/g or 37 kJ/g).

Of all fats found in food, 2 — ALA and linoleic acid (LA, 18:2 n-6) — cannot be synthesized in the human body, yet are necessary for proper physiological functioning. These 2 fats are called essential fatty acids. The essential fatty acids can be converted in the liver to long-chain polyunsaturated fatty acids, which have a higher number of carbon atoms and double bonds. These long-chain polyunsaturated fatty acids retain the omega type (n-3 or n-6) of the parent essential fatty acids.

ALA and LA comprise the bulk of the total polyunsaturated fatty acids consumed in a typical North American diet. Typically, LA comprises 89% of the total polyunsaturated fatty acids consumed, while ALA comprises 9%. Smaller amounts of other polyunsaturated fatty acids make up the remainder ¹. Both ALA and LA are present in a variety of foods. For example, LA is present in high concentrations in many commonly used oils, including safflower, sunflower, soy, and corn oil. ALA, which is consumed in smaller quantities, is present in leafy green vegetables and in some commonly used oils, including canola and soybean oil. Some novelty oils, such as flaxseed oil, contain relatively high concentrations of ALA, but these oils are not commonly found in the food supply.

The Institute of Medicine suggests that, for adults 19 and older, an adequate intake (AI) of ALA is 1.1–1.6 g/day, while an adequate daily intake of LA is 11–17 g/day ². Recommendations regarding AI differ by age and gender groups, and for special conditions such as pregnancy and lactation.

As shown in Figure 1.1, EPA and DHA can act as competitors for the same metabolic pathways as AA. In human studies, the analyses of fatty-acid compositions in both blood phospholipids and adipose tissue showed similar competitive relationship between omega-3 long-chain polyunsaturated fatty acids and AA. General scientific agreement supports an increased consumption of omega-3 fatty acids and reduced intake of omega-6 fatty acids to promote good health. However, for omega-3 fatty acid intakes, the specific quantitative recommendations vary widely among countries not only in terms of different units — ratio, grams, total energy intake — but also in quantity ³. Furthermore, there remain numerous questions relating to the inherent complexities about omega-3 and omega-6 fatty acid metabolism, in particular regarding the inter-relationships between the 2 fatty acids. For example, it remains unclear to what extend ALA is converted to EPA and DHA in humans, and to what extend high intake of omega-6 fatty acids compromises any benefits of omega-3 fatty acid consumption. Without resolution of these 2 foundational questions, it remains difficult to study the importance of omega-6 to omega-3 fatty acid ratio.

Metabolic Pathways of Omega-3 and Omega-6 Fatty Acids

Omega-3 and omega-6 fatty acids share the same pools of enzymes and go through the same oxidation pathways while being metabolized (Figure 1.1). Once ingested, ALA and LA can be elongated and desaturated into long-chain polyunsaturated fatty acids. LA is converted into gamma-linolenic acid (GLA, 18:3 n-6), an omega-6 fatty acid that is a positional isomer of ALA. GLA, in turn, can be converted to the long-chain omega-6 fatty acid, arachidonic acid (AA, 20:4 n-6). ALA can be converted, to a lesser extent, to the long-chain omega-3 fatty acids, EPA and DHA. However, the conversion from parent fatty acids into long-chain polyunsaturated fatty acids occurs slowly in humans, and conversion rates are not well understood. Because of the slow rate of conversion and the importance of long-chain polyunsaturated fatty acids to many physiological processes, humans must augment their level of long-chain polyunsaturated fatty acids by consuming foods that are rich in these important compounds. Meat is the primary food source of AA, while fish is the primary food source of EPA.

The specific biological functions of fatty acids depend on the number and position of double bonds and the length of the acyl chain. Both EPA and AA are 20-carbon fatty acids and are precursors for the formation of prostaglandins, thromboxane, and leukotrienes — hormone-like agents that are members of a larger family of substances called eicosanoids. Eicosanoids are localized tissue hormones that seem to be one of the fundamental regulatory classes of molecules in most higher forms of life. They do not travel in the blood, but are created in the cells to regulate a large number of processes, including the movement of calcium and other substances into and out of cells, dilation and contraction of muscles, inhibition and promotion of clotting, regulation of secretions including digestive juices and hormones, and control of fertility, cell division, and growth ⁴.

As shown in Figure 1.1, the long-chain omega-6 fatty acid, AA, is the precursor of a group of eicosanoids including series-2 prostaglandins and series-4 leukotrienes. The omega-3 fatty acid, EPA, is the precursor to a group of eicosanoids including series-3 prostaglandins and series-5 leukotrienes. The series-2 prostaglandins and series-4 leukotrienes derived from AA are involved in intense actions (such as accelerating platelet aggregation and enhancing vasoconstriction and the synthesis of inflammatory mediators) in response to physiological stressors. The series-3 prostaglandins and series-5 leukotrienes that are derived from EPA are less physiologically potent than those derived from AA. More specifically, the series-3 prostaglandins are formed at a slower rate and work to attenuate excessive series-2 prostaglandins. Thus, adequate production of the series-3 prostaglandins, which are derived from the omega-3 fatty acid, EPA, may protect against heart attack and stroke as well as certain inflammatory diseases like arthritis, lupus, and asthma ⁴. In addition, animal studies, have demonstrated that omega-3 fatty acids, such as EPA and DHA, engage in multiple cytoprotective activities that may contribute to antiarrhythmic mechanisms ⁵. Arrhythmias are a common cause of “sudden death” in heart disease.

In addition to affecting eicosanoid production as described above, EPA also affects lipoprotein metabolism and decreases the production of other compounds - including cytokines, interleukin 1ß (IL), and tumor necrosis factor a (TNF-a) - that have pro-inflammatory effects. These compunds exert pro-inclammatory cellular actions that include stimulating the production of collagenases and inreasing the expression of adhesion molecules necessary for leukocyte extravasation ⁶. The mechanism responsible for the suppression of cytokine production by omega-3 fatty acids remains unknown, although suppression of eicosanoid production by omega-3 fatty acids may be involved. EPA can also be converted into the longer chain omega-3 form of DPA, and then further elongated and oxygenated into DHA. EPA and DHA are frequently referred to as very long chain omega-3 fatty acids. DHA, which is thought to be important for brain development and functioning, is present in significant amounts in a variety of food products, including fish, fish liver oils, fish eggs, and organ meats. Similarly, AA can convert into an omega-6 form of DPA. Studies have reported that omega-3 fatty acids decrease triglycerides (Tg) and very low density lipoprotein (VLDL) in hypertriglyceridemic subjects, with a concomitant increase in high density lipoprotein (HDL). However, they appear to increase or have no effect on low density lipoprotein (LDL). Omega-3 fatty acids apparently lower Tg by inhibiting VLDL and apolipoprotein B-100 synthesis and decreasing post-prandial lipemia ⁷. Omega-3 fatty acids, in conjunction with transcription factors (small proteins that bind to the regulatory domains of genes), target the genes governing cellular Tg production and those activating oxidation of excess fatty acids in the liver. Inhibition of fatty acid synthesis and increased fatty acid catabolism reduce the amount of substrate available for Tg production ⁸.

As noted earlier, omega-6 fatty acids are consumed in larger quantities (>10 times) than omega-3 fatty acids. Maintaining a sufficient intake of omega-3 fatty acids is particularly important since many of the body's physiologic properties depend upon their availability and metabolism.

Population Intake of Omega-3 Fatty Acids in the United States

The major source of omega-3 fatty acids is dietary intake of fish, fish oil, vegetable oils (principally canola and soybean), some nuts including walnuts, and dietary supplements. Two population-based surveys, the third National Health and Nutrition Examination (NHANES III) 1988-94 and the Continuing Food Survey of Intakes by Individuals 1994-98 (CSFII) surveys, are the main source of dietary intake data for the U.S. population. NHANES III collected information on the U.S. population aged =2 months. Mexican Americans and non-Hispanic African-Americans, children =5 years old, and adults = 60 years old were over-sampled to produce more precise estimates for these population groups. There were no imputations for missing 24-hour dietary recall data. A total of 29,105 participants had complete and reliable dietary recall. Complete descriptions of the methods used and fuller analyses are available in the report Effects of Omega-3 Fatty Acids on Cardiovascular Disease, under “Methods: Method to Assess the Dietary Intake of Omega-3 Fatty Acids in the US population” and “Results: Population Intake of Omega-3 Fatty Acids in the United States”. CSFII 1994-96, popularly known as the What We Eat in America survey, addressed the requirements of the National Nutrition Monitoring and Related Research Act of 1990 (Public Law 101–445) for continuous monitoring of the dietary status of the American population. In CSFII 1994-96, an improved data-collection method known as the multiple-pass approach for the 24-hour recall was used. Given the large variation in intake from day-to-day, multiple 24-hours recalls are considered to be the best suited for most nutrition monitoring and will produce stable estimates of mean nutrient intakes from groups of individuals ⁹. In 1998, the Supplemental Children's Survey, a survey of food and nutrient intake by children under age of 10, was conducted as the supplement to the CSFII 1994-96. The CSFII 1994-96, 1998 surveyed 20,607 people of all ages with over-sampling of low-income population (<130% of the poverty threshold). Dietary intake data by individuals of all ages were collected over 2 nonconsecutive days by use of two 1-day dietary recalls.

Table 1.1 reports the NHANES III survey mean intake ± the standard error of the mean (SEM), as well as, the median and range for each omega-3 fatty acid. Distributions of EPA, DPA, and DHA were very skewed; therefore, the means and standard errors of the means should be used and interpreted with caution. Table 1.2 reports the CSFII survey mean and median intakes for each omega-3 fatty acid, along with SEMs, as reported in Dietary Reference Intakes by the Institute of Medicine ².

Dietary Sources of Omega-3 Fatty Acids

Omega-3 fatty acids can be found in many different sources of food, including fish, shellfish, some nuts, and various plant oils. Table 1.3 lists the amount of omega-3 fatty acids in some commonly consumed fish, shellfish, nuts, and edible oils, selected from the USDA website http://www.nal.usda.gov/fnic/foodcomp (accessed November 3, 2003; Finfish and Shellfish Products: sr16fg15.pdf; Fats and Oils: sr16fg04.pdf; and Nut and Seed Products: sr16fg12.pdf) ¹⁰

Improvement of Lipoproteins

Elevated serum low density lipoprotein (LDL) and depressed high density lipoprotein (HDL), especially when accompanied by elevated triglycerides (Tg), are well-known risk factors for CVD. Studies have reported that omega-3 fatty acids decrease Tg and very low density lipoprotein (VLDL) in hypertriglyceridemic subjects, with a concomitant increase in HDL. However, they appear to increase or have no effect on LDL. Omega-3 fatty acids apparently lower Tg by inhibiting VLDL and apolipoprotein B-100 (apo B-100) synthesis and decreasing post-prandial lipemia ⁷. Omega-3 fatty acids, in conjunction with transcription factors (small proteins that bind to the regulatory domains of genes), target the genes governing cellular Tg production and those activating oxidation of excess fatty acids in the liver. Inhibition of fatty acid synthesis and increased fatty acid catabolism reduce the amount of substrate available for Tg production ⁸.

Numerous other lipids and associated proteins are involved in lipid metabolism and thus possibly in atherogenesis and CVD; although they are less commonly measured. These include, among others, lipoprotein (a) [Lp(a)]; apolipoproteins (apo) A-I, B-48, B-100, C-III; and free fatty acids.

Reduction of Thrombosis

Blockage of coronary, cerebral and peripheral vessels due to thrombosis is a leading cause of CVD. Omega-3 fatty acids affect the clotting system in a number of ways. EPA competes with AA for the cyclo-oxygenase enzyme, thus reducing thromboxane A₂ (TX), a thrombotic agent. DHA may further inhibit cyclo-oxygenase ¹⁷. Omega-3 fatty acids also inhibit TXB₂ production, platelet aggregation, and platelet adhesion, although much less so than aspirin. Omega-3 fatty acids also lead to endothelial formation of prostaglandin I₃ (PG), PGI₂, and nitrous oxide, all of which reduce vasoconstriction ^{17, 18}. However, knowledge about the role of omega-3 fatty acids on coagulation factors and fibrinolysis is incomplete.

Many markers of coagulability exist, including the numerous factors involved in the clotting cascade, homocysteine, bleeding time, and platelet aggregation. Except among people with specific hypercoagulable conditions, it is not clear that any of these measures, among others, are predictive of CVD or that modification of their levels modifies risk of CVD.

Reduction of Inflammation, Atherogenesis, and Leukocyte Activity

Awareness of the effect of inflammation on atherogenesis (atheromatous plaque formation) and the risk of cardiovascular events is increasing. Leukocytes (white blood cells) are the blood cells that respond to injury or infection with a protective inflammatory response and an immune response. However, leukocytes are prominent cells in the atheromatous plaque in major blood vessels, which suggests that early plaque formation has an inflammatory component. PGE₂ and leukotriene B₄ (LT) have pro-inflammatory biological actions, and together they can cause vascular leakage and extravasation of fluid. The omega-6 fatty acid, AA, is the progenitor of both PGE2 and LTB4 via the cyclo-oxygenase and 5-lipo-oxygenase enzymatic pathways, respectively. EPA is the omega-3 homologue of AA; the 2 fatty acids differ only in that EPA has 1 additional double bond at the third carbon. EPA can thus inhibit AA metabolism competitively via the enzymatic pathways and can suppress production of the omega-6 fatty acid eicosanoid inflammatory mediators. Although EPA promotes the formation of PGE3 and LTB5, these eicosanoids are far less active as pro-inflammatory agents than the corresponding derivatives of AA ⁸. Furthermore, other pro-inclammatory factors, such as IL-1ß and TNF-a, can be suppressed by the effect of long-chain polynsaturated fatty acids on lipoprotein metabolism ⁶.

C-reactive protein (CRP) is a well-described marker of inflammation and rises in response to injury, infection, and other inflammatory stimuli. In patients with either angina or risk factors for atherosclerosis, increased CRP has been associated with increased relative risk of nonfatal myocardial infarction and overall cardiovascular mortality ¹⁹. It is unclear whether reduction in CRP would result in reduced risk of CVD. Trials commonly measure other inflammatory markers including IL-6 and vascular cell adhesion molecule 1 (VCAM-1). Less is known about their association with CVD.

Reduction of Arrhythmia

Cardiac arrhythmias can be fatal, causing sudden death, or can result in stroke, myocardial infarction, congestive heart failure, and peripheral embolisms, among other types of CVD. Animal studies have shown that fatal ventricular fibrillation could be essentially abolished by high-level feeding with omega-3 fatty acids ²⁰. Omega-3 fatty acids appear to act in multiple ways to prevent arrhythmias. Various animal and in vitro experiments have shown that omega-3 fatty acids directly modulate sodium, potassium, and calcium channels ²¹. By incorporating into cell membrane phospholipids, the excitation-contraction coupling that can result in arrhythmia is reduced ²². Omega-3 fatty acids also modulate various intracellular enzymes involved in controlling the contraction and relaxation cycles of myocytes ²³. EPA and DHA also affect adrenoceptors, membrane proteins whose function in the heart is to transmit the neuroendocrine message of the catecholamines (adrenaline and its derivatives) ²⁴. The activity of DHA is thus similar in principle to that of ß-blockers, a group of key cardiovascular drugs used to decrease the cardiac effects of catecholamines. Omega-3 long-chain polyunsaturated fatty acids also appear to act similarly to another group of cardiovascular durgs, calcium channel blockers, by increasing intracellular calcium sequestration and interfering with receptor-operated calcium channels, thus lowering calcium influx ²². The effect of omega-3 fatty acids on prostanoids and leukotrienes also theoretically reduces the arrhythmia potential of cardiac myocytes.

The risk of ventricular arrhythmia is most commonly measured by 24 hour ambulatory electrocardiography recordings, in which a continuous electrocardiogram (ECG) is taken for generally 24 hours. Various measures of heart rate variability are calculated, primarily based on the standard deviation (SD) of the duration of time between heart beats. Other common ECG measurements are also followed as indicators or risk of arrhythmia or cardiac ischemia.

Blood Pressure

Hypertension is well recognized as one of the leading causes of CVD. The recent Joint National Committee report (JNC 7) emphasizes the risks of blood pressure that is even slightly elevated above 120/80 mm Hg ²⁵. Lifestyle modification, including reduction of sodium and alcohol intake, weight loss, diets high in fruits and vegetables and low-fat dairy products, and exercise has been shown to reduce blood pressure, often as much as medication use. Early investigations into the way in which fatty fish consumption may lower CVD found that omega-3 fatty acids possibly reduce blood pressure ²⁶. While the mechanisms for such an effect remain uncertain, the most compelling hypothesis is that by altering the balance between vasoconstrictive TXA2 and vasodilatory PGI3, as described in the section on inflammation, overall blood vessel capacitance increases and thus blood pressure falls ²⁷. However, the baseline balance of vasoactive and regulatory hormones may be altered in people with frank hypertension or other types of CVD The question thus arises whether the effect of omega-3 intake on blood pressure is altered in people with hypertension.

Diabetes

Although long-chain omega-3 fatty acids appear to have an overall beneficial effect on CVD, their effect on glucose homeostasis is less clear. Omega-3 fatty acids may, in fact, have a detrimental effect on glucose tolerance ²⁸. Theoretical benefits of omega-3 fatty acids to diabetic management include reducing Tg, increasing HDL, increasing glucose-induced insulin secretion, and possibly lowering insulin resistance ^{28, 29}. However, omega-3 fatty acids may worsen glucose tolerance in patients with clear cut diabetes and may, in fact, worsen insulin resistance ²⁸.

Thus, important questions relate to the level of markers of glucose tolerance, such as fasting blood glucose (FBS), glycohemoglobin or hemoglobin A_1c (Hgb A_1c), and fasting insulin levels, in people with both diabetes and insulin resistance and people without glucose tolerance impairment.

Cardiovascular Diagnostic Tests

The metabolic effects of omega-3 fatty acids on lipoproteins, thrombosis, inflammation, arrhythmia and blood pressure all have potential effects on blood vessels and the heart, which eventually can lead to clinical CVD. In addition, there are numerous diagnostic tests of cardiovascular health that are known to be predictive of future cardiovascular events both in people with and without a known history of CVD. Improvements in these diagnostic tests are commonly used as indicators of effective prophylaxis or treatment.

Among the tests of vascular health that have been assessed in omega-3 fatty acid trials are coronary arteriography (to measure coronary vessel stenosis), carotid intima-media thickness (IMT, which measures the thickness of the carotid artery wall, a measure of atherosclerosis), carotid Doppler ultrasonography or magnetic resonance arteriography (to measure carotid and extra-carotid stenosis), ankle brachial index (to measure peripheral blood flow), and endothelium-dependent vasorelaxation (an invasive or minimally invasive test of endothelial function). Other useful diagnostic tests measure heart function, including the exercise tolerance test (treadmill or stress test) and cardiac ultrasonography (which measures heart wall, chamber and valve structure and function).

Omega-3 Fatty Acids Search Strategy

A wide variety of search terms were used to capture the many potential sources of omega-3 fatty acids. Search terms used include the specific fatty acids, fish and other marine oils, and specific plant oils (flaxseed, linseed, rapeseed, canola, soy, walnut, mustard seed, butternut, and pumpkin seed). These terms were used in all search strategies.

Cardiovascular Search Strategy

The primary search strategy was designed to address both the clinical and intermediate outcomes of CVD in humans (Appendix A.1). In order to identify CVD outcomes in human studies, the search was divided into 3 categories consisting of controlled trials, other studies, and reviews. These 3 categories were further divided into English and non-English subsets.

Diabetes

Because specific terms referring to diabetes had been omitted from the primary search strategy, a supplemental search strategy was conducted on March 29, 2003. The diabetes supplemental search strategy included relevant search terms for diabetes. This search strategy resulted in an additional 410 citations for screening (Appendix A.2).

Supplemental Searches

Because some studies evaluated the effect of nuts on CVD outcomes without specifying in the abstract the type of nuts used in the study, we performed a supplemental Medline search on July 30, 2003 using the term “nut” as a text word for studies of CVD (Appendix A.3). Furthermore, upon noting that a number of relevant articles were missing from our search strategy, we performed a supplemental search on July 1, 2003. This search included terms specific to the CVD risk factor and intermediate markers outcomes of interest (Appendix A.4).

Overall

The number of citations for the final results of the database searches is approximate. Because the 5 main databases used in the search employ different citation formats, duplicate publications were encountered. The UO EPC eliminated most of the duplicate publications, however, because of many different permutations it was impossible to identify all of them. We eliminated duplicate publications as we encountered them.

Ongoing automatic updates of Medline searches were conducted using the CVD search strategy. The last automatic update was on April 19, 2003. The UO EPC conducted a final update search of the other databases on April 10, 2003.

Abstract Screening

All abstracts identified through the literature search were screened manually. At this stage, eligibility criteria were loosely defined to include all English language primary experimental or observational studies that evaluated any potential source of omega-3 fatty acids in at least 5 human subjects, irrespective of the study outcomes reported in the abstract. We excluded abstracts that clearly included only subjects who had a non-CVD-related condition (such as cancer, schizophrenia, or organ transplant), letters and abstracts.

Full Article Inclusion Criteria

Articles that passed the abstract screening process were retrieved and the full articles were screened for eligibility. Articles were rejected during this round based on the following criteria: review articles, inappropriate human population, pediatric studies and those conducted on subjects less than 19 years old, no mention of omega-3 fatty acid dietary supplements or fish consumption, daily dose of omega-3 fatty acid greater than 6 g, fewer than 5 subjects in omega-3 fatty acid arm(s), prospective interventional studies of less than 4 weeks duration, crossover studies with less than 4 week washout between treatments, and no appropriate outcome of interest reported. Studies that reported only the tissue level of omega-3 fatty acid without explicitly reporting the amount of omega-3 fatty acid consumed were also excluded. Studies that reported only lipid data among the outcomes of potential interest with fewer than 20 subjects were excluded during screening because of the large number of such studies and limited resources. In addition, with the exception of studies of Mediterranean diets and studies that reported fish servings, studies were excluded if no specific data were reported about omega-3 fatty acid consumption. Specific sources of omega-3 fatty acids considered acceptable included fish oils, dietary fish, canola (rapeseed) oil, soybean oil, flaxseed or linseed oil, walnuts or walnut oil, and mustard seed oil. Other sources were eligible if omega-3 fatty acid levels were reported to be greater than control. For each study that was rejected, the reason(s) for rejection was noted.

The exclusion criterion of more than 6 g per day for non-adverse event clinical outcomes was based on discussions with the TEP, in which it was agreed that omega-3 fatty acid intake above this amount is impractical and has little relevance on health care recommendations. Therefore, the inclusion criterion for the maximum daily intake was set at 6 g per day. The definition of dose of omega-3 fatty acids varied greatly across studies. Thus, the maximal allowable dose may have applied to total daily omega-3 fatty acid, total EPA plus DHA, or a total of other combinations of omega-3 fatty acids. The total did not refer to total fish oil. Short duration studies (less than 4 weeks) and crossover studies with washout periods less than 4 weeks were excluded since, it was agreed, a metabolic steady-state of omega-3 fatty acids is likely not achieved for about 4 weeks.

Sometimes there were multiple publications of the same study reporting interim results or different outcomes. We identified and grouped articles belonging to the same overall study and used data from the latest publication, supplemented by data from earlier publications, as appropriate.

In addition, a list of approximately 100 potential markers of CVD (e.g., coronary intima media thickness) and risk factors (e.g., hypertension, C-reactive protein) was reviewed in detail. Because of limited time and resources, 22 factors were chosen from this list for definite inclusion. A second list of factors was evaluated for possible inclusion if time and resources allowed (see Table 3.1 in Results section). Studies that reported on none of these factors were rejected.

Because of the large number of studies available for analysis, for most outcomes of interest we decided to confine analysis to the largest randomized trials for each outcome evaluated. For outcomes with few studies, all studies were included regardless of study design or sample size (minimum of 5 subjects). We used a lower sample size threshold for crossover studies because these studies are more strongly powered for a given number of subjects than parallel studies. We generally aimed for approximately 20 to 25 studies for analysis. For studies of platelet aggregation, we used the additional inclusion criterion that platelet aggregation data must be presented in a numerical format; articles that reported platelet aggregation results only graphically were not analyzed. This additional criterion was used because of the particular difficulty in estimating data from graphs for this outcome and because of the large number of specific outcomes reported in each study. Specific criteria used are listed in Table 3.1 and described in each outcome section in Chapter 3.

Incorporation of omega-3 fatty acids into phospholipids is very commonly reported by studies, often as proof of treatment compliance. Again because of limited time and resources, we limited our review of studies examining omega-3 fatty acid incorporation (or the association between dietary omega-3 fatty acid intake and tissue levels of omega-3 fatty acids) to the larger randomized trials that met eligibility criteria for either intermediate or clinical outcomes. We based this decision on the assumption that this sample of studies should not be biased. In addition, because the primary research question concerns correlation between dietary intake and blood levels of omega-3 fatty acids, for these analyses we have included only prospective, intervention trials to avoid biases and inaccuracies inherent to retrospective or survey-based studies. We have limited measurable levels to those most commonly reported and most practically measured, including erythrocyte, platelet cell membrane, and plasma phospholipids.

Common Elements for Grading the Methodological Quality of Randomized Controlled Trials in Evidence Reports

As part of the overall omega-3 fatty acid project, the 3 collaborating EPCs agreed to use the Jadad Score and adequacy of random allocation concealment as elements to grade individual randomized controlled trials ^{41, 42}. We also agreed that individual EPCs might add other elements to this core set, as we deemed appropriate. All EPCs agreed that studies should not be graded using a single numerical quality score, as this has been found to be unreliable and arbitrary ⁴³.

The Jadad Score assesses the quality of randomized controlled trials using 3 criteria: adequacy of randomization, double blinding, and drop outs ⁴¹. A study that fully meets all 3 criteria gets a maximum score of 5 points. Adequacy of allocation concealment was assessed using the criteria described by Schulz et al., as adequate, inadequate, or unclear ⁴².

Generic Summary Quality Grade for Studies

The Jadad and Schulz scores address only some aspects of the methodological quality of randomized controlled trials. Potential biases due to reporting and analytic problems in the study are ignored. In this evidence report, we applied a 3-category grading system (A, B, C) to each randomized trial. We have used this grading system in most of our previous EPC evidence reports, as well as in several evidence based clinical practice guidelines ⁴⁴. This scheme defines a generic grading system for study quality that is applicable to each type of study design (i.e., randomized controlled trial, cohort study, case-control study):

1. Least bias; results are valid. A study that mostly adheres to the commonly held concepts of high quality, including the following: a formal randomized study; clear description of the population, setting, interventions and comparison groups; appropriate measurement of outcomes; appropriate statistical and analytic methods and reporting; no reporting errors; less than 20% dropout; clear reporting of dropouts; and no obvious bias.

2. Susceptible to some bias, but not sufficient to invalidate the results. A study that does not meet all the criteria in category A. It has some deficiencies but none likely to cause major bias. Study may be missing information making assessment of the limitations and potential problems difficult.

3. Significant bias that may invalidate the results. A study with serious errors in design, analysis, or reporting. These studies may have large amounts of missing information or discrepancies in reporting.

Studies that reported multiple results of interest to this report could receive different quality grades for different outcomes if there were reporting or methodological issues with specific outcomes but not others. We did not grade the few non-randomized studies that were analyzed.

Applicability

Applicability addresses the relevance of a given study to a population of interest. Every study applies certain eligibility criteria when selecting study subjects. Most of these criteria are explicitly stated (i.e., disease status, age, sex). Some may be implicit or due to unintentional biases, such as those related to study country, location (e.g., community vs. specialty clinic), or factors resulting in study withdrawals. The question of whether a study is applicable to a population of interest (such as Americans) is distinct from the question of the study's methodological quality. For example, due to differences in the background diets an excellent study of Japanese men may be very applicable to people in Japan, but less applicable to Japanese-American men, and even less applicable to African-American men. The applicability of a study is thus dictated by the questions and populations that are of interest to those analyzing the studies.

In this report, the focus is on the US population, as specified in the Scope of Work for this series of evidence reports. We also address specific subgroups within that population (i.e., healthy Americans, Americans with CVD, and Americans with diabetes or dyslipidemia), as specified. To capture the potential applicability of studies to the different populations of interest as defined in the scope of work we define the following target population categories:

GEN General population. Typical healthy people similar to Americans without known CVD, diabetes or dyslipidemia.

CVD Cardiovascular disease population. Subjects with a history of or currently with cardiac, peripheral vascular, or cerebrovascular disease, as defined by the author. In addition studies of hypertensive patients were included.

DM Diabetic population. Subjects with any type of diabetes, including type I (DM I), type II (DM II), insulin dependent (IDDM) and non-insulin dependent (NIDDM), as defined by the authors.

DysLip Population with dyslipidemia, either elevated total cholesterol, LDL, or Tg, or low levels of HDL, as defined by the authors.

One study was classified as CVD Risk because it included a combination of subjects with known CVD, diabetes, dyslipidemia and other potential CVD risk factors. In addition, some studies received multiple classifications (CVD/DM or DM/DysLip), when inclusion criteria included multiple conditions.

Even though a study may focus on a specific target population, limited study size, eligibility criteria and the patient recruitment process may result in a narrow population sample that is of limited applicability, even to the target population. To capture this parameter, we categorize studies within a target population into 1 of 3 levels of applicability ⁴⁴:

1. Sample is representative of the target population. It should be sufficiently large to cover both sexes, a wide age range, and other important features of the target population including baseline dietary intake broadly similar to that of the US population.

2. Sample is representative of a relevant sub-group of the target population, but not the entire population. For example, while the Nurses Health Study is the largest such study and the results are highly applicable to women, it is nonetheless representative only of women. A fish oil study in Japan, where the background diet is very different from that of the US, would also fall into this category.

3. Sample is representative of a narrow subgroup of subjects only, and not well applicable to other subgroups. For example, a study of male college students or a study of a population on a highly controlled diet.

In the summary tables, each study receives a combined applicability grade comprised of the target population (GEN, CVD, DM, and DysLip) and the 3-level grade (I, II, III).

Sample Size

The study sample size provides a quantitative measure of the weight of the evidence. In general, large studies provide more precise estimates of effect and associations. In addition, large studies are more likely to be generalizable; however, large size alone does not guarantee broad applicability.

Overall Effect ^{48, 49, 52, 53, 60–78}

Across the 23 studies there was a wide range of effects of omega-3 fatty acids on total cholesterol, although in most studies the net effect was small and generally of an increase in total cholesterol. Most studies found net increases of between 0% and 6% (approximately 0 to 14 mg/dL). Only 3 studies found that the changes in total cholesterol in subjects on omega-3 fatty acids were significantly different than control. Notably, the directions of the treatment effects were not consistent across these studies.

Sub-populations

Only 5 of the studies included generally healthy subjects, 3 of which were all male^{66, 67, 72}. Net effects were generally small but inconsistent in direction. Most of the studies included subjects with a variety of types of CVD. There was no clear consistent effect among the 12 studies. Two studies evaluated subjects at increased risk of CVD with different sets of treatments and came to different conclusions. Sirtori et al. found no effect with fish oil in approximately 900 individuals with dyslipidemia and either hypertension, diabetes or glucose intolerance ⁷⁷. Singh et al. reported a large, highly significant reduction in total cholesterol with an Indo-Mediterranean diet in approximately 1,000 people with either hypercholesterolemia, hypertension, diabetes, angina or myocardial infarction ⁷⁶. However, this study found that subjects on the Indo-Mediterranean diet lost significantly more weight (3 kg) than those on the control diet. In addition, they reported uniform highly significant effects on all serum markers despite widely ranging effects. A number of statistical calculation errors were also found.

While no study evaluated a population of all diabetic subjects, Natvig et al., in an early Norwegian trial of linseed oil supplements, reported a sub-analysis of the 98 diabetic subjects and found that the effect of linseed oil was similar in both all subjects and specifically in diabetic subjects, but that total cholesterol decreased by a small amount more in the diabetic subjects ⁷². The difference was not significant.

Covariates

No subgroup analyses based on covariates were reported. Two studies performed regressions. Bairati et al. reported no change in total cholesterol effect after adjusting for age, sex, baseline lipid level, lipid treatment, body mass index and alcohol use ⁶⁰. Mori et al. performed a regression adjusting for change in weight and found a highly significant “group effect” increase in total cholesterol with omega-3 fatty acids (P < .001) ⁷¹. This study also found larger relative net increases in total cholesterol among subjects on a 40% fat diet, but no net effect (and a decrease in absolute change) in subjects on a 30% fat diet. No clear difference was seen between the 5 studies that included only men and the remaining studies ^{61, 66, 67, 71, 72}.

Dose and Source Effect

Three studies compared different sources - and doses - of marine oil supplements ^{62, 66, 74}. Grimsgaard et al. found a significantly greater decrease in total cholesterol with purified EPA than DHA in healthy, middle-aged men ⁶⁶. Brox et al. found a substantially greater decrease in total cholesterol with higher omega-3 fatty acid dose cod liver oil supplement than seal oil supplement in healthy subjects with elevated total cholesterol; although they imply that the difference was not statistically significant ⁶². Osterud et al. found varying degrees of net increases of total cholesterol with different marine oil supplements in healthy subjects ⁷⁴. No clear pattern was evident among different doses of omega-3 fatty acids and dose effect of marine oil supplements was evident across the studies.

Hanninen et al. compared 5 fish diets ⁶⁷. No significant effect on total cholesterol was seen with any diet and there was no dose effect based on frequency of fish consumption.

Among subjects on a higher fat diet, there was no clear difference in effect based on source of EPA+DHA among men studied by Mori et al. ⁷¹. Despite an apparent larger net increase in total cholesterol among subjects consuming both fish oil margarine and fish oil supplements compared to those consuming only fish oil margarine or rapeseed and linseed margarine, Finnegan et al. found no differences in effect among the treatments ⁵³.

The 4 studies of ALA all reported net increases in total cholesterol, but there was no apparent difference compared to fish and fish oil studies.

Exposure Duration

In 7 studies, total cholesterol levels varied by similar amounts in treatment and control arms at multiple time points ^{49, 53, 67, 69, 73, 75, 77}. No differences in effect were seen at times ranging from 5 weeks to 2 years. No effect across studies is evident based on duration of intervention or exposure.

Sustainment of Effect

No study reported data on an effect after ceasing omega-3 fatty acid treatment.

Overall Effect ^{48, 49, 52, 53, 60, 63–66 68–71 76, 79}

The effect of omega-3 fatty acid consumption was fairly uniform across studies. Most found a net increase in LDL with treatment, although the range of effects varied substantially. Most studies found net increases of LDL of 10 mg/dL or less, although the complete range of mean net effects was a decrease of 19 mg/dL to an increase of 21 mg/dL. As with a number of other outcomes, Singh et al. found a discordant result ⁷⁶. In this case, they reported a large, highly significant reduction in LDL with an Indo-Mediterranean diet in subjects at risk for CVD. However, as previously noted, this study found a difference in weight loss between the 2 interventions and reported uniform highly significant effects on all serum markers despite widely ranging effects; also, a number of statistical calculation errors were found.

Sub-populations

Only a single study included generally healthy subjects and no study included exclusively diabetics. Most of the studies included subjects with CVD. There was no clear difference among the 10 studies of CVD populations compared to the 3 dyslipidemia studies or single study of healthy subjects.

Covariates

No subgroup analyses based on covariates were reported. Two studies performed regressions. Bairati et al. reported that the effect of fish oil supplements on LDL (a net increase) was reduced and became borderline non-significant (P = .06) after adjusting for age, sex, baseline lipid level, lipid treatment, body mass index and alcohol use ⁶⁰. Mori et al. performed a regression adjusting for change in weight and found a highly significant “group effect” increase in LDL with omega-3 fatty acids (P < .001) ⁷¹. In contrast to their findings for total cholesterol, they reported similar effects on LDL among subjects on a 40% fat diet and on a 30% fat diet.

Dose and Source Effect

Mori et al. found no difference in effect among men consuming various doses of EPA+DHA either as supplements or as dietary fish ⁷¹. Finnegan et al. noted a particularly large increase in LDL in the fish oil margarine/fish oil supplement arm compared to other arms, but the differences were not statistically significant ⁵³. Grimsgaard found no difference in effect on LDL level between purified EPA and purified DHA ⁶⁶.

The 2 studies of ALA reported smaller net changes in LDL, but it is not clear that this represents a real difference in effect.

Exposure Duration

In 3 studies, LDL levels varied by similar amounts in treatment and control arms at multiple time points ^{49, 53, 69}. No differences in effect were seen at times ranging from 8 weeks to 2 years. No effect across studies is evident based on duration of intervention or exposure.

Sustainment of Effect

No study reported data on an effect after ceasing omega-3 fatty acid treatment.

Overall Effect ^{48, 49, 52, 53, 60, 62–66 68–71 73–76 79}

The effect of omega-3 fatty acid consumption was generally consistent across the 19 studies. Most found a small net increase in HDL with treatment of up to 3 to 5 mg/dL, although 7 found a small net decrease or no effect in at least one tested study arm. Six of the studies reported that the net increase in HDL was statistically significant.

Sub-populations

Across studies, there is no clear difference in effect among the 11 studies of CVD populations, the 4 studies of dyslipidemic patients, the 3 studies of healthy subjects, or the study of Indians at increased risk of CVD. No study included only diabetic patients.

Covariates

No subgroup analyses based on covariates were reported. Two studies performed regressions. Bairati et al. reported that the effect of fish oil supplements on HDL (a net increase) was reduced and became borderline non-significant (P = .06) after adjusting for age, sex, baseline lipid level, lipid treatment, body mass index and alcohol use ⁶⁰. Mori et al. performed a regression adjusting for change in weight and found a highly significant “group effect” increase in HDL with omega-3 fatty acids (P < .001) ⁷¹. In contrast with their findings for total cholesterol, they reported similar effects on HDL among subjects on a 40% fat diet and those on a 30% fat diet.

Dose and Source Effect

Three studies compared different sources - and doses - of marine oil supplements ^{62, 66, 74}. Grimsgaard et al. found a small difference in effect between purified EPA and DHA, but the net increase in HDL was significantly larger in men consuming DHA than those consuming EPA ⁶⁶. In studies by Brox et al. and Osterud et al., somewhat different net effects were seen with the different types of oils; however, neither study reported on whether the oils differed from each other on their effect on HDL ^{62, 74}. No dose effect of marine oil supplements was evident across the studies.

Mori et al. found no difference in effect among men consuming various doses of EPA+DHA either as supplements or as dietary fish ⁷¹. All doses and sources of omega-3 fatty acids resulted in significant increases in HDL. Finnegan et al. reported no difference in effect with different omega-3 fatty acid treatments ⁵³.

Only 2 studies tested ALA supplementation, with minimal effect.

Exposure Duration

Five studies reported data on time trends of HDL levels. Leng et al., de Lorgeril et al. and Finnegan et al. reported no difference in HDL levels at multiple time periods between 8 weeks and 2 years. ^{49, 53, 69}. In contrast, Nilsen et al. reported a steady increase in HDL in patients with recent myocardial infarctions who started fish oil supplements at 6 weeks (+8%), 6 months (+14%), and 12 months (+19%); patients on corn oil had variable HDL levels (-0.3%, +4%, and +7%, respectively). Sacks et al. reported that HDL levels were unchanged at 3 months in healthy subjects taking fish oil supplements compared to control - decreasing by about 1.5 mg/dL in both - but that HDL returned to baseline at 6 months, resulting in a small net difference compared to control. No clear effect across studies is evident based on duration of intervention or exposure.

Sustainment of Effect

No study reported data on an effect after ceasing omega-3 fatty acid treatment.

Overall Effect ^{48, 49, 52, 53, 60, 63–68 70, 71, 73, 74, 76, 77, 79, 81}

With few exceptions, Tg levels in the 19 studies decreased by substantial amounts in subjects taking omega-3 fatty acids, both in absolute amount and compared to control groups. The changes in Tg were generally highly significant.

Sub-populations

The 3 studies of healthy subjects, whose mean Tg levels were normal, generally found net decreases in Tg levels of about 10% to 25%. Eleven studies included subjects with a variety of types of CVD, all with mean Tg levels above 150 mg/dL. With the exception of Maresta et al., the 11 studies reported net decreases in Tg of between about 10% to 30%, most of which were statistically significant ⁸¹. There was no obvious difference between the study by Maresta et al. of patients undergoing PTCA and other studies to explain the discordant finding.

Two studies evaluated subjects at increased risk of CVD with different sets of treatments. Both of these studies found large, significant reductions in Tg. Two of 3 studies of dyslipidemic patients reported large net decreases in Tg of 20% or 33%. Finnegan et al., in a study of moderately hyperlipidemic patients, found different effects of omega-3 fatty acid consumption on Tg depending on dose and source ⁵³. No study evaluated a population of only diabetic subjects.

Covariates

Nilsen et al. found similar decreases in Tg among men and women, where the difference in significance level can be ascribed mostly to sample size ⁷³. Two studies that performed regressions both found no substantial change in the significant Tg reduction after adjusting for age, sex, baseline lipid level, lipid treatment, body mass index and alcohol use ⁶⁰ or change in weight ⁷¹. Grimsgaard et al. reported the effect of purified EPA and DHA on Tg in quartiles of baseline Tg ⁶⁶. While the authors did not discuss whether the effect of omega-3 fatty acids was associated with baseline Tg level, there does appear to be a trend toward greater reduction of Tg in subjects with higher baseline Tg. Those in the lowest quartile had a net reduction of approximately 7 mg/dL (10 – 14%); those in the middle two quartiles had net reductions of between 15 mg/dL and 27 mg/dL (14 – 30%); and those in the highest quartile (128 mg/dL – 319 mg/dL) had net decreases in Tg of about 50 mg/dL (about 28%). Across studies, the average net decrease in Tg level was larger in studies with higher mean baseline levels, as indicated by Figure 3.1, in which the meta-regression is not adjusted for dose of omega-3 fatty acid or study size. After adjusting for dose and the study variance, the association across studies remains statistically significant. In a separate analysis comparing different percentages of fat in the diet, Mori et al. also found nearly identical effects in subjects on 30% or 40% fat diets who were consuming similar amounts of omega-3 fatty acids ⁷¹.

Dose and Source Effect

The 4 studies that compared different doses of marine oil supplements found that the greatest net decrease in Tg level occurred in study arms receiving the highest dose of EPA+DHA, although none of the articles reported whether there was a significant trend within the study. Across studies there was a clear trend toward greater percent decrease in Tg with higher doses, regardless of source (Figure 3.2). At least a 10% reduction in Tg was found in most studies with doses of at least 1.7 g per day of marine oil supplementation. Most study arms with doses of at least 3 g per day of marine oil supplements resulted in at least a 20% reduction in Tg. Among the studies of dietary fish, only the 2 arms with high omega-3 fatty acid fish diets in Mori, et al. achieved at least a 20% reduction of Tg ⁷¹.

Grimsgaard et al., overall, found no difference in effect between purified EPA and purified DHA, although the net decreases in Tg were consistently greater in the DHA group than in the EPA group across quartiles of baseline Tg ⁶⁶. Across studies, and within the Mori et al. study ⁷¹, the source of the EPA+DHA, whether as a supplement or from dietary fish, does not appear to make a difference. In contrast, the effect of ALA is uncertain. The single study that evaluated pure ALA supplementation, Finnegan et al., found increases in Tg levels in subjects on both 4.5 g and 9.5 g per day of ALA margarine (the latter dose is not included in the summary table) ⁵³. Both Singh et al. and de Lorgeril et al. provided ALA in the context of a Mediterranean diet, which also included higher dietary fish intake ^{49, 76}.

Exposure Duration

The effect of duration of intervention or exposure was somewhat inconsistent among the 4 studies that reported data on Tg levels at different time points in studies of omega-3 fatty acids. Hanninen et al. found progressive decreases of Tg at 5 and 12 weeks in group of subjects consuming higher amounts of fish ⁶⁷. Similarly, Nilsen et al found progressive decreases in men, but not in a small group of women, at 6 weeks, 6 months and 12 months ⁷³. Sirtori et al. found that the effect of lower dose fish oil supplementation to reduce Tg occurred by 2 months and remained stable at 4 and 6 months ⁷⁷. In contrast, Finnegan et al. reported a significant decrease (15%) in mean Tg levels after 2 months which was not sustained at 6 months in the EPA+DHA arms ⁵³. Across studies, there is no apparent correlation between study duration and fish oil supplement effect, even after grouping studies by fish oil dosage.

Sustainment of Effect

No study reported data on an effect after ceasing omega-3 fatty acid treatment.

Overall Effect ^{49, 55, 58, 62, 83–92}

Across the 14 studies there is no consistent effect on Lp(a) levels of omega-3 fatty acid consumption compared to control. In approximately one-third of the studies the omega-3 fatty acid study arms had a net increase in Lp(a) level compared to control; in the remaining studies the net decrease in Lp(a) level was generally small and non-significant. Only 2 studies reported a statistically significant difference between the effect of omega-3 fatty acid and control, both of which found a net decrease in Lp(a). However, the variability of Lp(a) levels among subjects within all the studies resulted in wide confidence intervals which limited the likelihood of statistically significant findings.

Sub-populations

The 5 studies that evaluated generally healthy subjects found no consistent effect of omega-3 fatty acids on Lp(a). Marckmann et al. found a large net increase of Lp(a) with fish oil supplement use and Deslypere et al. found a large net increase of Lp(a) in 1 of 3 treatment arms ^{85, 89}. The remaining studies (and study arms) reported generally small effects, which were not uniform in direction. Five studies evaluated subjects with known CVD, one of which included only patients with hypertriglyceridemia on simvastatin. The apparent large decrease in Lp(a) in the latter study, Durrington et al., occurred because the median Lp(a) level rose by less in the fish oil supplement group than the corn oil group ⁸⁶. Again no consistent effect was seen. In the only study of diabetic subjects, Luo et al. found a statistically significant net reduction of Lp(a) of about 20% with fish oil supplementation ⁸⁸. The 4 studies of subjects with dyslipidemia (including the one with subjects with CVD on simvastatin) all found that subjects on marine oil supplements had a net decrease in Lp(a) compared to control; however, none of the changes was significant.

Eritsland et al. found that the effect on Lp(a) was not related to age or sex ⁸⁷. The 2 studies that excluded pre-menopausal women both found small, non-significant, net reductions in mean Lp(a) with fish oil supplements or fish diet ^{58, 83}. The 4 studies of men generally found small, non-significant, net increases in Lp(a) ^{84, 85, 89, 91}. No study included only women.

Covariates

As shown in the summary table, Eritsland et al. found a differential effect of omega-3 fatty acids based on baseline Lp(a) level in patients referred for coronary artery bypass graft surgery ⁸⁷. Those with Lp(a) in the upper quintile (≥ 20 mg/dL) had a small but significant absolute and net reduction in Lp(a), while the remaining subjects did not. A similar comparison between subjects with elevated baseline Tg (≥ 245 mg/dL) and those with lower Tg found no difference in effect.

Dose and Source Effect

Only 2 studies directly compared different doses of fish oil supplements or different oils. Deslypere et al. reported no effect on Lp(a) at any of 3 doses of fish oil supplements, although the mean Lp(a) level rose by almost 50% after 1 year in subjects on the highest dose ⁸⁵. Brox et al. found no difference between similar doses of cod liver oil and seal oil supplements ⁶². Across studies no differences could be discerned based on marine oil dose or omega-3 fatty acid-rich diet.

Exposure Duration

Two studies reported Lp(a) data at different time periods. de Lorgeril et al. found no difference in effect on Lp(a) at 8, 52, and 104 weeks in a study of Mediterranean diet ⁴⁹. Prisco et al. also found no difference in effect at 2 and 4 months in a study of fish oil supplements ⁹¹. Across studies there is no apparent relationship between effect and duration of intervention or exposure.

Sustainment of Effect

Both Prisco et al. and Deslypere et al. reported no difference between Lp(a) levels while subjects were on fish oil supplements and at multiple time points up to 6 months after stopping supplementation ^{85, 91}.

Overall Effect ^{48, 49, 52, 62, 66, 67, 85, 86, 88, 89, 93–109}

Across the 27 studies, effects of omega-3 fatty acids on apo A-I levels were generally heterogeneous but small. Most studies found a small net change in apo A-I with omega-3 fatty acid consumption. Three-quarters of studies found net changes between -5% and +5% (-7 to +10 mg/dL). No study found a large net increase in apo A-I level. A small number of studies found larger net decreases of up to 18% reductions (-33 mg/dL).

Sub-populations

Eight studies evaluated healthy people, all single-sex groups (7 male^{66, 85, 89, 95, 97, 100, 110}, 1 female⁹⁶), mostly of university students. Four studies evaluated diabetic patients. Thirteen studies evaluated patients with dyslipidemia, 2 of which were also of patients with CVD. There was one additional study of patients with CVD. There were no clear patterns of treatment effect or differences in effect among the sub-populations.

Covariates

Silva et al. reported that sex, body mass index, hypertension, and non-insulin dependent diabetes did not affect the fish oil or soya oil supplement effect on lipid parameters including apo A-I in hyperlipidemic subjects ¹⁰⁷. No other study evaluated correlations or sub-analyses based on apo A-I. Agren et al. (1988) compared the effect of daily fish with daily fish with a low saturated fat diet in male university students ⁹⁵. Among subjects on a fish and low saturated fat diet, apo A-I levels remained essentially unchanged compared to those on a regular diet. In contrast, subjects on a fish diet who were not told to lower their saturated fat intake had a significant net decrease in apo A-I that was among the largest net decreases across studies. However, no comparison was made between the 2 treatment groups, nor were any explanations for the difference examined or discussed. Three studies compared fish oil to placebo oil supplements in dyslipidemic patients who were all taking either atorvastatin or simvastatin ^{98, 99, 106}. The effects of fish oil supplementation on apo A-I were small in all 3 studies. The effects were not uniform in direction.

Dose and Source Effect

Neither Deslypere et al. nor Hanninen et al. reported a dose dependent effect on apo A-I of either fish oil supplements or different frequencies of fish meals ^{67, 85}. No dose effect was seen across studies of EPA+DHA either.

Five studies compared different sources of omega-3 fatty acids. Grimsgaard et al. found a small but significant net decrease in apo A-I with purified EPA compared to a smaller, non-significant, net increase with purified DHA; the difference between the 2 omega-3 fatty acids was statistically significant (P = .008) ⁶⁶. Brox et al. compared 2 sources of marine oil supplements: cod liver and seal oil ⁶². No effect was found with either treatment. Cobiac et al. found no treatment effect with either fish oil supplementation or with a fatty fish diet ¹⁰⁰. Silva et al. found similarly large, significant reductions in apo A-I level in subjects taking either fish oil or soya oil supplements; however, no non-omega-3 fatty acid was used as a control ¹⁰⁷. Agren et al. (1996) compared fish oil supplementation, algae DHA oil supplementation, and fatty fish diet and also found no difference in effect on apo A-I among the groups ⁹⁷.

Exposure Duration

Two studies reported apo A-I levels at multiple time points. Neither Hanninen et al. nor de Lorgeril et al. found any time-related effects of omega-3 fatty acids on apo A-I, at 5 and 12 weeks, and 8, 52, and 104 weeks, respectively ^{49, 67}.

Sustainment of Effect

Three studies followed subjects after stopping the intervention. Jensen et al. and Deslypere et al. found no change in apo A-I levels 8 weeks and 6 months, respectively, after stopping fish oil supplements ^{85, 103}. In contrast, Agren et al. (1988) reported that 5 months after a 15 week trial of dietary fish apo A-I levels remained at lowered levels in the fish diet group who had no limitation of saturated fat; however, they do not indicate what these students' diets were at subsequent follow-up ⁹⁵.

Overall Effect

Total apo B ( Table 3.8 ) ^{48, 49, 53, 66, 67, 71, 85, 86, 88–90, 93, 95–101, 103–106, 108, 109}. Across the 25 studies, we found little consistency in the effect of omega-3 fatty acids on apo B levels. About half the studies found a small net increase and half a small net decrease in apo B levels. Only 2 studies found significant changes in individual study arms, but Deslypere et al. found a significant decrease and Mori et al. found a significant increase ^{71, 85}.

Apo B-100 ( Table 3.9 , top) ^{50, 52, 62, 107} and LDL apo B ( Table 3.9 , bottom) ^{93, 94, 108, 111–113}. The 4 studies of apo B-100 found a range of effects with omega-3 fatty acid consumption. Two found a decreases in level of less than 5%; the other 2 studies found net increases of 2% and 15%. In contrast, large, significant net increases in LDL apo B were found in 4 of 6 studies (20 to 45 mg/dL).

Sub-populations

Total apo B. The heterogeneity of effects seen across all studies is apparent among the 10 studies of healthy populations (8 of which were in men^{66, 67, 71, 85, 89, 95, 97, 100} and one of which was in women⁹⁶), the 10 studies of dyslipidemic populations (subjects in 2 of which also had CVD), and the 3 studies of CVD populations (including those studies with subjects with dyslipidemia). The 4 studies of diabetics, one of which included insulin-dependent diabetics, all found small, non-significant, net increases in total apo B.

Apo B-100 and LDL apo B. The 2 apo B-100 studies of dyslipidemic patients reported small net decreases in apo B-100, while the study of patients undergoing coronary bypass surgery showed a small net increase and the study of healthy, male college students found a larger net increase in apo B-100. The 5 LDL apo B studies of dyslipidemic or diabetic subjects found generally large increases in LDL apo B, while the single study of hypertensive subjects showed a small net decrease.

Covariates

Total apo B. Nenseter et al. performed a subanalysis based on age of the effect of a low-omega-3 fatty acid fish powder ⁹⁰. Subjects between ages 30 and 52 years had a significantly greater rise in apo B level compared to subjects 53 to 70 years old; furthermore age negatively correlated with the rise in apo B (r = -0.40, P < .04). The authors also imply that the effect was not correlated with sex. Mori et al. performed a regression adjusting for change in weight and found a highly significant “group effect” increase in apo B with omega-3 fatty acids (P<.01) ⁷¹. Agren et al. (1988), in a study of male university students, found no difference in effect between 2 fish diets that differed in the amount of low saturated fats ⁹⁵. Three studies compared fish oil to placebo oil supplements in dyslipidemic patients who were all taking either atorvastatin or simvastatin ^{98, 99, 106}. The effects of fish oil supplements on apo B were small in all. They were not uniform in direction.

Apo B-100 and LDL apo B. Silva et al. reported that any effect of fish oil and soya oil supplements on apo B was not correlated with sex, BMI, hypertension, or diabetes in hyperlipidemic patients ¹⁰⁷. Schectman et al. found that changes in LDL apo B did not correlate with baseline differences in diet or with individual changes in diet or body weight ⁹³. Other studies did not correlate findings with possible covariates. The small number of studies limits hypothesis generating of possible effect mediators across studies.

Dose and Source Effect

Total apo B. Among studies of fish oil supplements, Deslypere et al. found a significant net decrease in apo B in subjects on the highest dose of omega-3 fatty acids but smaller non-significant net decreases with smaller doses ⁸⁵. Among the individual study arms, apo B levels fell in the arm with a higher dose of fish oil but rose in the lower dose arms (and the olive oil arm). No dose effect was seen across fish oil supplement studies. Among studies of dietary fish, Hanninen et al. reported a trend in effect related to different frequencies of fish meals ⁶⁷. Subjects most frequently consuming fish had the largest, significant reduction in apo B (compared to baseline). Subjects with intermediate frequencies of fish consumptions (average of 1.5 and 2.3 meals per week) had smaller reductions in apo B with P values (compared to baseline) of less than .10. Subjects on only about 1 fish meal per week had a non-significant increase in apo B.

Five studies compared different sources of omega-3 fatty acids. Grimsgaard et al. found no difference in effect between purified EPA and purified DHA ⁶⁶. Mori et al. compared a variety of doses of fish oil supplements and combinations of dietary fish and supplemental fish oil, along with higher and lower percentage fat diets ⁷¹. Overall, significant net increases in apo B were seen in the subjects who consumed fish oil supplements and were on non-fish diets, but smaller, non-significant increases were seen in the subjects who were on fish diets, regardless of fish oil supplementation or percent fat in the diet. Cobiac et al. similarly found that subjects on fish oil supplement had a net increase in apo B while those on dietary fish had almost no change ¹⁰⁰. While neither change was statistically significant, there was a trend toward a difference between the 2 treatments (P = .10). In contrast, Agren et al. (1996) reported small non-significant net reductions in apo B with fish oil and algae DHA oil supplementation and no effect with fatty fish diet; although they do not comment on potential differences between groups ⁹⁷. Finally, Finnegan et al. reported no effects on apo B and no differences among people consuming different omega-3 fatty acids from margarine and/or supplements ⁵³.

Apo B-100 and LDL apo B. Neither Brox et al. nor Silva et al. found a difference in effect of different omega-3 fatty acids on apo B-100 levels ^{62, 107}. Radack et al. (1990) found a similar large increase in LDL apo B in 2 groups of hypertriglyceridemic patients consuming different doses of fish oil supplements ¹¹³. While the increase was greater in the group consuming a higher dose of fish oil, no analysis was done to compare the effect in the 2 arms.

Exposure Duration

Total apo B. While the authors do not describe an effect of duration of fish consumption, the data at 5 and 12 weeks in Hanninen et al. may suggest that any effects of dietary fish on apo B do not occur until after 5 weeks ⁶⁷. At 5 weeks there were essentially no changes in apo B in any of the study arms, compared to significant and near significant reductions in arms with more frequent fish consumption. In de Lorgeril et al. a Mediterranean and ALA margarine diet had no effect on apo B at 8 weeks, 1 year, and 2 years.

Apo B-100 and LDL apo B. In their study of apo B-100, DeLany et al. found that while there was no difference in effect between 5 g fish oil supplementation and no oil at 5 weeks, there was a significant increase over time at 0, 2, and 5 weeks in subjects on fish oil supplements ⁵⁰. However, this analysis included 5 subjects who took 20 g fish oil supplements. There was also a small increase in apo B-100 levels in subjects not consuming oil supplements. Radack et al. (1990) reported no change in LDL apo B level between measurements at 8, 12, and 20 weeks ¹¹³.

Sustainment of Effect

Total apo B. Three studies followed subjects after stopping the intervention. Both Jensen et al. and Agren et al. (1988) found no change in apo B levels 8 weeks and 5 months, respectively, after stopping fish oil supplements ^{95, 103}. Deslypere et al. found that 6 months after stopping supplements apo B levels rose to similar levels in all groups except those who had been on the lowest dose fish oil, although no analysis was performed on follow-up data ⁸⁵.

Apo B-100 and LDL apo B. Although Radack et al. (1990) measured LDL apo B levels 4 weeks after stopping treatment ¹¹³, no study reported whether changes in apo B-100 or LDL apo B levels were sustained.

Meta-Regression ¹¹⁴

Geleijnse et al. collected trials of fish oil supplementation and blood pressure through March 2001. Eligibility criteria were: (1) randomized design, (2) adult study population, and (3) publication after 1966. Trials were excluded if they included sick or hospitalized patients, including kidney disease and diabetic patients, or if the intervention was shorter than 2 weeks duration. A total of 36 trials with 50 omega-3 fatty acid study arms were analyzed. Of note, 6 of these studies did not meet our eligibility due to high omega-3 fatty acid dose (3), short washout period in crossover trial (2), or short study duration (1).

The range of trial duration was 3 to 52 weeks and doses of omega-3 fatty acids were less than 1.0 g/day in 1 trial, 1.0 to 1.9 g/day in 5 trials, 2.0 to 2.9 g/day in 4 trials, and 3.0 to 15.0 g/day in 26 trials.

The mean net reduction (controlling for placebo arms) in systolic and diastolic blood pressure, weighted for study size, was -2.1 mm Hg (95% confidence interval -3.2, -1.0) and -1.6 mm Hg (-2.2, -1.0), respectively. The mean reductions in systolic and diastolic blood pressures were somewhat smaller in the 22 double blinded studies. Data on univariate and multivariate weighted meta-regression analyses performed on study subgroups based on mean age, sex, mean baseline blood pressure, and mean body mass index are reported. Briefly, systolic and diastolic blood pressure reductions were significantly larger in older (mean age ≥ 45 years) than younger populations, and in hypertensive (blood pressure ≥ 140/90 mm Hg) compared to normotensive populations. A lack of studies in women precluded adequate analysis based on sex. Body mass index was not associated with blood pressure response to fish oil supplementation. In addition, trial duration and fish oil dose were not associated with effect.

Overall Effect in Diabetes Studies ^115–120

Across the 6 studies of diabetic patients, there were generally small, non-significant effects of fish oil supplements on systolic (Table 3.10) and diastolic (Table 3.11) blood pressure. Overall, these study results were similar to the findings of the meta-regression among non-diabetic populations in their small, but generally inconsistent net effects. One study reported a small significant reduction in mean diastolic pressure (-2 mm Hg) and 2 reported significant reductions in mean systolic pressure (-3 and -6 mm Hg).

Covariates

Haines et al., who found non-significant small net increases in blood pressure, reported that neither sex nor Hgb A_1c levels were related to the effect of fish oil supplements on blood pressure ¹¹⁵. No study analyzed data based on age. Across studies there was no clear difference among populations with type I or type II diabetes, and there were insufficient data to comment on age, sex, menopausal status, race, weight or other variables.

Dose and Source Effect

No study compared different doses of omega-3 fatty acids. Woodman et al. compared purified EPA and purified DHA and found a net fall in mean 24 hour ambulatory systolic blood pressure in subjects on EPA and a net increase in diastolic pressure; however, there was no statistical difference between the 2 treatments ¹²⁰. Across studies, there is no apparent difference in effect on systolic blood pressure based on fish oil supplement dose. However, the largest, and significant, reductions in diastolic pressure were found in the 2 studies with the smallest fish oil supplementation doses.

Exposure Duration

In 3 studies no differences in effect are noted based on duration of intervention or exposure at 3 and 6 weeks ¹¹⁵, 6 and 12 weeks ¹¹⁸, or 6 and 12 months ¹¹⁹.

Sustainment of Effect

No study reported blood pressures after subjects stopped treatment.

Overall Effect ^{77, 85, 88, 93, 102, 103, 106, 115, 117–126}

Across the 18 studies, omega-3 fatty acids had a very small, if any, effect on Hgb A_1c levels compared to control. The range of net effects across the studies was -0.4% to +1.0%. Only 1 study reported a statistically significant reduction in Hgb A_1c; however, this study by Jain et al. found one of the smaller net changes of all studies ¹¹⁷.

Sub-populations

As expected, the large majority of studies evaluating Hgb A_1c included diabetic patients. Fourteen studies analyzed diabetic populations, 3 of which were also dyslipidemic. An additional 2 studies analyzed dyslipidemic patients; 1 included patients with untreated hypertension; and 1 evaluated healthy monks.

While none of the 4 studies of dyslipidemic patients had net reductions in Hgb A_1c levels, given the small differences in almost all studies, there are no clear difference in effect in the different populations, including diabetic patients.

Covariates

Schectman et al. found that the effect of fish oil supplements on Hgb A_1c did not correlate with baseline differences in diet or with individual changes in diet or body weight ⁹³. Toft et al. and Westerveld et al. reported no change in effect of fish oil supplements on Hgb A_1c after adjustment for body weight ^{125, 126}. Likewise, Haines et al reported no relationship between effect on Hgb A_1c and sex ¹¹⁵. Three studies were notable for including only men ^{85, 88}, or because all subjects were taking simvastatin ¹⁰⁶. The effect found in these studies was not clearly different than that found in studies.

Dose and Source Effect

Two studies compared different doses of fish oil supplements. Deslypere et al., in a 1 year study of healthy Belgian monks, reported no difference in the effect of 3 doses of fish oil or olive oil ⁸⁵. Westerveld et al. also reported no difference in the effect of 2 different doses of fish oil, purified EPA, or olive oil in non-insulin dependent diabetics ¹²⁶. Across studies, there was no apparent dose effect of fish oil supplements. The only study of dietary fish found a lack of effect similar to the fish oil supplement studies. Woodman et al. compared purified EPA to DHA in type II diabetics ¹²⁰. No difference was noted between the 2 treatments.

Exposure Duration

Two studies reported treatment effect at multiple time points. In Haines et al. there was a transient drop in Hgb A_1c by 0.6% (0.5% net) at 3 weeks which reverted to baseline at 6 weeks ¹¹⁵. The change was not statistically significant. Rossing et al. found no difference in effect between 6 and 12 months ¹¹⁹. Across studies there was no apparent effect of treatment duration.

Sustainment of Effect

Jensen et al., in a crossover study, found that Hgb A_1c remained unchanged 8 weeks after stopping oil supplementation.

Overall Effect ^{52, 53, 68, 76, 77, 103, 116, 117, 120, 123, 125, 127–132}

The effect of omega-3 fatty acids on FBS was inconsistent across the 17 studies. Four studies found large and/or near-significant net increases in FBS compared to control; 3 found large and/or significant net decreases in FBS and the rest found small non-significant changes. Across the studies, the net effect ranged between a decrease of 29 mg/dL over 8 weeks and an increase of 25 mg/dL over 6 weeks. Interpretation of the overall data is further complicated by inconsistent patterns of effect within individual study arms. In omega-3 fatty acid arms and in control arms, FBS increased from baseline in half the arms and either decreased or remained unchanged in the other half.

Sub-populations

Seven studies evaluated diabetic populations, 2 of which also had dyslipidemia; an additional 5 studies evaluated patients with dyslipidemia. Three studies included subjects who had CVD or were at increased risk for CVD (due to either diabetes or dyslipidemia). Two studies were of healthy populations.

The findings within the diabetic populations were inconsistent. The largest net decrease in FBS was found by Jensen et al. in the only study of insulin-dependent diabetics ¹⁰³, while the largest net increase in FBS with omega-3 fatty acids was seen in Woodman et al. in one of the studies of type II diabetics ¹²⁰. Furthermore in each of the 3 groups of subjects on fish oil supplements in these 2 studies, FBS rose by approximately 10 or 20 mg/dL; the large difference in net effect is due to the difference in effect of the olive oil control (+49 mg/dL and -7 mg/dL, respectively). In the remaining studies of diabetics, the change in FBS was in the same direction in omega-3 fatty acid arms and control arms; in 6 omega-3 study arms FBS rose from 10 mg/dL to 23 mg/dL; in 4 arms FBS fell from -2 mg/dL to -16 mg/dL. In studies of diabetics, factors other than omega-3 fatty acid consumption - such as those related to population characteristics, other treatments, or study design - appear to have had a greater effect on change in FBS than the omega-3 fatty acid treatment itself.

Among the 7 studies of dyslipidemic populations, 2 of which were also diabetic, all found a small non-significant net effect of omega-3 fatty acids on FBS that ranged from -4 to +5 mg/dL. Only Dunstan et al. found large changes in individual omega-3 fatty acid arms, which were related primarily to exercise level and were similar to the changes in the no fish control arms ¹²⁷.

The 4 studies of CVD patients or people with an elevated risk of CVD all found small absolute and net changes in FBS with omega-3 fatty acid consumption. Only Singh et al. found a significant net change and had a relatively large absolute change (-8 mg/dL) in FBS, although notably about 20% of the subjects were diabetic, two-thirds were vegetarian, and those subjects on the Indo-Mediterranean diet on average lost 3 kg more weight than controls ⁷⁶. In addition, this study reported uniform, highly significant effects on all serum markers despite widely ranging effects. A number of statistical calculation errors were also found.

The single study of a healthy population, by Freese et al., found small differences in FBS with 2 different omega-3 fatty acid treatments (in opposite directions) ¹²⁸.

Covariates

Schectman et al. found that changes in FBS did not correlate with baseline differences in diet or with individual changes in diet or body weight ⁹³. Two studies of diabetics reported data on associations between effect and other variables. Hendra et al. reported that the change in FBS was unrelated to change in weight ¹¹⁶. Woodman et al. reported that the significant effect compared to olive oil was unchanged after adjusting for age, sex, and BMI ¹²⁰. In Mori, et al. (1999), a study of obese hypertensive subjects, the direction of the absolute and net changes in FBS appear related to whether subjects were on a weight-reduction diet or not (those on a weight maintaining diet had increases in FBS, while those on a weight-reduction diet had reductions in FBS); however, they reported no interaction between fish diet and weight loss on FBS ¹³¹. No patterns across studies are evident based on reported data on covariates.

Dose and Source Effect

No study directly compared doses of the same source of omega-3 fatty acids. In comparisons of EPA and DHA, Woodman et al. reported no difference in effect on FBS ¹²⁰; however, Mori et al. (2000) reported a trend toward increasing FBS with EPA, but no change with DHA ¹³². Freese et al. reported a significant increase from baseline in FBS with fish oil supplementation compared to no change with linseed oil; however the difference between the 2 treatments was reported to be non-significant ¹²⁸. In a comparison of multiple sources of omega-3 fatty acids, Finnegan et al. found no significant differences in effect between various doses of either fish oils or plant oils ⁵³. Across studies, there was no discernable difference in effect based on either fish oil dose or omega-3 fatty acid source among diabetic or dyslipidemic populations.

Exposure Duration

Two studies measured FBS levels at multiple time points. Hendra et al. found that FBS rose with fish oil supplements at both 3 and 6 weeks, although the net difference with control was significant only at 3 weeks ¹¹⁶. In a longer study that measured FBS at 2, 4, and 6 months, Finnegan et al. found no treatment effect at any time period ⁵³. The heterogeneity does not appear to be related to study duration.

Sustainment of Effect

Jensen et al., in a crossover study which found that FBS rose by large amounts in both the high-dose cod liver oil and olive oil supplement arms, found that FBS fell back near baseline levels 8 weeks after stopping oil supplementation, although none of the levels were significantly different from each other ¹⁰³. Freese et al., who compared fish oil to linseed oil supplements, reported that FBS, which had risen in the fish oil arm, returned to baseline during a 12 week follow-up period ¹²⁸.

Overall Effect ^{52, 53, 68, 77, 88, 89, 106, 120, 122, 125, 129, 131–134}

Baseline levels of fasting insulin varied broadly across studies. In general, studies of non-insulin-dependent diabetics and obese subjects (under “Studies of “Hyperglycemic” Subjects”) had higher mean insulin levels than dyslipidemic, hypertensive, or healthy patients (under Studies of “Euglycemic” Subjects). However, within each population grouping the range of insulin levels remained broad. Mean insulin levels varied within studies also. In 6 studies, baseline insulin levels differed between omega-3 fatty acid arms and control arms by 20% to 60%. Among these, Toft et al. reported a significant difference at baseline and Chan et al. reported no significant difference; the remaining studies did not comment ^{125, 133}. In an attempt to standardize across studies, given the large variation in insulin levels, we calculated net differences in terms of percent change from baseline instead of absolute changes.

Across the 15 studies there were a wide range of apparent treatment effects ranging from net changes of -28% to +29% (or -22 pmol/L in Dunstan et al. ¹²² to +34 pmol/L in Chan et al. ¹³³). Approximately one-third of the omega-3 fatty acid study arms had net percent changes of either greater than +10%, between -10% and +10%, or less than -10%.

Sub-populations

Nine of the studies reported data on essentially euglycemic populations. The remaining 6 studies evaluated diabetic or obese populations in which the fasting insulin level may be of less value. While the studies with hyperglycemic subjects all had elevated mean fasting insulin levels, there was a wide range of mean insulin levels in the studies of euglycemic subjects.

Among the studies of euglycemic subjects, the heterogeneity of effect was similar to the heterogeneity seen across all studies. The heterogeneity was particularly apparent among the studies of dyslipidemic patients.

Covariates

Among the studies of euglycemic subjects, Mori et al. (1999) reported no interaction between dietary fish intake and weight loss on insulin levels ¹³¹. However, a weight loss diet resulted in a reduction of insulin levels, regardless of fish consumption. In addition, there was a net decrease in insulin levels in subjects who were on a weight loss diet with fish compared to a net increase in insulin in subjects who were on a weight-maintaining diet. Otherwise, studies did not attempt to correlate the effect on insulin of covariates. The 3 studies that either included only euglycemic men ^{89, 132} or excluded pre-menopausal women ¹³¹ had a wide range of effects on insulin levels. Thus, no potential sex effect could be seen.

No study of hyperglycemic subjects reported a correlation between insulin and covariates. As in studies of euglycemic subjects the effects on insulin found among the 2 studies of hyperglycemic men ^{88, 133} and the study that excluded pre-menopausal women ¹²⁰ were heterogeneous.

Dose and Source Effect

Finnegan et al. compared plant oil margarine to 2 doses of fish oil (as margarine and as both margarine and supplement) and to omega-6 fatty acid margarine ⁵³. None of the differences in insulin levels was statistically significant and the article does not comment on the relative effects of different treatments. However, dyslipidemic subjects on ALA margarine had an absolute and net decrease in fasting insulin, while subjects on low dose fish oil had a small absolute increase in insulin that was less than the increase in the control group, and subjects on high dose fish oil had an increase in insulin similar to controls. Across the studies, the effect on insulin does not appear to be associated with fish oil dose.

Both Mori et al. (2000) and Woodman et al. compared purified EPA to DHA, although in different populations ^{120, 132}. No difference was noted between the 2 treatments in both studies.

Exposure Duration and Sustainment of Effect

Only Finnegan et al. measured insulin levels at multiple time points ⁵³. They reported no treatment-time interaction with insulin levels at 2, 4, and 6 months. No study measured insulin levels after ceasing omega-3 fatty acid consumption.

Overall Effect ^{56, 99, 136–138}

No study found a significant effect of omega-3 fatty acid consumption on CRP level. However, CRP levels increased relative to subjects who were on control oils in most study arms among the 4 randomized trials. In contrast, the cross-sectional study did find that CRP levels were lower among subjects who ate fish regularly (fish score >4) but the difference was not statistically significant.

Sub-populations and Covariates

No study directly compared the effect of omega-3 fatty acids with placebo in different populations. There was no clear difference in effect across studies based on population. Baseline CRP levels varied across studies; although the reason for the different CRP levels is not apparent. Madsen et al. reported that when the 11 subjects with baseline CRP greater than 2 mg/L were analyzed separately, no difference in effect was seen with fish oil supplementation (as in all subjects) ¹³⁷. Likewise, the effect of omega-3 fatty acids does not appear to differ across studies based on average baseline CRP.

The trial by Chan et al. was a factorial study with fish oil supplements and atorvastatin (40 mg/day) in obese men who had a substantially higher baseline CRP than a separate group of 10 lean men (0.49 mg/L) ¹³⁹. While atorvastatin did significantly reduce CRP levels (by 0.73 mg/L) there was no interaction with fish oil.

Dose and Source Effect

No study compared different sources of omega-3 fatty acids. Any differences in effect due to differing sources across studies could not be appreciated among the few studies. The cross-sectional study did not find an association between fish score (amount of fish in diet) and CRP level.

Exposure Duration

Junker et al. evaluated CRP levels at both 2 and 4 weeks. No differences were noted between baseline and either 2 or 4 weeks ⁵⁶. Mezzano et al. evaluated CRP levels at 30 days and 90 days (and also at 60 days after 30 days of added red wine). CRP was unchanged at all observation points.

Sustainment of Effect

No study re-examined CRP after subjects stopped taking omega-3 fatty acids.

Overall Effect ^{46, 56, 69, 74, 85, 89, 90, 100, 115, 116, 138, 140–152}

Across the 24 studies there was no consistent effect on fibrinogen levels of omega-3 fatty acid consumption compared to control. Approximately half the omega-3 fatty acid study arms resulted in a net increase in fibrinogen level compared to control; in the other half there was either a net decrease or no effect on fibrinogen level. Only 4 studies reported a statistically significant difference between the effect of omega-3 fatty acid and control. In 3 of these, the net decrease of fibrinogen ranged from approximately 5% to 20%. One study reported a significant net increase of fibrinogen of 11%.

Sub-populations

Thirteen of the studies evaluated generally healthy subjects. No consistent effect was found specifically in this population. Four studies evaluated subjects with known CVD: 2 studies of patients with stable claudication (Gans et al. and Leng et al.) ^{69, 144}, one of patients who were undergoing coronary bypass (Eritsland et al.) ¹⁴², and one of subjects with hypertension (Toft et al.) ¹⁵². All 4 studies found no effect of omega-3 fatty acids on fibrinogen levels. Seven studies included subjects with diabetes and/or dyslipidemia. Again, there was no consistent effect. However, the largest (significant) net decrease in fibrinogen was found by Radack et al. in a group of 10 subjects with hyperlipoproteinemia types IIb or IV on a moderate dose of fish oil supplement ¹⁵¹. A significant net increase in fibrinogen was seen by Haines et al. among 19 subjects with insulin-dependent diabetes on a high dose of fish oil supplement, although the effect was not related to Hgb A1c level. ¹¹⁵.

In the study of patients undergoing coronary bypass, Eritsland et al. found that the (lack of) effect of omega-3 fatty acids on fibrinogen was unchanged after adjusting for multiple factors including age and sex ¹⁴². Seven studies included only men ^{46, 85, 100, 138, 140, 147, 149}. The distribution of effects was similar in this subset of studies as in the whole set. Three of these studies of men and an one additional study included only younger adults (generally less than 30 or 40 years old) ^{46, 138, 140, 146}. These studies had results similar to studies of broader age ranges or of older subjects. Overall, the studies provided insufficient data on race or ethnicity to allow analysis of these subpopulations. Almost half the studies were performed in Scandinavia and Finland; most of the remaining are from northern Europe and Australia. Notably the study by Radack et al., which showed the largest benefit from omega-3 fatty acids and was the only study to show a dose effect (see below), was the only study performed in the United States ¹⁵¹.

Covariates

Eritsland et al., Haines et al. and Toft et al. found no association of effect of omega-3 fatty acids on fibrinogen with various factors including sex, baseline and change in weight, baseline blood pressure, change in lipids or insulin, or cardiovascular, lipid or antithrombotic drug use among patients with cardiovascular disease ^{115, 142, 152}. Mezzano et al. found no interaction of wine consumption with a Mediterranean diet in a multiphase trial ¹³⁸. No differences were found among studies with run-in phases of either high- or low-fat diets. No study quantified baseline fish consumption. Radack et al. reported that the relative effect of higher dose fish oil supplements was greater with higher baseline fibrinogen values (r = -0.59, P < .01) ¹⁵¹.

Dose and Source Effect

Two studies compared different doses of the same omega-3 fatty acid supplements. Radack et al. found that subjects with dyslipidemia who took 6 g of fish oil supplements (2.2 g EPA+DHA) for 20 weeks had a relatively large, statistically significant net reduction in fibrinogen ¹⁵¹. This effect was significantly greater than in the subjects who took 3 g of fish oil (1.1 g EPA+DHA), who had no effect. Deslypere et al., however, found no difference in effect across 3 doses of fish oil supplements (3.4 g, 2.2 g, and 1.1 g EPA+DHA) in monks who took fish oils for 1 year. Across all studies the effect is not related to omega-3 fatty acid dosage.

Hansen et al. (1993a) reported a possible trend toward greater effect of fish oil ethyl esters than fish oil triglycerides ¹⁴⁷. Osterud et al. found no difference among different marine oils ⁷⁴. Two studies evaluated ALA oils. Both found no effect with dietary flaxseed oil or rapeseed oil supplements ^{46, 56}.

Three studies compared fish oil supplements with other sources of omega-3 fatty acids ^{100, 140, 143}. Cobiac et al. found a small significant reduction in fibrinogen only among the subjects consuming dietary fish; however the significance of the difference between the 2 treatments was not reported ¹⁰⁰. Overall, there were no clear differences in effect of different sources of omega-3 fatty acids.

Exposure Duration

Across studies, there was no apparent effect on fibrinogen of duration of consumption of omega-3 fatty acids in studies that reported data from 2 weeks to 2 years. Seven studies reported fibrinogen levels at various time points ^{56, 69, 85, 115, 138, 149, 151}. Although mean fibrinogen levels varied with time in most studies, no study found a difference in effect related to time.

Sustainment of Effect

Two studies, which both found no effect of omega-3 fatty acids on fibrinogen levels, reported no further change after stopping treatment. Deslypere et al. reported no difference in fibrinogen levels up to 6 months after 1 year of treatment ⁸⁵. Freese et al. likewise found no difference 4 weeks after finishing 4 weeks of treatment ¹⁴³.

Overall Effect

Factor VII (Table 3.17) ^{46,56,74,89,90,115,116,138,140–143,145–147,149,150,152,155}. There is little consistency in effect across the 19 studies of factor VII activity. In general, the net change in factor VII in subjects consuming omega-3 fatty acids is small (7% change from baseline or less), although a nearly equal number of studies found net increases as found net decreases in levels.

Factor VIII (Table 3.18) ^{46,84,85,115,138}. Five studies reported data on factor VIII activity. (It is unclear whether Conquer et al. measured factor VIII activity or antigen ⁸⁴.) There is no consistent effect across studies, with some finding a net increase and some a net decrease in factor VIII level.

von Willebrand Factor (Table 3.19)^{46,69,84,85,89,147,149,150,156}. Nine studies reported data on various measurements of vWF using different measurement methods. Some studies were not explicit about the specific measurement used. Most studies found a net decrease in vWF level (of up to a 13% reduction from baseline), although in only 1 study was the difference with placebo reported to be statistically significant.

Sub-populations

Factor VII. A small, inconsistent effect across studies was found among the 10 studies of a general population, the 3 studies of populations with CVD, and the 4 studies of people with dyslipidemia. The only statistically significant effects - both net increases in factor VII - were seen in 2 of the 3 studies of diabetic patients (one of which included only diabetics with dyslipidemia). The large increase in factor VII found by Hendra et al. in a 6 week study of fish oil versus olive oil supplements in non-insulin dependent diabetics was noted to be unexpected in light of a large decrease in Tg level ¹¹⁶.

Factor VIII. The single study of insulin dependent diabetics found a larger net increase of factor VIII than the studies of general populations, although the difference in this study was not significant. No study measured factor VIII in CVD or dyslipidemic populations.

von Willebrand Factor. With the exception of a low-dose arm in 1 study, the 6 studies of general populations found either net decreases or no effect in vWF, although none was statistically significant. The single study of a CVD population was the only study to find an overall net increase in vWF level, although Leng et al. was also an anomaly in that the oil analyzed was primarily gamma-linolenic acid (GLA, 18:3 n-6), an omega-6 fatty acid, with a small amount of EPA ⁶⁹.The only study to find a large, statistically significant decrease in vWF was 1 of the 2 studies of dyslipidemic patients. No study evaluated diabetic patients.

Covariates

Factor VII. Haines et al. found no association between change in factor VII with fish oil supplementation and either sex or Hgb A_1c in insulin dependent diabetics ¹¹⁵. In contrast, in a study of non-insulin dependent diabetics, Dunstan et al. reported a significant positive association between the changes in factor VII and fasting blood sugar with a fatty fish diet; however, dietary fish significantly affected factor VII levels only in subjects who were not in a moderate exercise program ¹⁴¹. Eritsland et al. reported no change in (lack of) effect of fish oil supplements in patients undergoing coronary bypass surgery after controlling for multiple factors including age, sex, weight, blood pressure, diabetes and CVD medications ¹⁴².

In possible contrast to the rest of the studies, only 1 of the 6 studies of male subjects, 3 of which were of younger men, found a net increase in factor VII; however all effects were small ^{46, 89, 138, 140, 147, 149}. One study in which all subjects were on simvastatin ¹⁵⁰ found a non-significant effect of fish oil supplements similar to other studies.

Factor VIII. Haines et al. found no relationship between effect of fish oil supplementation in insulin dependent diabetics who were taking aspirin on factor VIII and either sex or Hgb A_1c ¹¹⁵. All other studies were in men, most of whom were under age 40 years. There were no other data relating to other covariates.

von Willebrand Factor. No study reported on correlations between effect on vWF and covariates. Notably, though, only 2 of the studies included women ^{69, 150}. The effect of fish oil supplements in patients on simvastatin was similar to the effect of fish oil alone in other studies ¹⁵⁰.

Dose and Source Effect

Factor VII. No study compared different doses of the same omega-3 fatty acid source. Across studies there does not appear to be a dose effect. Four studies compared different sources of omega-3 fatty acids. Hansen et al. (1993a) found no difference between fish oil triglycerides and fish oil ethyl esters ¹⁴⁷. Osterud et al. reported no difference in effect of different marine oils ⁷⁴. Freese et al. compared similar doses of fish oil and linseed oil supplements and found no difference between the 2 oils ¹⁴³. Agren et al. also did not report a difference in effect among fish oil supplementation, algae DHA oil supplementation, and fatty fish diet ¹⁴⁰.

Factor VIII. Only Deslypere et al. compared different doses of fish oil supplements ⁸⁵. They reported no difference in effect of fish oil on factor VIII related to dose. None of the studies of fish oil supplements showed more than a marginal decrease in factor VIII level. In contrast, the single study of a flaxseed oil diet found a non-significant, approximately 6% net decrease in factor VIII activity and the single study of Mediterranean diet found a highly significant, approximately 7% net reduction in factor VIII activity. In the latter study, Mezzano et al. also found significant reductions in factor VII activity and fibrinogen levels, in contrast to most other studies ¹³⁸. They found no association between the effect on factor VIII and either ABO blood type (which is related to factor VIII level) or CRP, as a marker of inflammation.

von Willebrand Factor. Deslypere reported no difference in effect on vWF after 1 year in monks taking 3 different doses of fish oil supplements ⁸⁵. Hansen found similar effects among men taking either fish oil triglycerides or fish oil ethyl ester ¹⁴⁷. Across studies, though, the study by Seljeflot et al., which tested the largest dose of omega-3 fatty acid supplementation, found the largest, significant decrease in vWF. However, the study of mackerel paste diet, with a similar omega-3 fatty acid level, found no effect. The single study of plant oils found a non-significant decrease in vWF with an ALA-rich flaxseed oil diet that was similar to most marine oil studies.

Exposure Duration

Factor VII. Five studies measured factor VII levels at different time periods, ranging from 2 to 16 weeks ^{56, 115, 138, 149, 155}. No differences were seen in factor VII levels at any time point.

Factor VIII. Three studies measured factor VIII activity at different time periods. Haines et al. found no effect of fish oil supplements on factor VIII at either 3 or 6 weeks ¹¹⁵. Deslypere et al. did find an occasional significant decrease of factor VIII from the second trial month on in multiple measurements done between 4 weeks and 12 months ⁸⁵. However, this effect was also seen in the olive oil group and no net differences were found. Mezzano et al. found similar responses to Mediterranean diet at both 1 and 3 months ¹³⁸.

von Willebrand Factor. Three studies measured vWF at different time periods. Muller et al. found no change in vWF in either study arm at both 3 and 6 weeks ¹⁴⁹. Both Deslypere et al. and Leng et al. found that vWF levels fluctuated at different time points ranging from 3 weeks to 1 year, but that there were no differences among arms ^{69, 85}.

Sustainment of Effect

Factor VII. Only Freese et al. reported data on factor VII levels after stopping treatment ¹⁴³. There was no difference 4 weeks after finishing 4 weeks of treatment compared to either pre- or post-treatment levels.

Factor VIII and von Willebrand Factor. Only Deslypere et al. reported data on factor VIII activity and vWF after stopping treatment ⁸⁵. There was a large increase in factor VIII activity in all study arms, including the olive oil group, at both 1 and 2 months after stopping treatment. There were no differences between fish oil supplement and control groups. There was no difference in vWF after treatment.

Overall Effect ^{54, 57, 108, 115, 116, 128, 140, 157–160}

Within the 11 studies, heterogeneous effects of omega-3 fatty acids were generally found depending on the aggregating agent, the dose of agent, and the measurement metric used. However, in most studies either no effect on platelet aggregation was found with omega-3 fatty acids or no difference in effect was seen between treatments and controls.

Sub-populations

Seven studies were performed in generally healthy individuals. Salonen et al., Junker et al., and Wensing et al. all found no effect of omega-3 fatty acid consumption and no difference with control groups in healthy men, non-obese individuals and elderly individuals, respectively ^{56, 159, 160}. Freese et al. (1994) found no significant effect from rapeseed oil supplements in male students; however, they did find an apparent comparative effect since Trisun sunflower oil, which was used as the comparison, significantly increased platelet aggregation ⁵⁴. Hansen et al., Freese et al. (1997a), and Agren et al. found mixed effects in younger individuals (Agren at al. in male students), with significantly decreased platelet aggregation in some study arms with some specific tests ^{128, 140, 157}.

Two studies evaluated hypercholesterolemic subjects, both of which found no effect of omega-3 fatty acids on measures of platelet aggregation. An additional 2 studies included diabetic patients. Haines et al. reported no effect among insulin-dependent diabetics, while Hendra et al. reported small, but significant increases in spontaneous platelet aggregation among type 2 diabetics ^{115, 116}. However, in the latter study it was also reported, without supporting evidence, that epinephrine-induced aggregation was unaffected by either treatment or control. No studies specifically included patients with known or suspected CVD.

Covariates

Hansen et al., recognizing that male and female sex hormones have different effects on platelet function, made an a priori evaluation of the potentially different effect of cod liver oil supplementation on platelet aggregation in men and women ¹⁵⁷. Healthy, young, normolipemic men and women were included in the study. A large, significant decrease in platelet aggregation with low dose collagen was seen in men on cod liver oil supplements, but not in women (P < .01 men vs. women). Otherwise the effect of fish oil was generally mixed and not different between the sexes. No explanation was offered for why the effect would have been seen only with low-dose collagen aggregation. In contrast, Haines et al. made the blanket statement that the baseline variables smoking, alcohol consumption, and sex were not related to the response to fish oil supplementation ¹¹⁵. Four other studies included only men ^{54, 57, 140, 159}. No clear difference was seen between these studies and studies that included both men and women. No other covariate was specifically analyzed in any study.

Dose and Source Effect

No study compared different doses of the same type of oil. Among the studies of fish oil supplements or diets, there was no clear association across studies between dose and change in platelet aggregation.

No significant effect was seen in any of the studies of plant oil supplements or diets, regardless of dose. Two studies compared fish oil (EPA+DHA) to linseed oil (ALA). Freese et al (1997a) was inconclusive regarding a difference between fish oil and linseed oil supplements ¹²⁸. However, Wensing et al. reported that platelet aggregation was prolonged by greater amounts in subject who consumed fish oil shortening compared to those who consumed linseed oil shortening ¹⁶⁰. Agren et al. compared 3 sources of EPA and/or DHA ¹⁴⁰. Collagen aggregation was reduced in subjects on both fish oil supplementation and fish diet, but not in those consuming pure DHA oil. From this, they concluded that while omega-3 fatty acids impair platelet aggregation, DHA is less potent than fish oil or dietary fish at moderate doses.

Exposure Duration

Three studies measured platelet aggregation at different time points. Haines et al. and Junker et al. reported data at 3 and 6 weeks, and 2 and 4 weeks, respectively, but did not comment on a potential time effect ^{56, 115}. However, no apparent difference in effect was seen between the earlier and later times. Kwon et al. noted that with 2 mg/L collagen aggregation a significant decrease in platelet aggregation was found at 3 weeks on canola oil diet, which reverted to baseline by 8 weeks ⁵⁷.

Sustainment of Effect

Freese et al. (1997a) reported that the decrease in collagen-induced aggregation in the fish oil supplement arm did not return to baseline during a 12 week follow-up period, although, the other tests did ¹²⁸.

Overall Effect ^{63, 64, 81, 161–169}

All studies compared a single dosage of fish oil supplementation to control. Definitions of restenosis, however, were not uniform as noted in the footnotes of the summary table. In particular, 3 studies included abnormal exercise tolerance tests (ETT) as a potential definition of restenosis ^{166, 167, 169}. The results of random effects model meta-analysis are presented in both the Table 3.21 and Figure 3.3. Overall, although there is heterogeneity among the studies, there is a trend toward a net reduction of coronary artery restenosis with fish oil supplementation. The meta-analysis estimate is a lowering of risk of 14% (95% confidence interval -29%, +3%).

Sub-populations and Covariates

Most studies included all patients who were undergoing first PTCA, therefore with known or suspected coronary artery disease. No study restricted eligibility to patients with either diabetes or dyslipidemia. A number of studies performed multivariate analysis including diabetic, lipid, and cardiovascular variables, generally finding no association between these covariates and restenosis in the randomized trials. Only Bairati et al. commented about the effect of multivariate analysis on the relative risk of restenosis from fish oil supplement treatment ¹⁶¹. The authors reported that after controlling for history of hypertension, myocardial infarction, and diabetes, and for smoking, body mass index, angina class, degree of stenosis, location and number of stenoses, and ejection fraction, the inverse association between fish oil supplementation and restenosis was stronger and of higher statistical significance (because of a higher risk profile in the fish oil group).

Reis et al. and Kaul et al. both compared relative risk of restenosis in men and women; neither found a significant difference in effect, although both found a higher (worse) relative risk in women than in men ^{166, 169}. In men, the relative risks of restenosis were 1.33 and 1.29, respectively, compared to 2.20 and 1.78 in women. Notably, though, these 2 studies had the lowest control rates (the rate of restenosis in the control arm, a commonly used metric to estimate the underlying severity of disease) and were the only 2 studies with relative risks substantially greater than 1.0. Interestingly, the 1 study which was restricted to men, Dehmer et al., had about the lowest relative risk of restenosis among the studies.

Dose and Source Effect

No study compared doses of fish oils and all evaluated only fish oil. Across studies, no effect is apparent based on dose of fish oil supplement.

Exposure Duration

Each study evaluated restenosis at one time point only. Across studies, the duration of treatment does not appear to correlate with the relative risk of restenosis. In fact, both the longest study ¹⁶⁸ (12 months) and the shortest study ¹⁶³ (approximately 3–4 months) had similarly, low and statistically significant relative risks of restenosis.

Sustainment of Effect

No study re-evaluated for restenosis after stopping treatment.

Overall Effect ^{51, 79, 172, 173}

The only placebo-controlled randomized trial found small, non-significant net thickening of carotid IMT, using 4 different measurements at 24 months, with fish oil supplementation. The uncontrolled cohort of subjects consuming ALA margarine had a significant thickening in IMT at 2 years. However, the absolute change in IMT in this cohort of subjects was similar to the absolute change in IMT in the fish oil supplementation arm in the randomized trial (an absolute increase of between 0.05 mm and 0.11 mm in the study by Angerer et al.) ^{79, 172}. The cross-sectional studies both found that people with greater dietary intake of omega-3 fatty acids, either as total linolenic acid or as fish, had significantly thinner IMTs than those with less intake.

Sub-populations and Covariates

Other than study design, the primary difference between the studies that found no effect and the studies that found a beneficial effect of omega-3 fatty acids is that the former were both trials in patients with cardiovascular disease and the latter were both studies of generally healthy individuals. There is insufficient data, however, to conclude that the differences were due to study populations. There is no evidence among people with diabetes or hyperlipidemia. Bemelmans et al. performed a regression analysis of predictors of change in IMT among subjects taking ALA margarine ¹⁷². Age, sex, blood pressure, LDL, and weight were not predictive of change in IMT. In addition, change in intake of polyunsaturated fatty acids, cholesterol and alcohol were not predictive of change in IMT. Change in intake of saturated fatty acids (SFA) was positively associated, and change in intake of fruit was negatively associated, with change in IMT in univariate analysis but not in multivariate analysis (although it is not clear what factors were included in multivariate analysis since none was significant).

In the cross-sectional study, IMT was greater in older than younger subjects in both the fishing and farming villages. Among younger villagers, IMT was non-significantly lower in the fishing village than the farming village; however, in subjects in their seventh and eighth decades IMT was marginally greater in the fishing village.

Dose and Source Effect, Exposure Duration, Sustainment of Effect

There are insufficient data to draw conclusions regarding dose effect, oil type, duration of intervention or exposure, or sustainment of effect after stopping omega-3 fatty acids.

Overall Effect ^{64, 174–178}

The 3 randomized trials each found a small relative improvement in exercise capacity in subjects with coronary artery disease who took fish oil supplements compared to those who took olive oil supplements. However, with a single exception, exercise capacity measurements improved in all study arms, regardless of whether subjects consumed fish oil or olive oil supplements. The maximum double product (heart rate multiplied by blood pressure) fell by a non-significant amount in the olive oil arm in Salachas et al. ¹⁷⁴.

Warren et al. evaluated 7 patients with stable angina who took cod liver oil supplements for 6 weeks ¹⁷⁸. Exercise workload and time to ischemia improved, although the changes were not significant. The ratio of resting to exercise workload fell significantly. Verheugt et al. studied 5 men with moderate to severe exercise-induced angina ¹⁷⁷. They were given fish oil for 6 months. The patients' angina was sufficiently severe that all ETTs both before and after treatment were discontinued because of angina symptoms. Essentially no change was found in either exercise duration or maximal ST depression. Toth et al. enrolled 10 men with coronary artery disease and hyperlipidemia ¹⁷⁶. They fish oil supplements for 2 months. A variety of measures of cardiac function significantly improved.

Overall, given the small number of studies and subjects, the different metrics used across studies, and the lack of placebo control in half the studies, only limited conclusions can be drawn about the effect of omega-3 fatty acids in improving cardiac function in patients with coronary artery disease. The studies suggest that fish oil consumption may benefit exercise capacity among patients with coronary artery disease, although the effect may be small.

Sub-populations, Dose Effect, Duration, Sustainment of Effect

There is no evidence regarding different doses, duration of fish oil consumption, other omega-3 fatty acids, the effect in various sub-populations, or sustainment of effect.

Overall Effect ^181–183

One randomized controlled trial was performed in 60 healthy volunteers who took either low or high dose fish oil supplements, or olive oil capsules for 12 weeks ¹⁸³. No significant effect was found either within study arms or compared to olive oil. The authors concluded that among all subjects, fish oil supplementation had no effect on heart rate variability.

In a randomized trial of 49 patients who had had a recent myocardial infarction and had a ventricular ejection fraction below 0.40 those who consumed fish oil supplements (for 12 weeks) had a significant increase in SDNN compared to controls ¹⁸¹. The authors concluded that omega-3 fatty acids may increase heart rate variability in survivors of myocardial infarction which may be protective against ventricular arrhythmias and mortality.

The same patients with recent myocardial infarction were divided at baseline into 3 groups based on their regular level of fish consumption ¹⁸². Both groups who consumed at least 1 fish meal per week had greater SDNN than those who did not consume fish, though the difference was not statistically significant. This finding may suggest that dietary fish consumption increases SDNN and thus is protective against ventricular arrhythmia.

Sub-populations and Covariates

Neither study directly compared healthy subjects with those with CVD. Neither examined subjects with either diabetes or dyslipidemia. While the effect of fish oil supplementation appeared greater in the study of subjects with recent myocardial infarction, there is insufficient evidence to compare the effect in subjects with or without heart disease.

In the study of healthy subjects, sub-group analyses based on sex and baseline SDNN suggested that the effect of fish oil supplementation was greatest in the 18 men with below median (<150 msec) baseline SDNN. However, data were not reported for the other 3 subgroups (women and those with above median SDNN).

Dose and Source Effect and Exposure Duration

The study among healthy subjects compared low and high dose fish oil supplementation. While it appears that there may be a trend toward increasing SDNN with higher dose fish oil, it is noteworthy that the subjects on high dose fish oil had no change in their SDNN while those on olive oil had a decrease in SDNN. Both trials lasted 12 weeks. There is no evidence regarding the effect of duration of intervention or exposure.

Sustainment of Effect

Neither study re-examined subjects after stopping fish oil supplementation.

Summary (Table 3.26)

Meta-regression revealed direct relationships between dose of consumed EPA+DHA and changes in measured levels of EPA and DHA, either as plasma or serum phospholipids, platelet phospholipids, or erythrocyte membranes. The correlation between dose and change in level appears to be fairly uniform, where 1 g supplementation of EPA and/or DHA is associated with, approximately, a 1% increase in EPA+DHA level. Granulocyte and monocyte membrane phospholipid levels also increased by roughly similar amounts after omega-3 fatty acid supplementation in individual studies. In these studies, ALA level did not change significantly after supplementation in any blood marker. In most studies, there was a decrease in arachidonic acid (AA, 20:4 n-6) level, which corresponded to the increase in EPA+DHA level.

Among eligible studies, only 3 included ALA supplementation arms ^{53, 143, 160}. The dose of ALA in these 3 studies ranged from 4.5 to 9.5 g/d. The studies consistently found an increase in both ALA and EPA levels in the blood markers, at these doses of ALA. In contrast, there was no significant change in DHA level when lower dose of ALA was used (up to 6.8 g/d) but in the study arm that received 9.5 g/d ALA a significant increase in DHA level was also found.

Plasma or Serum Phospholipid Composition ^{48, 53, 62, 66, 74, 90, 97, 100, 101, 120, 129, 131, 132, 146, 157, 184} (Table 3.27, Figure 3.4)

EPA/DHA. For plasma and serum phospholipid composition, 16 randomized trials with 30 omega-3 fatty acid arms were initially included; however, we excluded 1 study that reported only total omega-3 fatty acid dose and levels ¹³¹. Among the 15 trials of EPA and/or DHA supplementation (which had 28 treatment arms), the dose of EPA+DHA ranged from 0.2 to 5.8 g/day. Study populations include general healthy population, and people with diabetes, dyslipidemia or cardiovascular diseases. Meta-regression shows a significant dose-response relationship between the dietary EPA and DHA supplementations and the changes in EPA+DHA compositions in plasma or serum phospholipids across studies. Across studies, the effect was similar regardless of source of EPA or DHA. Three studies compared purified EPA to purified DHA ^{66, 120, 132}. All found that purified EPA increased EPA and decreased DHA in plasma phospholipid and that purified DHA increased DHA by about 4 to 7 times as much as EPA in plasma phospholipid; however, combined EPA+DHA was increased by about the same amount by both fatty acids.

Meta-regression equation (r² = 0.45, P < .001): Change in Plasma/Serum EPA+DHA Level (%) = 0.93 × [EPA+DHA Intake (g/day)] + 1.41

Because 4 studies reported only EPA levels, we re-analyzed the data with only the 12 studies with a complete EPA and DHA profile of plasma/serum phospholipids. As expected, since no study excluded DHA levels, the revised meta-regression equation indicates that the EPA+DHA level increases by a greater amount for each unit of omega-3 fatty acid supplementation and the r² was greater than in the meta-regression that included all studies.

Meta-regression equation (r² = 0.63, P < .001): Change in Plasma/Serum EPA+DHA Level (%) = 1.24 × [EPA+DHA Intake (g/day)] + 0.89

ALA. One study also evaluated 2 linseed/rapeseed oil supplementation doses, which included primarily ALA with minimal EPA and DHA ⁵³. Finnegan et al. found that with higher dose ALA (9.5 g/d), EPA, DHA and ALA levels all significantly increased. With lower dose ALA (4.5 g/d), EPA and ALA levels rose by a degree consistent with the lower dose of omega-fatty acids; although DHA levels did not change. In the remaining study arms of fish oils and sunflower oils, small amounts of ALA (<= 1.5 g/d) did not affect ALA levels. In this study, a daily dose of 9.5 g or 4.5 g ALA (with 0.3 g EPA+DHA) had similar effects on plasma EPA levels as a daily dose of 1.7 g or 0.8 g EPA+DHA (with 1.4 g ALA), respectively. The plasma level of AA did not decrease in either ALA arm.

Platelet Phospholipid Composition ^{68, 71, 95, 96, 101, 116, 122, 123, 132, 137, 143, 163} (Table 3.28, Figure 3.5)

EPA/DHA. For platelet phospholipid composition, we analyzed 12 randomized trials with 21 omega-3 fatty acid arms. All of these studies evaluated EPA and/or DHA supplementation. One treatment arm was ALA; therefore, there were 20 EPA and/or DHA treatment arms. The dose of EPA+DHA ranged from 0.8 to 5.9 g/day. Study populations include general healthy population and people with diabetes, dyslipidemia, or cardiovascular diseases. Meta-regression results show a significant dose-response relationship between the dietary EPA and DHA supplementations and the changes in EPA+DHA compositions in platelet phospholipids across studies. Studies that used fish or fish combined with fish oil supplement treatments generally had greater increases in platelet phospholipid EPA+DHA amounts than studies of fish oil supplements. This effect was seen in Mori, et al. (1994), which compared fish, fish oil supplements, and combination fish and fish oil ⁷¹. They reported that the largest increase in DHA occurred in the groups consuming fish. In contrast to the finding in plasma phospholipids, Mori et al. (2000) reported that platelet EPA+DHA levels rose more in subjects taking DHA than in subjects taking EPA, although it is not reported whether this difference is statistically significant ¹³².

Meta-regression equation (r² = 0.52, P < .001): Change in Platelet EPA+DHA Level (%) = 0.74 × [EPA+DHA Intake (g/day)] + 1.16

As was the case for plasma/serum phospholipid levels, the re-analysis of the platelet phospholipid data that excluded the 2 studies without a complete EPA and DHA profile indicates a larger increase in EPA+DHA level and a larger r² than in the complete meta-regression.

Meta-regression equation (r² = 0.72, P < .001): Change in Platelet EPA+DHA Level (%) = 0.80 × [EPA+DHA Intake (g/day)] + 1.25

ALA. One study also evaluated linseed oil supplementation, which included only ALA without EPA or DHA ¹⁴³. Freese et al. found that a 5.9 g/d ALA supplementation significantly increased EPA and ALA platelet phospholipid levels. However, the effect on EPA levels was small in comparison to the effect of a similar dose of fish oil (+0.41% vs. +3.32% for 5.2 g/d EPA+DHA). In addition, DHA levels were unaffected. The AA level decreased in the ALA arm.

Erythrocyte Membrane Phospholipid Composition^{79, 88, 95, 96, 101, 115, 134, 141, 160, 175} (Table 3.29, Figure 3.6)

EPA/DHA. For erythrocyte membrane phospholipid composition, 10 randomized trials with 15 omega-3 fatty acid arms were included. All of these studies evaluated EPA and/or DHA supplementation. One study included 2 ALA treatment arms; therefore, there were 13 EPA and/or DHA treatment arms. The dose of EPA+DHA ranged from 0.8 to 4.6 g/day. Study populations include general healthy population and people with diabetes, dyslipidemia or cardiovascular diseases. Meta-regression results show no significant dose-response relationship between the dietary EPA and DHA supplementations and the changes in EPA plus DHA compositions in platelet phospholipids. No clear difference is seen in effect based on source of omega-3 fatty acids. No study compared different sources of EPA+DHA oil.

Meta-regression equation (r² = 0.11, P = .14): Change in Erythrocyte EPA+DHA Level (%) = 0.63 × [EPA+DHA Intake (g/day)] + 3.22

The re-analysis of the data, excluding 1 study by Green et al. who did not report the change in DHA levels, greatly affected slope and statistical significance of the meta-regression equation ¹⁰¹. The large effect of this single study can be explained by outlier status of the study. The change in EPA level reported in this study is considerably lower than the change in EPA+DHA levels in studies with similar supplementation doses.

Meta-regression equation (r² = 0.39, P < .02): Change in Erythrocyte EPA+DHA Level (%) = 1.05 × [EPA+DHA Intake (g/day)] + 2.69

ALA. One study also evaluated a diet enriched in ALA and that contained no EPA or DHA among both young (16–33 years old) and old (60–78 years old) subjects ¹⁶⁰. Wensing et al. found that a 6.8 g/d ALA supplementation significantly increased both EPA and ALA levels but not DHA level. The effects on the changes in EPA and ALA compositions were larger among older subjects than among younger subjects. The higher dose ALA (6.8 g/d) had a smaller effect on EPA levels (+0.20% and +0.40%, for younger and older subjects, respectively) than a lower dose of EPA+DHA (1.6 g/d, +1.30%). The AA level decreased among old subjects while it increased among young subjects.

Granulocyte Membrane Phospholipid Composition ¹³⁷ (Table 3.30)

One randomized controlled trial examined the changes of EPA+DHA composition in granulocyte membrane phospholipids after fish oil supplementation. Madsen et al. found that EPA and DHA compositions in granulocyte phospholipids significantly increased after 12 weeks of fish oil supplement treatment, while no significant changes were found in the placebo group ¹³⁷. In addition, the change in DHA profile was significantly larger in the higher-dose fish oil supplementation group than in the lower-dose fish oil group.

Monocyte Membrane Phospholipid Composition ¹⁴⁶ (Table 3.31)

One crossover study examined the changes of EPA+DHA composition in monocyte phospholipids after cod-liver oil supplementation. Hansen, et al. showed the EPA profile in monocyte phospholipids significantly increased, while the arachidonic acid profile significantly decreased after 8 weeks of cod liver oil supplement treatment compared to the no treatment controls ¹⁴⁶.

Effect on Triglycerides and Other Serum Lipids

The strongest, most consistent effect of omega-3 fatty acids was among the 19 studies of Tg. Most of these studies reported a net decrease in Tg of about 10% to 33%. The effect was dose-dependent and generally consistent among healthy subjects and patients with CVD, dyslipidemia, or at elevated risk of CVD. The effect was also greater in studies with higher mean baseline Tg. However, 1 of 2 studies of plant oils (ALA) found a net increase in Tg. Limited data suggest that the effect is not related to sex, age, weight, background diet, or lipid treatment. The effect of duration of intervention is unclear and there were no data regarding sustainment of effect. In addition, no study of diabetic patients had sufficient number of subjects to be analyzed.

The effect of omega-3 fatty acids on other serum lipids was weaker. The 23 studies of total cholesterol and the 19 studies of HDL we analyzed were heterogeneous, but mostly found small (0% to 6%), non-significant net increases in levels of both lipids. The 15 analyzed trials of LDL were fairly uniform in finding small net increases in LDL. The effect of plant oils (ALA) on these lipoproteins was possibly weaker but similar to the effect of marine oils. No differences in effect were seen among different populations, including the diabetic subjects who were evaluated in a sub-analysis. One study found a larger net increase in total cholesterol among subjects on a higher fat diet compared to those on a lower fat diet, but this effect was not seen for other lipids. A single study of fish oil reported a steady increase in HDL levels over time beginning at 6 weeks and ending at 12 months. No other studies found an effect of time on lipids and no other covariates were reported to interact with fish oil effects on lipids.

One study compared the effect of purified EPA to purified DHA on these 4 lipids. The results were mixed. EPA lowered total cholesterol significantly (and substantially) more than DHA, DHA increased HDL by a small but significant amount more than EPA, and the effects of the 2 oils were similar in their lack of effect on LDL and their ability to lower Tg.

Effect on Blood Pressure

A recent meta-regression of the effect of fish oils on blood pressure found a small but significant reduction in both systolic and diastolic blood pressure of about 2 mm Hg. The effect was stronger in older and hypertensive populations. Because the meta-regression excluded diabetic populations, we evaluated the 6 randomized studies of diabetics and found similar results. One study reported that neither sex nor Hgb A_1c levels were related to the fish oil effect on blood pressure. No study analyzed plant oils. One study reported no significant difference in blood pressure effect of purified EPA compared to purified DHA.

Effect on Restenosis after Coronary Angioplasty

We performed a meta-analysis of the 12 randomized trials that reported restenosis rates after coronary angioplasty. All evaluated fish oils. We found heterogeneity of results across studies but an overall trend toward a net reduction of relative risk of 14% with fish oil intake. Two studies reported no significant difference in effect between men and women.

Effect on Exercise Capacity and Heart Rate Variability

The 6 available studies examining exercise tolerance testing suggest that fish oil consumption may benefit exercise capacity among patients with coronary artery disease, although the effect may be small. Three analyses of heart rate variability in 2 study populations concluded that fish oil supplementation among patients with recent myocardial infarction, and dietary fish consumption in healthy people, improves heart rate variability, which may, in turn, reduce the incidence of ventricular arrhythmias. However, fish oil supplementation did not improve heart rate variability in the same healthy population.

Effect on Other Cardiovascular Risk Factors and Intermediate Markers

The effects of omega-3 fatty acids on the other outcomes that we evaluated were either small or inconsistent across studies.

Apolipoproteins. No consistent effect was found across 14 studies of Lp(a), although one study reported a small but significant net decrease in subjects with elevated baseline Lp(a) levels compared to those with lower baseline levels. There were insufficient studies to compare different omega-3 fatty acids. The 27 studies of apo A-I that we analyzed generally found no effect or either a small increase or decrease in level with omega-3 fatty acid consumption. Limited evidence suggested that purified EPA may decrease apo A-I levels while DHA has no effect, and that there is no difference in effect between fish oils and ALA. There was little consistency of effect in the 25 studies of total apo B. The 4 available studies of apo B-100 found a range of effects from a 5% decrease to a 15% increase in level. Most of the 6 studies of LDL apo B found large, significant net increases in LDL apo B with omega-3 fatty acid consumption.

C-reactive protein. The 5 available studies of CRP found no effect with fish oil supplementation or dietary fish.

Measures of hemostasis. No consistent effect was found among the 24 analyzed studies of fibrinogen, the 19 analyzed studies of factor VII, or the 5 available randomized trials of factor VIII. The 9 randomized trials of vWF mostly found a small, non-significant decrease in level with omega-3 fatty acid consumption. The results among the 11 analyzed studies of platelet aggregation were heterogeneous depending on aggregating agent, dose of agent, and measurement metric used, however, generally no effect was found with omega-3 fatty acid intake. The few studies that compared types of omega-3 fatty acids found no difference in effect on these measures of hemostasis, with the exception that 2 studies came to opposite conclusions regarding whether fish oil prolonged platelet aggregation by a greater degree than ALA, and 1 study concluded that DHA may be less potent at prolonging platelet aggregation than EPA.

Carotid intima-media thickness. The 4 available studies of carotid IMT were heterogeneous. The randomized trial found no effect of fish oil but 2 cross-sectional studies found that dietary omega-3 fatty acid was correlated with thinner IMT; the cohort study of plant oil margarine was inconclusive.

Glucose tolerance. Overall, the studies of markers of glucose tolerance found no consistent effect of omega-3 fatty acids. There was a wide range of net effects of omega-3 fatty acids on fasting blood sugar across the 17 analyzed studies. Heterogeneity was present regardless of the make-up of the study population, although the range of effect was widest among diabetic patients. Within studies there were no apparent differences in effect of different omega-3 fatty acids on fasting blood sugar. Among the 18 analyzed studies of Hgb A_1c there was no substantial significant effect of omega-3 fatty acid consumption, regardless of study population. A single study found no difference in effect of purified EPA and purified DHA on Hgb A_1c. The 15 randomized trials of fasting insulin levels were very heterogeneous. Similar heterogeneity existed among the 9 studies of generally euglycemic populations as among the studies of diabetics and obese subjects. Within studies there were no apparent differences in effect of different omega-3 fatty acids on fasting insulin levels.

Tissue Levels of Fatty Acids

Meta-regression of 30 studies revealed direct relationships between dose of omega-3 fatty acids consumed and changes in measured levels of eicosapentaenoic acid (EPA, 20:5 n-3) and docosahexaenoic acid (DHA, 22:6 n-3), either as plasma or serum phospholipids, platelet phospholipids, or erythrocyte membranes. The correlation between dose and change in level appears to be fairly uniform, where 1 g supplementation of EPA and/or DHA corresponds to approximately a 1% increase in EPA+DHA level. Granulocyte and monocyte membrane phospholipid levels also increased after omega-3 fatty acid supplementation in individual studies.