Chapter  113:  Effects of Omega-3 Fatty Acids on Cancer

The Recognition of Essential Fatty Acids

Dietary fat has long been recognized as an important source of energy for mammals, but in the late 1920s, researchers demonstrated the dietary requirement for particular fatty acids, which came to be called essential fatty acids. It was not until the advent of intravenous feeding, however, that the importance of essential fatty acids was widely accepted: Clinical signs of essential fatty acid deficiency are generally observed only in patients on total parenteral nutrition who received mixtures devoid of essential fatty acids or in those with malabsorption syndromes. These signs include dermatitis and changes in visual and neurological function. Over the past 40 years, an increasing number of physiological functions, such as immunomodulation, have been attributed to the essential fatty acids and their metabolites, and this area of research remains quite active.^{1, 2}

Fatty Acid Nomenclature

The fat found in foods consists largely of a heterogeneous mixture of triacylglycerols (triglycerides)--glycerol molecules that are each combined with three fatty acids. The fatty acids can be divided into two categories, based on chemical properties: saturated fatty acids, which are usually solid at room temperature, and unsaturated fatty acids, which are liquid at room temperature. The term “saturation” refers to a chemical structure in which each carbon atom in the fatty acyl chain is bound to (saturated with) four other atoms, these carbons are linked by single bonds, and no other atoms or molecules can attach; unsaturated fatty acids contain at least one pair of carbon atoms linked by a double bond, which allows the attachment of additional atoms to those carbons (resulting in saturation). Despite their differences in structure, all fats contain approximately the same amount of energy (37 kilojoules/gram, or 9 kilocalories/gram).

The class of unsaturated fatty acids can be further divided into monounsaturated and polyunsaturated fatty acids. Monounsaturated fatty acids (the primary constituents of olive and canola oils) contain only one double bond. Polyunsaturated fatty acids (PUFAs) (the primary constituents of corn, sunflower, flax seed, and many other vegetable oils) contain more than one double bond. Fatty acids are often referred to using the number of carbon atoms in the acyl chain, followed by a colon, followed by the number of double bonds in the chain (e.g., 18:1 refers to the 18-carbon monounsaturated fatty acid, oleic acid; 18:3 refers to any 18-carbon PUFA with three double bonds).

Table 1.1 Nomenclature of omega-3 fatty acids

Table 1.1 Nomenclature of omega-3 fatty acids

Names		Abbreviations
Trivial	IUPAC*	Carboxyl-reference	Omega-reference	Other
Linolenic acid	9,12,15-octadecenoic acid	18:3Δ9 12 15	18:3n-3	ALA
	alpha-linolenic acid		18:3 (ω-3)	α-LA
				LNA
				α-LNA
Docosahexaenoic acid	4,8,12,15,19- docosahexaenoic acid	22:6Δ4 8 12 15 19	22:6n-3	DHA
	cervonic acid		22:6 (ω-3)
Docosapentaenoic acid	7,10,13,16,19- docosapentaenoic acid	22:5Δ7 10 13 16 19	22:5n-3	DPA
			22:5 (ω-3)
Eicosapentaenoic acid Icosapentaenoic acid	5,8,11,14,17- eicosapentaenoic acid	20:5Δ5 8 11 14 17	20:5n-3	EPA
Timnodonic acid			20:5 (ω-3)

*

IUPAC=International Union of Pure and Applied Chemistry.

Finally, PUFAs can be categorized according to their chain length. The shorter-chain 18-carbon n-3 and n-6 PUFAs are precursors to the longer 20- and 22-carbon PUFAs, called very-long-chain PUFAs (VLCPUFAs).

Fatty Acid Metabolism

Mammalian cells can introduce double bonds into all positions on the fatty acid chain except the n-3 and n-6 position. Thus, the shorter-chain alpha-linolenic acid (ALA, chemical abbreviation: 18:3n-3) and linoleic acid (LA, chemical abbreviation: 18:2n-6) are essential fatty acids. No other fatty acids found in food are considered ‘essential’ for humans, because they can all be synthesized from the shorter chain fatty acids.

Figure 1.1 Classical omega-3 and omega-6 fatty acid (more...)

   Figure 1.1 Classical omega-3 and omega-6 fatty acid synthesis pathways and the role of omega-3 fatty acids in regulating health/disease markers

Figure 1.1

The omega-6 fatty acid LA is converted to gamma-linolenic acid (GLA, 18:3n-6), an omega-6 fatty acid that is a positional isomer of ALA. GLA, in turn, can be converted to the longer-chain omega-6 fatty acid, arachidonic acid (AA, 20:4n-6). AA is the precursor for certain classes of an important family of hormone-like substances called the eicosanoids (see below).

The omega-3 fatty acid ALA (18:3n-3) can be converted to the long-chain omega-3 fatty acid, eicosapentaenoic acid (EPA; 20:5n-3). EPA can be elongated to docosapentaenoic acid (DPA 22:5n-3), which is further elongated, desaturated, and beta-oxidized to produce docosahexaenoic acid (DHA; 22:6n-3). EPA and DHA are also precursors of several classes of eicosanoids and docosanoids, respectively, are known to play several other critical roles, some of which are discussed further below.

The conversion from parent fatty acids into the VLC PUFAs-EPA, DHA, and AA-appears to occur slowly in humans. In addition, the regulation of conversion is not well understood, although it is known that ALA and LA compete for entry into the metabolic pathways.

Physiological Functions of EPA and AA

As stated earlier, fatty acids play a variety of physiological roles. The specific biological functions of a fatty acid are determined by the number and position of double bonds and the length of the acyl chain.

Both EPA (20:5n-3) and AA (20:4n-6) are precursors for the formation of a family of hormone-like agents called eicosanoids. Eicosanoids are rudimentary hormones or regulatory-molecules that appear to occur in most forms of life. However, unlike endocrine hormones, which travel in the blood stream to exert their effects at distant sites, the eicosanoids are autocrine or paracrine factors, which exert their effects locally - in the cells that synthesize them or adjacent cells. Processes affected include the movement of calcium and other substances into and out of cells, relaxation 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.¹

Figure 1.1

EPA (20:5 n-3) also affects lipoprotein metabolism and decreases the production of substances - including cytokines, interleukin 1β (IL-1β), and tumor necrosis factor α (TNF-α) - that have pro-inflammatory effects (such as stimulation of collagenase synthesis and the expression of adhesion molecules necessary for leukocyte extravasation [movement from the circulatory system into tissues]).¹ DPA (22:5n-3), the elongation product of EPA, is metabolized to DHA (22:6n-3). DHA (22:6n-3) is the precursor to a newly-described metabolite called 10,17S-docosatriene,⁴ which is part of a family of compounds called ‘resolvins.’⁵ They are synthesized in the brain in response to an ischemic insult and counteract the pro-inflammatory actions of infiltrating leukocytes by blocking interleukin 1-beta-induced NF-kappaB activation and cyclooxygenase-2 expression.⁶ DHA also plays a role in retinal rod outer segments by influencing membrane fluidity so as to optimize G protein coupled signaling.⁷ The mechanism responsible for the suppression of cytokine production by omega-3 LC PUFAs and VLCPUFAs remains unkown, although suppression of omega-6-derived eicosanoid production by omega-3 fatty acids may be involved, because the omega-3 and omega-6 fatty acids compete for common enzymes in the fatty acid metabolic pathway, including delta-6 desaturase, as well as the rate-limiting enzymes in the eicosanoid pathway - phospholipases A2, cyclooxygenase, and lipoxygenase.

Along with AA, DHA is the major PUFA found in the brain and is thought to be important for brain development and function. Recent research has focused on this role and the effect of supplementing infant formula with DHA (since DHA is naturally present in human breast milk but not in formula).

Dietary Sources and Requirements

Both ALA and LA are present in a variety of foods. LA is present in high concentrations in many commonly used oils, including safflower, sunflower, soy, and corn oil. ALA is present in some commonly used oils, including canola and soybean oil, and in some leafy green vegetables.

Table 1.2 Sources and proportions of omega-3 fatty (more...)

Table 1.2 Sources and proportions of omega-3 fatty acids in common foods and supplements

Food/supplement	EPA	DHA	DPA	ALA
	20:5n-3	22:6n-3	22:5n-3	18:3n-3
Foods/supplements in which total omega-3 fatty acids account for more than 50% of total PUFA
Fish
Anchovy
Halibut
Herring
Mackerel
Salmon
Sardine
Tuna
Canned, waterpacked
Fresh Bluefin
Oils/Supplements
Cod liver oils
Coromega*
Fish oil capsules*
Flaxseed/linseed oil*
Herring oil
MaxEPA*
Menhaden oil
Neuromins*
Omacor*
Ropufa*
Salmon oil
Sardine oil
Seeds and other foods
Flaxseeds/Linseeds
Spinach, cooked
Foods/supplements in which total omega-3 fatty acids are 10–50% of total PUFA
Oils
Black currant oil
Canola oil†
Mustard seed oils
Soybean oil
Walnut oil
Wheat germ oil
Other foods
Wheat germ
Human milk‡
Foods/supplements in which total omega-3 fatty acids are less than 10% of total PUFA
Efamol Marine*
Peanut butter
Soybeans
Olive oil
Walnuts

*

Dietary Supplement;

† Also called rapeseed oil;

‡ The amounts of ALA, EPA, and DHA in human milk vary greatly as a function of maternal diet; the amount of DHA rarely seems to exceed 25 percent of the total n-3 PUFA content (ALA is present in the greatest amount), but that content as well as the proportion of DHA is assumed to meet the requirements of the infant.

Table 1.3 Good food sources* of omega-3 fatty acids (more...)

Table 1.3 Good food sources* of omega-3 fatty acids

	EPA+DHA	ALA
Fish (3oz. Cooked)
Anchovy
Halibut
Herring, Atlantic
Pacific
Mackerel, Atlantic
Pacific
Salmon, Atlantic†
Sardines
Trout, Rainbow
Tuna, Albacore
Canned light, water-packed
Canned white, water-packed
Fresh Bluefin
Organ Meats (3 oz. Cooked)
Brain, lamb
Brain, pork
Thymus, calf
Other Foods
Caviar (1 oz.) ‡
Human breast milk (1c) ‡	§
Soybeans, cooked (1/2c)
Spinach, cooked (1/2c)
Tofu, regular (1/2c)
Walnuts (1/4c)
Wheat germ (1/4c) ‡
Oils (1 Tbs.)
Canola
Cod liver
Flaxseed/linseed
Herring
Menhaden
Salmon
Sardine
Soybean
Walnut
Wheat germ
Seeds
Flaxseeds/linseeds (1 Tbs.)

Source: Figures adapted from USDA, 2003;

*

Foods that provide (per serving) 10 percent or more of the Adequate Intake (AI) for ALA or the Acceptable Macronutrient Distribution Range (AMDR) for EPA and DHA (10 percent of the AMDR for ALA); an AI is a recommended average daily intake level based on observed or experimentally determined estimates of nutrient intake by a group of apparently healthy people (thus, assumed to be adequate) when an RDA cannot be determined; an AMDR is defined as “a range of intakes for a particular energy source that is associated with reduced risk of chronic disease while providing adequate intake of essential nutrients.”⁹;

† Farm-raised Atlantic salmon have nearly identical omega-3 fatty acid levels to wild Atlantic salmon and significantly more omega-3 fatty acids than wild Pacific salmon;

‡ Standard serving size not established;

§ See table note for Table 1.2.

Table 1.4 Estimates of the mean intake of LA, ALA, (more...)

Table 1.4 Estimates of the mean intake of LA, ALA, EPA, and DHA in the U.S. population from analysis of NHANES III data.*

	Grams/day		Percent energy intake/day
	Mean ± SEM	Median (range)†	Mean ± SEM	Median (range)†
LA (18:2n-6)	14.1 ± 0.2	9.9 (0 – 168)	5.79 ± 0.05	5.30 (0 – 39.4)
ALA (18:3n-3)	1.33 ± 0.02	0.90 (0 – 17)	0.55 ± 0.004	0.48 (0 – 4.98)
EPA (20:5n-3)	0.04 ± 0.003	0.00 (0 – 4.1)	0.02 ± 0.001	0.00 (0 – 0.61)
DHA (22:6n-3)	0.07 ± 0.004	0.00 (0 – 7.8)	0.03 ± 0.002	0.00 (0 – 2.86)

*

Based on analysis of a single 24-hour dietary recall from NHANES III data;

† Distributions are not adjusted for the over-sampling of Mexican -Americans, non-Hispanic African Americans, children five years old and under, and adults 60 years and over in the NHANES III dataset.

Table 1.5 Mean, range, and median usual daily Intakes (more...)

Table 1.5 Mean, range, and median usual daily Intakes (ranges) of n-6 and n-3 PUFAs, in the U.S. population, from analysis of CSFII data (1994 to 1998).*

	Mean (gms/d) (± SEM)†	Range of Means (gms/d) (±SEM)	Median (gms/d) (± SEM)†
LA (18:2n-6)	13.0 ± 0.1	6.7 ± 0.1–17.6 ± 0.5	12.0 ± 0.1
Total n-3 FA	1.40 ± 0.01	0.72 ± 0.02 – 1.86 ± 0.04	1.30 ± 0.01
ALA (18:3n-3)	1.30 ± 0.01	0.72 ± 0.02 – 1.73 ± 0.04	1.21 ± 0.01
EPA (20:5n-3)	0.028	0.002 – 0.049	0.004
DPA (22:5n-3)	0.013	0.001 – 0.019	0.005
DHA (22:6n-3)	0.057 ± 0.018	< 0.0005 ± 0.001	0.046 ± 0.013

Source: Adapted from Dietary Reference Intakes Report;⁹

*

Estimates are based on respondents' intakes on the first day of survey and were adjusted using the Iowa State University method;

† For all individuals.

Lacking sufficient evidence from research on the effects or correction of dietary deficiencies to establish Recommended Dietary Allowances (RDAs) for the essential fatty acids, the Food and Nutrition Board (FNB) of the Institute of Medicine⁹ has set adequate intakesⁱⁱ (AI) for the essential fatty acids, based on the average intakes of healthy CSFII participants. The AIs for the essential fatty acids vary by age group and sex, as well as for particular conditions such as pregnancy and breastfeeding. For ALA, the AI for men 19 and older, is 1.6 grams/day and the AI for (non-pregnant, non-breastfeeding) women is 1.1 grams/day. The AI for LA is 17 grams/day for men and 11 grams/day for women.

Table 1.6 The omega-3 fatty acid content, in grams (more...)

Table 1.6 The omega-3 fatty acid content, in grams per 100 g food serving, of a representative sample of commonly consumed fish, shellfish, fish oils, nuts and seeds, and plant oils.*

Food item	EPA	DHA	ALA
Fish (Cooked in dry heat unless otherwise specified)
Anchovy, European	0.8	1.3	-
Bass, Freshwater, Mixed Sp.	0.3	0.5	0.1
Bass, Striped	0.2	0.8	trace
Bluefish	0.3	0.7	-
Carp	0.3	0.3	0.3
Catfish, Channel, farmed	trace	0.1	0.1
Cod, Atlantic	trace	0.2	trace
Cod, Pacific	0.1	0.2	trace
Eel, Mixed Sp.	0.1	0.1	0.6
Flounder & Sole Sp.	0.2	0.3	trace
Grouper, Mixed Sp.	trace	0.2	-
Haddock	0.1	0.2	trace
Halibut, Atlantic and Pacific	0.1	0.4	0.1
Halibut, Greenland	0.7	0.5	0.1
Herring, Atlantic	0.9	1.1	0.1
Herring, Pacific	1.2	0.9	0.1
Mackerel, Atlantic	0.5	0.7	0.1
Mackerel, Pacific and Jack	0.7	1.2	0.1
Mullet, Striped	0.2	0.1	trace
Ocean Perch, Atlantic	0.1	0.3	0.1
Pike, Northern	trace	0.1	trace
Pike, Walleye	0.1	0.3	trace
Pollock, Atlantic	0.1	0.5	-
Pompano, Florida	0.2	0.5	-
Roughy, Orange	trace	-	trace
Salmon, Atlantic, Farmed	0.7	1.5	0.1
Salmon, Atlantic, Wild	0.4	1.4	0.4
Salmon, Chinook	1.0	0.7	0.1
Salmon, Chinook, Smoked (lox)	0.2	0.3	-
Salmon, Chum	0.3	0.5	trace
Salmon, Coho, Farmed	0.4	0.9	0.1
Salmon, Coho, Wild	0.4	0.7	0.1
Salmon, Pink	0.4	0.6	trace
Salmon, Pink, Canned	0.8	0.8	0.1
Salmon, Sockeye	0.5	0.7	0.1
Sardine, Atlantic, Canned in Oil	0.5	0.5	0.5
Sea bass, Mixed Sp.	0.2	0.6	-
Sea trout, Mixed Sp.	0.2	0.3	trace
Shark, Mixed Sp., battered and fried	0.3	0.4	0.2
Snapper, Mixed Sp.	0.1	0.3	0.1
Swordfish	0.1	0.7	0.2
Trout, Mixed Sp.	0.3	0.7	0.2
Trout, Rainbow, Farmed	0.3	0.8	0.1
Trout, Rainbow, Wild	0.5	0.5	0.2
Tuna, Fresh, Bluefin	0.4	1.1	-
Tuna, Fresh, Skipjack	trace	0.2	-
Tuna, Fresh, Yellowfin	trace	0.2	trace
Tuna, Light, Canned in Oil	trace	0.1	trace
Tuna, Light, Canned in Water	trace	0.2	trace
Tuna, White, Canned in Oil	trace	0.2	0.2
Tuna, White, Canned in Water	0.2	0.6	trace
Whitefish, Mixed Sp.	0.4	1.2	0.2
Whitefish, MixedSp., Smoked	trace	0.2	-
Wolf fish, Atlantic	0.4	0.4	trace
Shellfish (Raw)
Abalone, Mixed Sp., fried	0.1	0.1	0.1
Clam, Mixed Sp., moist heat	0.1	0.1	trace
Crab, Alaska King, moist heat	0.3	0.1	-
Crab, Blue, moist heat	0.2	0.2	-
Crayfish, Mixed Sp., Farmed	0.1	trace	trace
Lobster, Northern, moist heat	0.1	trace	trace
Mussel, Blue	0.3	0.5	trace
Oyster, Eastern, Farmed	0.2	0.2	0.1
Oyster, Eastern, Wild	0.3	0.3	0.1
Oyster, Pacific	0.9	0.5	0.1
Scallop, Mixed Sp.	0.2	0.2	-
Shrimp, Mixed Sp.	0.2	0.1	trace
Squid, Mixed Sp., fried	0.2	0.4	0.1
Fish Oils
Cod Liver Oil	6.9	11	0.9
Herring Oil	6.3	4.2	0.8
Menhaden Oil	13.2	8.6	1.5
Salmon Oil	13	18.2	1.1
Sardine Oil	10.1	10.7	1.3
Nuts and Seeds
Butternuts, Dried	-	-	8.7
Flaxseed			18.1
Walnuts, English	-	-	9.1
Plant Oils
Canola (Rapeseed)	-	-	9.3
Flaxseed Oil	-	-	53.3
Soybean Lecithin Oil	-	-	5.1
Soybean Oil	-	-	6.8
Walnut Oil	-	-	10.4
Wheat germ Oil	-	-	6.9

Source: Figures adapted from USDA, 2003;

*

Sp = species

Rationale for and Organization of this Report

Studies show that tissue levels of AA and EPA-derived eicosanoids influence many physiological processes, including calcium transport across cell membranes, angiogenesis, apoptosis, cell proliferation, and immune cell function. These processes are integral to the immune system and hence the pathogenesis of autoimmune disease such as arthritis, systemic lupus erythematosus, and asthma, as well as cancer. Epidemiological studies have suggested that groups of people who consume diets high in omega-3 FAs may experience a lower prevalence of some types of cancer, and many small trials have attempted to assess the effects of adding omega-3 fatty acids to the diet, either as omega-3 FA-rich foods or as dietary supplements (primarily fish oils). In addition, dietary omega-3 FA have been found to modulate tumor formation and proliferation in rodents.

In response to this evidence, a number of omega-3 FA-containing dietary supplements that claim to protect against a variety of conditions have appeared on the market. Thus, AHRQ and the National Institutes of Health (NIH) Office of Dietary Supplements (ODS) have requested a synthesis of the research to date on the health effects of diets rich in omega-3 FA.

The remainder of this report is organized into four chapters. Chapter Two describes the methods we used to identify and review studies related to the role of omega-3 FA in cancer. Specifically, the effects of omega-3 fatty acids on the incidence of cancer, on clinical outcomes after treatment of cancer, and on tumor growth differentiation and apoptosis. Chapter Three presents our findings related to the effects of omega-3 FA on those topics. Chapter Four presents our conclusions and recommendations for future research in this area.

Objectives

The topic of this report was nominated by the National Institutes of Health (NIH) Office of Dietary Supplements (ODS). The three participating Evidence-Based Practice Centers (EPCs) were asked to examine the effects of omega-3 fatty acids, in general, and on the following conditions: Cardiovascular Disease, Transplantation, Immune-Mediated Diseases, Gastrointestinal/Renal Diseases, Cancer, Neurology, Asthma, Child/Maternal Health, Eye Health, and Mental Health. The Southern California EPC (SCEPC) was responsible for examining Immune-Mediated Diseases and Gastrointestinal/Renal Diseases in Year 1 of the project and Cancer and Neurology in Year 2 of the project. This report pertains to cancer.

Scope of Work

The methodology that we used for this study included the following:

• Refining the preliminary questions provided by AHRQ,

• Convening a technical expert panel to advise the SCEPC on the study,

• Identifying sources of evidence in the scientific literature,

• Establishing inclusion/exclusion criteria for the articles identified in the scientific literature,

• Identifying potential evidence with attention to controlled clinical trials using omega-3 fatty acids,

• Evaluating potential evidence for methodological quality and relevance,

• Extracting data from studies meeting methodological and clinical criteria,

• Synthesizing the results,

• Performing further statistical analysis on selected studies,

• Performing pooled analyses where appropriate,

• Submitting the results to technical experts for peer review,

• Incorporating reviewers' comments into a final report for submission to AHRQ.

Original Proposed Key Questions

Preliminary questions for the project were developed by ODS in collaboration with the following NIH Institutes: (a) National Cancer Institute (NCI); (b) National Eye Institute (NEI); (c) National Heart, Lung, and Blood Institute (NHLBI); (d) National Institute of Alcohol Abuse and Alcoholism (NIAAA); (e) National Institute of Allergy and Infectious Diseases (NIAID); (f) National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS); (g) National Institute of Child Health and Human Development (NICHD); (h) National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK); (i) National Institute of Mental Health; and (j) National Institute of Neurological Disorders and Stroke (NINDS) The general and disease-specific questions that were originally proposed are detailed in Appendix A.1.

Technical Expert Panel

Each AHRQ evidence report is guided by a Technical Expert Panel (TEP). The TEP advises the SCEPC on refining the preliminary questions, determining the proper inclusion/exclusion criteria for the study and the populations of interest, establishing the proper outcomes measures, and conducting the appropriate analyses.

We convened a TEP that focused specifically on cancer. The TEP was composed of distinguished basic scientists and clinicians, with established expertise in omega-3 fatty acids, human nutrition, dietary assessment methods, cancer biology, and oncology. In addition to the experts that we identified, AHRQ and the relevant NIH Institute(s) recommended a number of industry experts. The members of our technical expert panel and a summary of their key comments and recommendations are listed in Appendix A.2.

Key Questions Addressed in this Report

Based on input from our TEP, the preliminary disease-specific questions were revised. The questions that are addressed in this report are as follows:

Tumor Incidence

• What is the evidence that omega-3 fatty acids reduce the incidence of tumors?

If omega-3 fatty acids influence the incidence tumors:

• For what type of tumors?

• Is there an inverse relationship with intake?

• Is there a temporal relationship with intake?

• What is the evidence that genes involved in omega-3 fatty acid transport or metabolism influence the magnitude or direction of the influence on tumor incidence?

• What is the evidence that the response to omega-3 fatty acids is dependent of the intake of antioxidants such as vitamin E or other bioactive food components?

• What is the evidence that the response is modified by the state of the immune system?

Effects on Clinical Outcomes After Cancer Treatment

• What is the evidence that omega-3 fatty acids alter the effects of cancer treatment on malignant tumors and clinical outcomes after cancer treatments?

• What is the evidence that the response to omega-3 fatty acids is dependent of the intake of antioxidants such as vitamin E or other bioactive food components?

• What is the evidence that the response is modified by the state of the immune system?

Tumor Behavior

• What is the evidence that omega-3 fatty acids alter the behavior of malignant tumors in terms of growth, differentiation, and apoptosis?

If omega-3 fatty acids influence the behavior of tumors:

• For what type of tumors?

• Is there an inverse relationship with intake?

• Is there a temporal relationship with intake?

• What is the evidence that genes involved in omega-3 fatty acid transport or metabolism influence the magnitude or direction of the influence on tumor behavior?

Identification of Literature Sources

Potential evidence for our study came from three sources: on-line library databases, the reference lists of all relevant articles, and industry experts.

Evaluation of Evidence

Extraction of Data

Grading Evidence

Data Synthesis

We performed both a qualitative and quantitative synthesis of the evidence. We performed a meta-analysis for those studies that sufficiently assessed interventions, populations, and outcomes to justify pooling. Only randomized controlled trials with a placebo comparator group were considered for meta-analysis. For the remaining studies and for those pertaining to the apoptosis, tumor growth, and differentiation question, we performed a qualitative analysis. For the cohort studies that assessed the effects of omega-3 fatty acids on tumor incidence we constructed summary tables for each type of cancer that detailed the age- and multivariate-adjusted risk ratios that were reported for each study arm. These tables are stratified by the specific categories of omega-3 fatty acids for which the risk ratios were reported, i.e. total omega-3, marine omega-3, ALA, EPA or DHA. Also included in these tables are strata for total fish intake which can reasonably be used as a surrogate for omega-3 consumption given the high omega-3 content of fish. Included in these tables is the median intake of the relevant omega-3 fatty acid for each study arm if it was reported. The categories of omega-3 fatty acids that we report are those that were reported in the included studies and were not identical across the different studies. These studies all calculated the intake of different categories of omega-3 fatty acids by comparing the food frequency diaries of study subjects to validated standard tables of nutritional components including omega-3 fatty acids. Total omega-3 intake includes all types of omega-3 fatty acids (ALA, EPA, DHA) that can be obtained from food. Fish intake describes the amount of fish consumed whereas marine omega-3 fatty acids describe the amount of ALA, EPA and DHA derived from marine sources.

Meta-Analysis

Trial Summary Statistics

Each trial contained one control or placebo group. Some trials contained more than one treatment (omega-3) group. In order not to double-count patients, we chose the most clinically relevant treatment group to enter our analysis, or in some cases combined treatment groups.

For those outcomes that were dichotomous, the summary statistic was a risk ratio, that is, the risk of the outcome in the treatment (omega-3) group divided by the risk of the outcome in the control or placebo group. A risk ratio greater than one indicates that the risk of the outcome in the treatment group is larger than that in the control or usual care arm. For example, if the risk ratio is 1.10, then patients in the treatment group are 1.10 times as likely to have the outcome as those in the control or placebo group.

For each study, we estimated the log risk ratio and its standard deviation. We conducted the analysis on the logarithmic scale for variance-stabilization reasons.¹⁷ We then back-transformed to the risk ratio scale for interpretability.

For those outcomes that were continuous, we extracted the follow-up means and standard deviations for the treatment and control or placebo groups, respectively. If a study did not report a follow-up mean, or a follow-up mean could not be calculated from the given data, the study was excluded from analysis. For studies that did not report a standard deviation or for which a standard deviation could not be calculated from the given data, we imputed the standard deviation by using those studies and groups that did report a standard deviation and weighting all groups equally, or we assumed that the standard deviation was 0.25 of the theoretical range for the specific measure in the study. For example, if a study measured pain on a 0–100 scale, we assumed the standard deviation was 25.

If all studies measured the outcome on the same scale or the measures could all be converted to the same scale, e.g., the summary statistic was the mean difference (MD) between the treatment group follow-up mean and the control or placebo group follow-up mean:

Mean difference = treatment follow-up mean - control follow-up mean

We estimated the standard deviation for that mean difference.¹⁸ If the studies used different measurements of the same outcome and we could not convert them all to the same scale, the summary statistic was an effect size. The effect size is the mean difference at follow-up divided by the pooled standard deviation. This summary statistic is unitless and indicates the number of standard deviations by which the treatment and control or placebo group means differ. We estimated an unbiased estimate¹⁹ of Hedges' g effect size²⁰ and its standard deviation. A negative mean difference or effect size indicates that the treatment is associated with a decrease in the outcome at follow-up as compared with the control or usual care group.

Performance of Meta-Analysis

In some cases, the trials were judged too clinically heterogeneous to combine. Furthermore, for each outcome, condition, and trial stratum combination, we required that at least three trials be available for pooling. In heterogeneous settings and those with insufficient data, we conduct only a descriptive analysis and present the study-level summary statistics but do not estimate a pooled effect.

For those conditions for which trials were determined to be clinically comparable and for which there were at least three trials, we estimated a pooled random-effects estimate²¹ by combining summary statistics across trials. We also report the chi-squared test of heterogeneity p-value.¹⁹

Forest plots were constructed for each setting. Each individual trial summary statistic is shown as a box whose area is inversely proportional to the estimated variance of the summary statistic in that trial. The trial's confidence interval is shown as a horizontal line through the box. The pooled estimate and its confidence interval are shown as a diamond at the bottom of the plot with a dotted vertical line indicating the pooled estimate value. A vertical solid line at one for dichotomous outcomes or at zero for continuous outcomes indicates no treatment effect.

All analyses and drawings of graphs were conducted in the statistical package Stata (Stata Statistical Software: Release 7.0 2001). The only exception was for the analysis of death. Given that deaths were rare, we used exact conditional inference to perform the pooling rather than applying the usual asymptotic methods that assume normality. Asymptotic methods require corrections if zero events are observed, and generally, half an event is added to all cells in the outcome-by-treatment (two-by-two) table in order to allow estimation, because these methods are based on assuming continuity. Such corrections can have a major impact on the results when the outcome event is rare. Exact methods do not require such corrections. We conducted the meta-analysis using the statistical software package StatXact (StatXact 4 for Windows 2000).

Sensitivity Analyses

We conducted post hoc sensitivity analysis for meta-analyses that exhibited significant (p<0.05) heterogeneity based on the chi-squared test of heterogeneity. In these sensitivity analyses, we removed the most outlying study chosen based on a visual inspection of the forest plot of the original meta-analysis, and estimated a new pooled estimate. We compared this pooled estimate to the original result as well as observed whether significant heterogeneity still remained.

Publication Bias

We assessed the possibility of publication bias by evaluating a funnel plot of summary statistics for asymmetry, which can result from the nonpublication of small trials with negative results. These funnel plots include a horizontal line at the fixed-effects pooled estimate and pseudo-95% confidence limits.²² If bias due to nonpublication exists, the distribution is asymmetric or skewed. Because graphical evaluation can be subjective, we also conducted an adjusted rank correlation test²³ and a regression asymmetry test²² as formal statistical tests for publication bias. The correlation approach tests whether the correlation between the effect sizes and their variances is significant, and the regression approach tests whether the intercept of a regression of the effects sizes on their precision differs from zero; that is, both formally test for asymmetry in the funnel plot. We acknowledge that other factors, such as differences in trial quality or true study heterogeneity, could produce asymmetry in funnel plots.

Interpretation of the Results

The mean difference pooled results are readily interpretable as they are measured in a clinically interpretable metric. To aid in interpreting the pooled effect size and risk ratio, whenever possible we back-transformed each pooled estimate to a specific metric. In order to do this, we multiplied each pooled effect size estimate by the average standard deviation of the most clinically relevant outcome measured across the trials, e.g., included in the pooled estimate.

Peer Review

Results of Literature Search

Tumor Incidence and Outcomes After Cancer Treatment

Figure 3.1 Literature flow to assess the effects of (more...)

   Figure 3.1 Literature flow to assess the effects of omega-3 FA on tumor incidence and treatment

Figure 3.1 displays the flow of the literature review to assess the effects of omega-3 FA on tumor incidence and treatment.

To assess the effects of omega-3 FA on tumor incidence and treatment, the University of Ottawa EPC e-mailed us a total of 4,729 citations as a result of their computerized library searches; our reviewers found 93 additional citations after reference mining; a request for unpublished data yielded one citation; peer reviewers of a draft of this report identified 11 more citations. In total we reviewed 4,834 citations. Our reviewers considered 1,238 of these article titles to be relevant to our research topics. We were able to retrieve 1,210 (98%) of these articles.

Of the articles retrieved, 356 were accepted for further review because they reported on results from randomized clinical trials, controlled clinical trials, or prospective cohort studies of omega-3 FA in the treatment of cancer. We rejected 854 at this stage: 283 were reviews and meta-analyses, 328 reported on a topic other than omega-3 FA, 112 did not report on a population of interest, 26 had descriptive study designs, 89 had other inappropriate study designs, 14 either reported on a condition other than those of interest or did not describe the effect of omega-3 FA on these outcomes, and two were written in foreign languages for which we did not have translators.

Of the 356 articles that went to further review, a total of 263 were rejected. Among those rejected, we were unable to compare the effect of omega-3 FA across study arms in 39. The remaining 224 were rejected for study design (i.e., case control/case series). Thus, a total of 93 articles were tentatively accepted for supplementary analysis. However, on further inspection, 41 of these articles did not report on outcomes of interest and/or we were not able to compare the effects of omega-3 FA across study arms, leaving 52 articles for the final analysis. Of these 52, 33 reported on cancer incidence and 19 reported on cancer treatment. Of the 19 articles that reported on cancer treatment, all reported on cancer surgery; none reported on chemotherapy or radiation therapy. Some articles assessed more than one cancer surgery outcome: 14 assessed post-operative complications, 13 assessed length of stay, 10 assessed mortality, 11 assessed nutrition, and three assessed body weight.

As noted above, an additional 11 articles not identified in our initial search were recommended by external reviewers who reviewed a first draft of this report. Among those studies, 3 met our inclusion criteria and were added to the report.

Tumor Behavior

Figure 3.2 Literature flow to assess the effects of (more...)

   Figure 3.2 Literature flow to assess the effects of omega-3 FA on tumor behavior

Figure 3.2 displays the flow of the literature reviews to assess the effects of omega-3 FA on tumor growth, differentiation, and apoptosis.

To assess the effects of omega-3 FA on tumor growth differentiation and apoptosis, the University of Ottawa EPC e-mailed us a total of 366 citations as a result of their computerized library searches, and our reviewers found three citations after reference mining, for a total of 369 citations. Our reviewers considered 82 of these article titles to be relevant to our research topics. We were able to retrieve 60 (73%) of these articles.

Of the 60 articles retrieved, 27 were accepted for further review, because they appeared to report on the effects of omega-3 FA (added to the diet or to cell cultures) on cancer development, apoptosis, or cell differentiation in laboratory animals or cell culture systems. The other 37 articles were rejected because they did not report on a topic of interest (26), were not about omega-3 FA (7), were not about supplementation (1), were about other mechanisms (2), were reviews (1), or were not about cancer development (1).

Summaries of the 27 accepted articles can be found in Appendix C. Table C.3.1 summarizes the findings for the systematic reviews and meta-analyses, and Table C.3.2 summarizes the findings for the nonsystematic reviews of tumor growth. Table C.3.3 summarizes the findings relevant to differentiation. Table C.3.4 summarizes the findings regarding apoptosis. Table C.3.5 summarizes the evidence related to a role for n-3 transport and metabolic enzyme genes. These findings are described qualitatively below as responses to the questions posed.

Tumor Incidence

Table 3.1 Prospective observational studies of cancer (more...)

Table 3.1 Prospective observational studies of cancer incidence by cancer type and cohort

Cohort	Cancer Type
	Aerodigestive, upper	Bladder	Breast	Colorectal	Lung	Lymphoma, Non-hodgkin's	Ovarian	Pancreatic	Prostate	Skin, BCC	Stomach
Aichi Prefecture Cohort, Japan					Takezaki, 200324
Alpha-tocopherol, Beta-Carotene Cancer Prevention Study								Stolzenberg-Solomon, 200225
Diet, Cancer and Health Study			Stripp, 200355
Fukuoka Prefecture Cohort, Japan											Ngoan, 200226
Hawaii Health Surveillance Program									LeMarchand, 199427
Health Professionals Follow-up Study				Giovannucci, 199430					Giovannucci, 199329;	Van Dam, 200031
									Augustsson, 2003;28
									Leitzman, 200457
Honolulu Heart Program	Chyou, 199532	Chyou, 199333
Iowa Women's Health Study				Bostick, 199434		Chiu, 199635
Japan Collaborative Cohort					Ozasa, 200136
Life Span Study			Key, 199952
Netherlands Cohort Study			Voorips, 200237	Goldbohm, 199438					Schuurman, 199939
New York University Women's Health Study				Kato, 199740
Norwegian National Health Screening Service Cohort			Vatten, 199041		Veierod, 199742
Norwegian Cohorts					Kvale, 198356
Nurses' Health Study			Holmes 199944;	Willett, 199046		Zhang, 199947	Bertone, 200248	Michaud, 200349
			Holmes, 200343
Seventh-day Adventist									Mills, 198950
Singapore Chinese Health Study			Gago-Dominguez, 200351
Swedish Twin Registry									Terry, 200153
Swedish Women in Mammography Screening Program				Terry, 200154

Cohort Characteristics

Table 3.2 Characteristics of cohorts that have described (more...)

Table 3.2 Characteristics of cohorts that have described the effects of omega-3 FA on cancer incidence

Cohort	Author, year	Cancer type	# subjects in cohort*	Birth years	Enrollment period	Observation period, exposure to omega-3	Ascertainment of omega-3 exposure	Observation period, cancer	Ascertainment of cancer	Base-population	Predominant race/ethnicity	Gender(s) in cohort
Aichi Prefecture Cohort, Japan	Takezaki, 200324	Lung	9,753	1917-1972	1986-1989	Enrollment	Food frequency questionnaire	ND	ND	Population of Aichi Prefecture	Japanese
Alpha-tocopherol, Beta-Carotene Cancer Prevention Study	Stolzenberg-Solomon, 200225	Pancreatic	27,111	1916-1938	1985-1988	Enrollment	Food frequency questionnaire about 1-year prior to enrollment	1985-1997	Tumor registry with medical records verification	Male smokers	Caucasian	Male
Diet, Cancer and Health Study	Stripp, 200355	Breast	29,875	1929-1947	1993-1997	Enrollment	Food frequency questionnaire	1993-2000	Cancer registry	Population of greater Copenhagen and Aarhus	Caucasian	Male and Female
												Female for substudy reported here
Fukuoka Prefecture Cohort, Japan	Ngoan, 2002 26	Stomach	13,250	1880-1974	1986-1989	Enrollment	Dietary questionnaire	Not stated	Not explicitly stated; infer death certificates from text.	Population of Fukuoka Prefecture	Japanese	Male and Female
Hawaii Health Surveillance Program	LeMarchand, 199427	Prostate	8,881	ND	1975-1980	1975-1980	Lifestyle questionnaire	1975-1989	Hawaii tumor registry	Hawaiians of Japanese, Caucasian, Filipino, Hawaiian or Chinese ancestry	Caucasian, Asian, Pacific Islander	Male
Health Professionals Follow-up Study	Augustsson, 200328	Prostate	51,529	1911-1946	1986	1986, 1990, 1994	Food frequency questionnaire	1986-1998	self-report or vital records confirmed by medical records review	Male dentists, optometrist, osteopaths, podiatrists, pharmacists, and veterinarians that responded to a postal questionnaire	Caucasian	Male
	Giovannucci, 199329	Prostate
	Giovannucci, 199430	Colorectal
	Leitzmann, 200457	Prostate
	VanDam, 200031	Skin, basal cell carcinoma
Honolulu Heart Program	Chyou, 1995 32	Upper Aero-digestive	8,006	1900-1919	1965-1968	1965-1968	Food frequency questionnaire and 24-hr diet recall history	1965-1993	Oahu hospitalizations for cancer and Hawaii Tumor Registry†	Institutionalized American men of Japanese ancestry residing on Oahu.	Hawaiians of Japanese ancestry	Male
	Chyou, 1993 33	Bladder
Iowa Women's Health Study	Bostick, 199434	Colorectal	41,837	1917-1931	1986	1986	Food frequency questionnaire re: prior 1-year	1986-1992	State Health Registry of Iowa	Women with valid Iowa driver's license	Caucasian	Female
	Chiu, 199635	Non-Hodgkin's lymphoma
Japan Collaborative Cohort	Ozasa, 200136	Lung	110,792	1909-1950	1988-1990	At enrollment	Food frequency questionnaire	1988-1997	Death certificates	Population of 19 prefectures in Japan	Japanese	Male and Female
Life Span Study	Key, 199952	Breast	Approx. 120,000	NR	1969-1970	1969-1970, 1979	Food frequency questionnaire	1969-1993, 1981-1983	Hiroshima and Nagasaki cancer Registries	Survivors of atomic bomb in Hiroshima or Nagasaki, Japan that were alive on September 1, 1969	Asian	Male and Female
Netherlands Cohort Study	Voorrips, 200237	Breast	62,573	1917-1931	1986	1986	Food frequency questionnaire	1986-1992	Regional cancer registries	Population	Caucasian/Dutch	Male and female
	Goldbohm, 199438	Colorectal
	Schuurman, 199939	Prostate
New York University Women's Health Study	Kato, 199740	Colorectal	14,727	1920-1957	1985-1991	At enrollment	Dietary questionnaire	1985-1992	Self report confirmed by medical records review supplemented by review of state cancer registries and National Death index	Women treated at the Guttman Breast Diagnostic Institute in New York City or at the Strax Breast Cancer Institute in Florida	Caucasian, Black, Hispanic	Female
Norwegian Cohorts	Kvale, 198356	Lung	16,713	NR	1964	One- time questionnaire between 1967 and 1969	Dietary questionnaire	From questionnaire until 1978	Cancer registry	Population	Caucasian	Male and Female
Norwegian National Health Screening Service Cohort	Vatten, 199041	Breast	14,729	1925-1942	1974-1977	At enrollment	Food frequency questionnaire and 24-hr diet recall history	11–14 years f/u, mean = 12	National Cancer Registry	Population of Norway	Caucasian	Male and Female
	Veierod, 199742	Lung
	Holmes, 200343	Breast
	Holmes, 199944	Breast
Nurses' Health Study	Willett, 199046	Colorectal	121,700	1921-1946	1976	1980, 1984, 1986, 1990, 1994	Food frequency questionnaire re: prior 1-year	1980-1994	Self-report or vital records confirmed by medical records review	US female registered nurses	Caucasian	Female
	Zhang, 1999 47	Non-Hodgkin's lymphoma
	Bertone, 200248	Ovarian
	Michaud, 200349	Pancreatic
Seventh-day Adventist	Mills, 198950	Prostate	ND	ND	1976	1976	Lifestyle questionnaire	1976-1982	Self-report confirmed by medical records review and Cancer registry	Seventh-day Adventist households in California	ND	Male and Female
Singapore Chinese Health Study	Gago-Dominguez, 200351	Breast	63,257	1919-1953	1993-1998	1-year prior to enrollment	Food frequency questionnaire	Enrollment -2000	Singapore Cancer registry	Permanent residents or citizens of Singapore living in government housing estates† speaking Hokkien or Cantonese	Asian	Male and Female
Swedish Twin Registry	Terry, 200153	Prostate	6272	1886-1925	1961	1967	Lifestyle questionnaire	1967-1997	National Cancer and death registries	Male twin pairs residing in Sweden in 1961	Caucasian	Male
Swedish women in mammography-screening program	Terry, 200154	Colorectal	61,463	1925-1939	1987-1990	6-months prior to enrollment	Food intake questionnaire	Enrollment-1998	Regional cancer registries	Participants of population-based mammography screening program	Caucasian	Female

*

Total number of subjects enrolled in cohort, number may differ from number of subjects in analyses of specific diseases;

† 86% of population lived in this type of housing at the time the cohort was formed.

Figure 3.3 Distribution of fish consumption by cohort (more...)

   Figure 3.3 Distribution of fish consumption by cohort relative to US intake reported in CSFII and NHANES III.*

Figure 3.4 Distribution of omega-3 consumption by cohort (more...)

   Figure 3.4 Distribution of omega-3 consumption by cohort relative to US intake reported in CSFII and NHANES III.*

Figure 3.5 Distribution of ALA consumption by cohort (more...)

   Figure 3.5 Distribution of ALA consumption by cohort relative to US intake reported in CSFII and NHANES III.*

Figure 3.6 Distribution of EPA consumption by cohort (more...)

   Figure 3.6 Distribution of EPA consumption by cohort relative to US intake reported in CSFII and NHANES III.*

Figure 3.7 Distribution of DHA consumption by cohort (more...)

   Figure 3.7 Distribution of DHA consumption by cohort relative to US intake reported in CSFII and NHANES III.*

Figures 3.3

3.7

Figures 3.5

3.7

Figures 3.3

3.7

Figure 3.5

Figure 3.5

Figures 3.6

3.7

Summaries of all evaluated studies can be found in Appendix C.1. The following sections describe the reported effects of omega-3 FA and the incidence of specific types of tumors.

Overall Effect of Omega-3 FA on Tumor Incidence

Figure 3.8 Risk of developing cancer for subjects with (more...)

   Figure 3.8 Risk of developing cancer for subjects with the highest grouping of fish intake relative to subjects with the lowest grouping of intake by cancer type

Figure 3.9 Risk of developing cancer for subjects with (more...)

   Figure 3.9 Risk of developing cancer for subjects with the highest grouping of omega-3 intake relative to subjects with the lowest grouping of intake by cancer type

Figure 3.10 Risk of developing cancer for subjects (more...)

   Figure 3.10 Risk of developing cancer for subjects with the highest grouping of ALA intake relative to subjects with the lowest grouping of intake by cancer type

Figure 3.11 Risk of developing cancer for subjects (more...)

   Figure 3.11 Risk of developing cancer for subjects with the highest grouping of EPA intake relative to subjects with the lowest grouping of intake by cancer type

Figure 3.12 Risk of developing cancer for subjects (more...)

   Figure 3.12 Risk of developing cancer for subjects with the highest grouping of DHA intake relative to subjects with the lowest grouping of intake by cancer type

Figures 3.8

3.12

Aerodigestive Tract Cancer

Table 3.3 Risk of upper aerodigestive cancer for different (more...)

Table 3.3 Risk of upper aerodigestive cancer for different categories of consumption of omega-3 FA, by category.*

Cohort	Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year				Age adjusted RR (95% CI)	Multivariate RR (95% CI)		Multivariate Adjustors
FISH
Honolulu Heart Program	1	NR	< 1 g/week	NR	1		Age, alcohol, number of cigarettes/day, number of years smoked.
Chyou, 199532	2	NR	2–4 g/week	NR	1.02	(0.65, 1.61)
	3	NR	≥ 5 g/week	NR	1.37	(0.70, 2.69)
	Total 7,995		p = 0.473‡

*

NR = Not Reported;

† = Number of people included in analysis;

‡ = test for trend.

Overall effect. We identified one study³² that evaluated the effect of fish consumption on the incidence of upper aerodigestive tract cancer, which was defined as squamous cell carcinoma of the oral cavity/pharynx, esophagus, or larynx. In this study, fish consumption had no significant effect on the incidence of aerodigestive tract cancer. Using fish consumption 1 time per week or less as the referent group, the relative risks of developing aerodigestive tract cancer were 1.02 (0.65–1.61) and 1.37 (0.70–2.69) for men consuming fish 2 to 4 times per week and ≥ 5 times per week or more, respectively (Table 3.3).

Sub-populations. The subjects in this one study were from a distinct population, institutionalized American men of Japanese ancestry who resided on the Hawaiian island of Oahu. Analyses of subpopulations were not performed.

Covariates. The effects of covariates on the effect of fish were not assessed.

Effects of dose, source, and exposure duration. Omega-3 dose was not defined in this study. Rather, the amount of fish consumed was described. As noted above, comparisons between different levels of fish consumption and a referent value did not reveal any statistically significant effects. Additionally, with testing across all exposure levels, the p-value for trend was 0.473. Duration of exposure was not defined in this study, and the effects of different durations of exposure were not tested; usual fish intake at baseline between 1965 and 1968 was determined but not assessed subsequently.

Sustainment of Effect. Sustainment of effect was not assessed.

Table 3.4 Relationship between methodologic quality (more...)

Table 3.4 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of upper aerodigestive cancer.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Honolulu Heart Program	III	Yes	NR	Yes	Yes	Yes
Chyou, 199532

*

NR = Not Reported

Quality and Applicability. See Table 3.4.

Bladder Cancer

Table 3.5 Risk of bladder cancer for different categories (more...)

Table 3.5 Risk of bladder cancer for different categories of consumption of omega-3 FA, by category.*

Cohort	Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year				Age adjusted RR (95% CI)	Multivariate RR (95% CI)		Multivariate Adjustors
FISH
Honolulu Heart Program	1	NR	≤ 1 times/week	NR	1		Age, smoking.
Chyou, 199333	2	NR	2–4 times/week	NR	0.90	(0.59, 1.39)
	3	NR	≥ 5 times/week	NR	0.67	(0.26, 1.67)
	Total 7,995		p = 0.377‡

*

NR = Not Reported;

† = Number of people included in analysis;

‡ = test for trend.

Overall effect. We identified one study³³ that evaluated the effect of fish consumption on the incidence of urinary bladder cancer. In this study, fish consumption had no significant effect on the incidence of bladder cancer. Using fish consumption 1 time per week or less as the referent group, the relative risks of developing bladder cancer were 0.90 (0.59–1.39) and 0.67 (0.26–1.67) for men consuming fish 2 to 4 times per week and 5 times per week or more, respectively (Table 3.5).

Sub-populations. The subjects in this one study were from a distinct population, institutionalized American men of Japanese ancestry who resided on the Hawaiian island of Oahu. Analyses of subpopulations were not performed.

Covariates. The effects of covariates on the effect of fish were not assessed.

Effects of dose, source, and exposure duration. Omega-3 dose was not defined in this study. Rather, the amount of fish consumed was described. As noted above, comparisons between different levels of fish consumption and a referent value did not reveal any statistically significant effects. Additionally, with testing across all exposure levels, the p-value for trend was 0.38. Duration of exposure was not defined in this study, and the effects of different durations of exposure were not tested; usual fish intake at baseline between 1965 and 1968 was determined, but not assessed subsequently.

Sustainment of effect. Sustainment of effect was not assessed.

Table 3.6 Relationship between methodologic quality (more...)

Table 3.6 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of bladder cancer.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Honolulu Heart Program	III	Yes	NR	Yes	Yes	Yes
Chyou, 199333

*

NR = Not Reported.

Quality and applicability. See Table 3.6

Breast Cancer

Table 3.7 Risk of breast cancer for different categories (more...)

Table 3.7 Risk of breast cancer for different categories of consumption of omega-3 FA, by category.*

Cohort		Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year					Age adjusted RR (95% CI)		Multivariate RR		Multivariate Adjustors (95% CI)
FISH
Diet, Cancer and Health Study		1	NR	0–26 g/day	1		1		Age, parity, number of births, age at first birth, BMI, benign breast tumor, years of school, use of HRT, duration of HRT use, alcohol.
Stripp, 200355		2	NR	27–39 g/day	1.01	(0.77. 1.32)	.99	(0.76, 1.30)
		3	NR	40–58 g/day	1.17	(0.89, 1.53)	1.12	(0.85, 1.47)
		4	NR	> 58 g/day	1.54	(1.18, 2.02)	1.47	(1.10, 1.98)
		Total 23,693
Nurses' Health Study		1	NR	≤ 0.13 servings/day	NR		1		Age, 2yr time period, total energy, alcohol intake, parity and age at first birth, BMI at age 18, weight change since 18, height in inches, family history of breast cancer, history of benign breast disease, age at menarche in years, menopausal status, age at menopausal and HRT use, duration of menopausal.
Holmes, 200343		2	NR	0.14–0.2 servings/day	NR		.98	(0.89, 1.08)
		3	NR	0.21–0.27 servings/day	NR		.97	(0.87, 1.08)
		4	NR	0.28–0.39 servings/day	NR		.99	(0.90, 1.09)
		5	NR	≥ 0.4 servings/day	NR		1.04	(0.93, 1.14)
		Total 88,647						p = 0.55‡
Life Span Study	Fish, not dry	1	NR	≤ 1 times/week	NR		1		Attained age, calendar period, city, age at time of bombing, and radiation dose.
Key, 199952		2	NR	2–4 times/week	NR		1.08	(0.84, 1.39)
		3	NR	≥ 5 times/week	NR		1.17	(0.90, 1.54)
		4	NR	Unknown	NR		0.92	(0.66, 1.29)
		Total 34,759						p = 0.21‡
	Fish, dry	1	NR	≤ 1 times/week	NR		1
		2	NR	2–4 times/week	NR		0.85	(0.64, 1.12)
		3	NR	≥ 5 times/week	NR		0.49	(0.24, 1.02)
		4	NR	Unknown	NR		0.77	(0.60, 0.98)
		Total 34,759						p = 0.03‡
Norwegian National Health Screening Service Cohort		1	NR	≤ 2 g/week	1§		NR		NR
Vatten, 199041		2	NR	≥ 2 g/week	1.2§	(0.8, 1.7)	NR
		Total 14,500				p = 0.24‡
OMEGA-3
Singapore Chinese Health Study		1	NR	NR	NR		1		Age at baseline interview, year of recruitment, dialect group, education, daily alcohol drinker, family history of breast cancer, age when period became regular, number of live births.
Gago-Dominguez, 200351		2	NR	NR	NR		0.82	(0.60, 1.1)
		3	NR	NR	NR		0.84	(0.62, 1.15)
		4	NR	NR	NR		0.87	(0.64, 1.18)
		Total 35,298						p = 0.40‡
ALA
Netherlands Cohort Study		1	NR	0.6	1		1		Age, history of benign breast cancer, breast cancer in one or more sisters, age at menarche, age at menopause, oral contraceptive use, parity, age at first childbirth, Quetelet index, education, alcohol use, current cigarette smoking, total energy intake, total energy-adjusted fat intake.
Voorips, 200237		2	NR	0.8	0.76	(0.58, 1.00)	0.78	(0.57, 1.05)
		3	NR	1.0	0.92	(0.71, 1.20)	1.03	(0.76, 1.39)
		4	NR	1.3	0.69	(0.52, 0.91)	0.74	(0.54, 1.00)
		5	NR	1.7	0.68	(0.51, 0.91)	0.70	(0.51, 0.97)
		Total 62,573				p = 0.001‡		p = 0.006‡
EPA
Netherlands Cohort Study		1	NR	0 g/d	1		1		Age, history of benign breast cancer, breast cancer in one or more sisters, age at menarche, age at menopause, oral contraceptive use, parity, age at first childbirth, Quetelet index, education, alcohol use, current cigarette smoking, total energy intake, total energy-adjusted fat intake.
Voorips, 200237		2	NR	0.01 g/d	1.18	(0.88, 1.56)	1.15	(0.84, 1.58)
		3	NR	0.02 g/d	1.14	(0.87, 1.50)	1.10	(0.82, 1.49)
		4	NR	0.04 g/d	1.23	(0.93, 1.62)	1.22	(0.90, 1.65)
		5	NR	0.08 g/d	1.03	(0.78, 1.37)	0.98	(0.72, 1.35)
		Total 62,573				p = 0.63‡		p = 0.87‡
DHA
Netherlands Cohort Study		1	NR	0.01	1		1		Age, history of benign breast cancer, breast cancer in one or more sisters, age at menarche, age at menopause, oral contraceptive use, parity, age at first childbirth, Quetelet index, education, alcohol use, current cigarette smoking, total energy intake, total energy-adjusted fat intake.
Voorips, 200237		2	NR	0.03	1.11	(0.83, 1.47)	1.10	(0.81, 1.51)
		3	NR	0.05	1.04	(0.78, 1.37)	1.03	(0.76, 1.40)
		4	NR	0.08	1.20	(0.91, 1.58)	1.21	(0.90, 1.64)
		5	NR	0.14	1.02	(0.77, 1.36)	1.00	(0.72, 1.37)
		Total 62,573				p = 0.62‡		p = 0.70‡

*

NR = Not Reported;

† = Number of people included in analysis;

‡ = test for trend;

§ = incidence rate ratio.

Overall effect. We identified seven studies^{37, 41, 43, 44, 51, 52, 55} from six different cohorts that evaluated the effect of omega-3 FA on the incidence of breast cancer. Breast cancer incidence relative to fish consumption was reported in four studies,^{41, 43, 52, 55} incidence relative to total and marine omega-3 fatty acid consumption was reported in one,⁵¹ and incidence relative to each of the specific omega-3 FA, DHA, EPA and ALA was reported in one.³⁷ No significant overall association with the incidence of breast cancer was found with fish, total omega-3 FA, DHA, or EPA consumption (Table 3.7). In one study,⁵⁵ women in the highest quartile of fish intake had an increased risk of breast cancer relative to women in the lowest quartile of fish intake (IRR 1.47; 95% CI 1.10, 1.98). Omega-3 FA consumption from marine sources and ALA consumption were associated with a reduced risk of developing breast cancer. Women in the highest quartile of consumption of marine omega-3 FA had a lower incidence of breast cancer than women in the lowest quartile of consumption (RR 0.72, 95% CI 0.53, 0.98). Women in the highest quintile of ALA consumption had a significantly lower incidence of breast cancer than women in the lowest quintile of consumption. This observation held true with adjustment for both age (RR 0.68; 95% CI 0.51, 0.91) and multiple variables (RR 0.70; 95% CI 0.51, 0.97). Associations between ALA consumption and breast cancer incidence were not significant for comparisons between the other quintiles and the lowest quintiles.

Sub-populations. All analyses were restricted to women of racial groups that were homogeneous within, but that differed across, the studies. The four studies that assessed the association between fish consumption and breast cancer incidence used cohorts from the US (Nurses Health Study), Denmark (Diet Cancer and Health Study), and Norway (Norwegian National Health Cohort). The study that assessed the association between the specific omega-3 FA ALA, DHA and EPA used a cohort of women residing in the Netherlands (Netherlands Cohort Study). The study that assessed the association between total omega-3 FA consumption and breast cancer risk used a cohort of Chinese women residing in Singapore (Singapore Chinese Health Study). In this last study, subgroup analyses revealed that the reduced incidence of breast cancer associated with marine omega-3 FA consumption was confined to postmenopausal women and to women with advanced stage disease (stage II or greater). The Nurses Health Study also compared the effect of marine omega-3 FA on premenopausal and postmenopausal women.⁴⁴ In this study, a small increased risk of breast cancer was seen among postmenopausal women (RR 1.09; 95% CI 1.02, 1.17), but no significant association was seen overall or for premenopausal women (Table 3.7).

Covariates. The effects of covariates on the effect of omega-3 FA on incidence of breast cancer were assessed in four of the studies. In one study, the risk of developing breast cancer associated with fish intake was not affected by family history of breast cancer, multivitamin use, or glycemic load in separate analyses.⁴³ In another study, occupational status and BMI did not affect the reported association between fish consumption and breast cancer incidence.⁴¹

One study examined the relationship between breast cancer incidence, marine omega-3 FA intake, and omega-6 FA intake.⁵¹ In this study, among subjects in the lowest quartile of marine omega-3 FA consumption, breast cancer risk increased significantly with increasing levels of omega-6 FA consumption (p for trend = 0.08). Relative to women in the lowest quartile of both omega-6 and marine omega-3 consumption, the relative risk of developing breast cancer for women in both the lowest quartile of omega-3 consumption and the highest quartile of omega-6 consumption was 1.87 (95% CI, 1.06, 3.27).

One study examined the relationship between fish intake, estrogen receptor (ER) positivity, and cancer incidence.⁵⁵ In this study, the incidence rate ratio (IRR) for breast cancer per mean intake of 25 g/d of fish was 1.14 (955 CI 1.03, 1.26) for ER-positive women and 1.00 (95% CI 0.81, 1.24) for ER-negative women.

Effects of dose, source, and exposure duration

Dose: Each of the studies assessed the effects of dose. No dose effect was observed for fish, total omega-3, DHA, or EPA consumption (Table 3.7). However, dose effects were demonstrated for marine omega-3 FA⁵¹ and ALA³⁷ (p for trend < 0.05).

Source: No effects were observed for fish in two studies.^{41, 43} One study demonstrated a reduced risk for marine omega-3 but not for total omega-3 FA.⁵¹ One study demonstrated reduced risk for ALA but not EPA or DHA.³⁷ (Table 3.7).

Exposure duration: Three of the studies identified assessed exposure at baseline only; the follow-up period in these studies ranged from 2 to 12 years.^{37, 41, 51} These studies did not assess the effect of exposure duration. Two cohorts assessed exposure at multiple time points. The Life Span Study⁵² and Nurses Health Study^{43, 44} collected dietary data at two and four time points, respectively. The Life Span Study found no difference in cancer risk associated with soy products (no association) using dietary data from either dietary survey; this study did not report the effect of exposure duration for fish on the risk of breast cancer. The Nurses Health Study assessed the associations of diet with breast cancer when the diet was assessed only at baseline and also when diet was updated over time without cumulatively averaging in prior intake;⁴³ results did not change with these analyses.

Sustainment of effect. None of the studies specifically assessed sustainment of effect.

Table 3.8 Relationship between methodologic quality (more...)

Table 3.8 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of breast cancer.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Diet, Cancer and Health Study	II	Yes	NR	Yes	Yes	Yes
Stripp55
Life Span Study	III	Yes	NR	Yes	Yes	Yes
Key, 199952
Netherlands Cohort Study	II	Yes	Yes	Yes	Yes	Yes
Voorips, 200237
Norwegian National Health Screening Service Cohort	II	Yes	NR	Yes	Yes	Yes
Vatten, 199041
Nurses' Health Study	II	Yes	Yes	Yes	Yes	Yes
Holmes, 200343
Singapore Chinese Health Study	II	Yes	NR	Yes	Yes	No
Gago-Dominguez, 200351

*

NR = Not Reported.

Quality and applicability. See Table 3.8

Colorectal Cancer

Table 3.9 Risk of colorectal cancer for different categories (more...)

Table 3.9 Risk of colorectal cancer for different categories of consumption of omega-3 FA, by category.*

Cohort		Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year					Age adjusted RR (95% CI)		Multivariate RR (95% CI)		Multivariate Adjustors
FISH
Health Professionals Follow-up Study		1	NR	8.4 g/d	1		NR		NR
Giovannucci, 199430		2	NR	20.9 g/d	0.85	(0.54, 1.33)	NR
		3	NR	31.0 g/d	1.05	(0.68, 1.61)	NR
		4	NR	47.8 g/d	0.80	(0.51, 1.26)	NR
		5	NR	83.4 g/d	1.06	(0.70, 1.60)	NR
		Total 47,949				p = 0.79‡
Netherlands Cohort Study		1	NR	0 g/d	NR		1		Age and energy.
Goldbohm, 199438		2	NR	0–10 g/d	NR		1	(0.68, 1.47)
		3	NR	10–20 g/d	NR		0.74	(0.48, 1.15)
		4	NR	> 20 g/d	NR		0.81	(0.56, 1.17)
		Total 3,111						p = 0.14‡
Nurses' Health Study		1	NR	< 1 g/month	1		NR		NR
Willett, 199046		2	NR	1–3 g/month	1.29	(0.70, 2.40)	NR
		3	NR	1 g/week	0.92	(0.49, 1.72)	NR
		4	NR	2–4 g/week	0.75	(0.35, 1.58)	NR
		5	NR	4 g/week	1.06	(0.36, 3.12)	NR
		Total 88,751				p = 0.09‡
New York University Women's Health Study		1	NR	NR	NR		1		Age, total calorie, place at enrollment and highest level of education.
Kato, 199740		2	NR	NR	NR		1.01	(0.62, 1.67)
		3	NR	NR	NR		0.65	(0.37, 1.13)
		4	NR	NR	NR		0.49	(0.27, 0.89)
		Total 14,727						p = 0.007‡
Omega-3
Iowa Women's Health Study		1	NR	< 0.03 g/day	1		1		Age, total energy intake, height, parity, total vitamin E, a total vitamin E by age interaction term, vitamin A supplement intake.
Bostick, 199434		2	NR	0.03–0.05 g/day	0.67	NR	0.82	(0.55, 1.24)
		3	NR	0.06–0.10 g/day	0.61	NR	0.77	(0.50, 1.17)
		4	NR	0.11–0.18 g/day	0.72	NR	0.96	(0.64, 1.43)
		5	NR	> 0.18 g/day	0.60	NR	0.70	(0.45, 1.09)
		Total 35,215			p = 0.04‡			p = 0.26‡
ALA
Swedish women in mammography-screening program	Colorectal	1	NR	0.45 g/d	NR		1		Age, BMI, education level, energy intake, intakes of red meat and alcohol, energy, dietary fiber, calcium, vitamin C, folic acid, Vitamin D, saturated fat, monounsaturated fat, polyunsaturated fat.
Terry, 200154		2	NR	0.50 g/d	NR		0.96	(0.73, 1.27)
		3	NR	0.54 g/d	NR		0.96	(0.72, 1.28)
		4	NR	0.70 g/d	NR		0.99	(0.75, 1.32)
		Total 61,463						p = 0.99‡
	Colon	1	NR	0.45 g/d	NR		1
		2	NR	0.50 g/d	NR		0.96	(0.68, 1.35)
		3	NR	0.54 g/d	NR		0.96	(0.67, 1.3)
		4	NR	0.70 g/d	NR		0.90	(0.63, 1.28)
		Total 61,463						p = 0.57‡
	Rectal	1	NR	0.45 g/d	NR		1
		2	NR	0.50 g/d	NR		0.95	(0.60, 1.52)
		3	NR	0.54 g/d	NR		0.92	(0.56, 1.49)
		4	NR	0.70 g/d	NR		1.11	(0.70, 1.78)
		Total 61,463
EPA
Swedish women in mammography-screening program	Colorectal	1	NR	0.03 g/d	NR		1		Age, BMI, education level, energy intake, intakes of red meat and alcohol, energy, dietary fiber, calcium, vitamin C, folic acid, Vitamin D, saturated fat, monounsaturated fat, polyunsaturated fat.
Terry, 200154		2	NR	0.05 g/d	NR		0.80	(0.68, 1.15)
		3	NR	0.07 g/d	NR		0.96	(0.73, 1.26)
		4	NR	0.09 g/d	NR		0.96	(0.72, 1.28)
		Total 61,463						p = 0.91‡
	Colon	1	NR	0.03 g/d	NR		1
		2	NR	0.05 g/d	NR		0.76	(0.54, 1.06)
		3	NR	0.07 g/d	NR		0.81	(0.58, 1.15)
		4	NR	0.09 g/d	NR		0.85	(0.60, 1.21)
		Total 61,463						p = 0.46‡
	Rectal	1	NR	0.03 g/d	NR		1
		2	NR	0.05 g/d	NR		1.17	(0.75, 1.83)
		3	NR	0.07 g/d	NR		1.29	(0.80, 2.06)
		4	NR	0.09 g/d	NR		1.25	(0.75, 2.06)
		Total 61,463						p = 0.35‡
DHA
Swedish women in mammography-screening program	Colorectal	1	NR	0.08 g/d	NR		1		Age, BMI, education level, energy intake, intakes of red meat and alcohol, energy, dietary fiber, calcium, vitamin C, folic acid, Vitamin D, saturated fat, monounsaturated fat, polyunsaturated fat.
Terry, 200154		2	NR	0.11 g/d	NR		0.88	(0.67, 1.15)
		3	NR	0.13 g/d	NR		0.87	(0.66, 1.15)
		4	NR	0.18 g/d	NR		0.90	(0.67, 1.20)
		Total 61,463						p = 0.52‡
	Colon	1	NR	0.08 g/d	NR		1
		2	NR	0.11 g/d	NR		0.84	(0.60, 1.17)
		3	NR	0.13 g/d	NR		0.74	(0.51, 1.06)
		4	NR	0.18 g/d	NR		0.88	(0.61, 1.26)
		Total 61,463						p = 0.41‡
	Rectal	1	NR	0.08 g/d	NR		1
		2	NR	0.11 g/d	NR		1.03	(0.66, 1.61)
		3	NR	0.13 g/d	NR		1.16	(0.73, 1.8)
		4	NR	0.18 g/d	NR		1.03	(0.62, 1.71)
		Total 61,463						P=0.79‡

*

NR = Not Reported;

† Number of people included in analysis;

‡ = test for trend.

Overall effect. We identified six studies^{30, 34, 38, 40, 46, 54} from six different cohorts that evaluated the effect of omega-3 FA on the incidence of colorectal cancer. Colorectal cancer incidence relative to fish consumption was reported in four studies,^{30, 38, 40, 46} incidence relative to total omega-3 fatty acid consumption was reported in one,³⁴ and incidence relative to each of the specific omega-3 FA, DHA, EPA and ALA was reported in one.⁵⁴ Among the studies that measured fish consumption, three found no association with the incidence of colorectal cancer;^{30, 38, 46} one study⁴⁰ demonstrated a reduced risk among subjects in the highest quartile of fish intake relative to subjects in the lowest quartile of fish intake (RR 0.49, 95% CI 0.27, 0.89). The one study that measured total omega-3 FA consumption³⁴ demonstrated a trend for reducing the risk of colorectal cancer with higher consumption of omega-3 FA when adjusting only for age. However, with adjustment for multiple variables no significant association was observed between omega-3 fatty acid consumption and the incidence of colorectal cancer. No significant association with the incidence of colorectal cancer was found with ALA, DHA, or EPA consumption⁵⁴ (Table 3.9).

Sub-populations. Three of the studies were among cohorts of women,^{34, 40, 46} one among a cohort of men,³⁰ and two among cohorts that included both men and women.^{38, 54} Among the latter, one study performed subgroup analyses among men and women and found no association between fish consumption and colon cancer for men or women.³⁸ The one study that demonstrated a favorable association between a source of omega-3 FA and incidence of colorectal cancer after adjustment for multiple variables was performed in a cohort of women.⁴⁰

Three of the studies assessed the incidence of colon cancer only^{34, 38, 46} and three assessed the incidence of colorectal cancer including cancers of the colon or rectum.^{30, 40, 54} In the one study that assessed the incidence of colon cancer, rectal cancer, and colorectal cancer,⁵⁴ there was no difference in the association between ALA, EPA, or DHA intake and the incidence of any of these types of cancer, i.e., there was no association in any case. The one study that demonstrated a favorable association between a source of omega-3 FA and incidence of colorectal cancer after adjustment for multiple variables included both cancers of the colon and rectum to define colorectal cancer.⁴⁰

Covariates. Although each of the studies performed multivariable analyses, the effects of specific covariates were not reported.

Effects of dose, source, and exposure duration

Dose: Each of the studies assessed the effects of dose. The one study⁴⁰ that demonstrated a reduced risk of colorectal cancer among subjects in the highest quartile of fish intake relative to subjects in the lowest quartile of fish intake also reported a significant test for trend across all quartiles (p = 0.007). However, comparisons of cancer incidence between the first quartile and each of the second and third quartiles of fish intake did not yield significant results. One additional study³⁴ demonstrated a trend for reducing the risk of colorectal cancer with higher consumption of omega-3 FA, when adjusting only for age. However, there was no significant dose effect with adjustment for multiple variables. None of the other studies demonstrated a dose effect.^{30, 38, 46, 54}

Source: One study demonstrated a reduced risk for fish;⁴⁰ three did not.^{30, 38, 46} One study demonstrated a reduced risk for omega-3 FA consumption that was not significant after adjustment for multiple variables.³⁴ One study assessed the effects of different types of omega-3 FA on the incidence of colorectal cancer and found no association with ALA, DHA, or EPA consumption.⁵⁴

Exposure duration: Four of the studies assessed exposure at baseline only,^{34, 38, 40, 54} and two assessed exposure at multiple time points. However, none specifically assessed the effect of exposure duration on the incidence of colorectal cancer.

Sustainment of Effect. None of the studies specifically assessed sustainment of effect.

Table 3.10 Relationship between methodologic quality (more...)

Table 3.10 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of colorectal cancer.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Health Professionals Follow-up Study	II	Yes	Yes	Yes	Yes	Yes
Giovannucci, 199430
Netherlands Cohort Study	II	Yes	NR	Yes	Yes	No
Goldbohm, 199438
Nurses' Health Study	II	Yes	Yes	Yes	Yes	Yes
Willett, 199046
New York University Women's Health Study	III	Yes	NR	Yes	Yes	Yes
Kato, 199740
Iowa Women's Health Study	II	Yes	NR	Yes	Yes	Yes
Bostick, 199434
Swedish women in mammography-screening program	II	Yes	No	Yes	Yes	NR
Terry, 200154

*

NR = Not Reported.

Quality and applicability. See Table 3.10

Lung Cancer

Table 3.11 Risk of lung cancer for different categories (more...)

Table 3.11 Risk of lung cancer for different categories of consumption of omega-3 FA, by category.*

Cohort		Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year					Age adjusted RR (95% CI)	Multivariate RR (95% CI)		Multivariate Adjustors
FISH
Aichi Prefecture Cohort, Japan		1	174	< 1 times/week	NR	1		Age, sex, smoke, occupation.
Takezaki, 200324		2	1,264	1–2 times/week	NR	0.99	(0.48, 2.03)
		3	1,360	≥ 3 times/week	NR	0.32	(0.13, 0.76)
		Total 5,885					p = 0.003‡
Japan Collaborative Cohort	Men	1	NR	≤ 1–2 times/week	NR	1§		Age, parent's history of lung cancer, smoking status, smoking index and time since quitting smoking.
Ozasa, 200136		2	NR	3–4 times/week	NR	1.12§	(0.87, 1.43)
		3	NR	almost every day	NR	1.03§	(0.79, 1.34)
		Total 42,940					p = 0.72‡
	Women	1	NR	≤ 1–2 times/week	NR	1
		2	NR	3–4 times/week	NR	0.73	(0.45, 1.21)
		3	NR	almost every day	NR	0.88	(0.52, 1.49)
		Total 55,308						p = 0.50‡
Norwegian Cohorts	Histologic verification	1	NR	< 10 times/month	NR	1		Age, cigarette smoking, region and urban/rural place of residence.
Kvale, 198356		2	NR	10–14 times/month	NR	NR
		3	NR	15–19 times/month	NR	NR
		4	NR	≥ 20 times/month	NR	0.82	NR
		Total 13785						p = 0.63‡
	Squamous and small-cell carcinomas	1	NR	< 10 times/month	NR	1
		2	NR	10–14 times/month	NR	NR
		3	NR	15–19 times/month	NR	NR
		4	NR	≥ 20 times/month	NR	0.98	NR
		Total 13785						p = 0.99‡
Norwegian National Health Screening Service Cohort		1	NR	<1 times/week		1§		Smoking status, gender, age at inclusion, attained age.
Veierod, 199742		2	NR	1–2 times/week		1.1||	(0.6, 2.2)
		3	NR	3–4 times/week		1.0||	(0.5, 2.1)
		4	NR	≥ 5 times/week		3.0||	(1.2, 7.3)
		Total 51,452						p = 0.2‡

*

NR = Not Reported;

† Number of people included in analysis;

‡ = test for trend;

§ Hazard Ratio;

|| Incidence Rate Ratio.

Overall effect. We identified three studies^{24, 42, 56} from three different cohorts that evaluated the effect of omega-3 FA on the incidence of lung cancer and one that evaluated the effect of omega-3 FA intake on death from lung cancer.³⁶ All of these studies assessed lung cancer incidence relative to fish consumption (Table 3.11). In one study,²⁴ fish consumption was associated with a reduced risk of lung cancer (RR 0.32, 95% CI 0.13, 0.76). In the other studies, no significant association was found between fish intake and lung cancer incidence^{42, 56} or death from lung cancer.³⁶

Sub-populations. Each of the cohorts was population-based and included men and women. The base population comprised residents of a single rural prefecture in Japan in one study,²⁴ 19 Japanese prefectures in another study,³⁶ and people residing in Norway in the other two.^{42, 56} One study reported the risk of dying from lung cancer stratified by gender.³⁶ This study found no significant association between fish consumption and death from lung cancer for either men or women (Table 3.11).

Covariates. The effects of different methods of cooking fish on the incidence of lung cancer were assessed in one study.²⁴ Consumption of fish that had been broiled or boiled was associated with reduced risk for lung cancer (p values for trend < 0.02). No significant reduction in risk of lung cancer was found for consumption of fish that was raw or deep-fried.

Effects of dose, source, and exposure duration

Dose: Three of the studies assessed the effects of dose.^{24, 36, 42} The study that reported a reduced risk of lung cancer with fish consumption, also reported a dose effect.²⁴ Subjects in each the middle and high consumption categories had a lower risk relative to subjects in the lowest category of consumption and the risk decreased with higher consumption (p for trend = 0.003). No overall or dose effect was observed in the other studies.^{24, 42}

Source: The source of omega-3 fatty acid was fish in each of the studies.

Exposure duration: Each of the studies assessed fish consumption at baseline only; the follow-up period in these studies ranged from 8 to 14 years. None of the studies assessed the effect of exposure duration.

Sustainment of effect. Neither of the studies specifically assessed sustainment of effect.

Table 3.12 Relationship between methodologic quality (more...)

Table 3.12 Relationship between methodologic quality and applicability for estimates of omega-3 fatty acid consumption on risk of lung cancer.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Aichi Prefecture Cohort, Japan	II	Yes	NR	Yes	Yes	NR
Takezaki, 200324
Japan Collaborative Cohort	II	Yes	NR	Yes	Yes	Yes
Ozasa, 200136
Norwegian Cohorts	II	Yes	NR	Yes	Yes	Yes
Kvale, 198356
Norwegian National Health Screening Service Cohort	II	Yes	NR	Yes	Yes	Yes
Veierod, 199742

*

NR = Not Reported.

Quality and applicability. See Table 3.12.

Lymphoma

Table 3.13 Risk of non-hodgkin's lymphoma for different (more...)

Table 3.13 Risk of non-hodgkin's lymphoma for different categories of consumption of omega-3 FA, by category.*

Cohort	Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year				Age adjusted RR (95% CI)		Multivariate RR (95% CI)		Multivariate Adjustors
FISH
Iowa Women's Health Study	1	NR	< 4 servings/ month	NR		1		Age and energy.
Chiu, 199635	2	NR	4–6 servings/ month	NR		0.94	(0.59, 1.49)
	3	NR	> 6 servings/month	NR		0.81	(0.49, 1.35
	Total 35,136						p = 0.42‡
Omega-3
Nurses' Health Study	1	NR	0.02 % of energy intake	1		1		Age, total energy, length of follow-up, geographic region, cigarette smoke, height in inches, saturated and trans unsaturated fats, fruit, vegetable intake.
Zhang, 199947	2	NR	0.03 % of energy intake	1.2	NR	1.2	NR
	3	NR	0.04% of energy intake	1.3	NR	1.4	NR
	4	NR	0.05% of energy intake	1.1	NR	1.2	NR
	5	NR	0.10% of energy intake	1.1	(0.7, 1.7)	1.4	(0.8, 2.2)
	Total 88,410				p = 0.90‡		Testing NR

*

NR = Not Reported;

† Number of people included in analysis;

‡ = test for trend.

Overall effect. We identified two studies from two different cohorts that evaluated the effect of omega-3 FA on the incidence of non-Hodgkin's lymphoma.^{35, 47} One study assessed incidence relative to fish consumption, the other relative to marine omega-3 fat consumption. Neither study found a significant association between fish intake and the incidence of non-Hodgkin's lymphoma (Table 3.13).

Sub-populations. Both cohorts were restricted to women. The Nurses Health Study cohort includes U.S. female registered nurses who responded to a mailed questionnaire.⁴⁷ The Iowa Women's Health Study cohort includes women who had valid Iowa driver's licenses at the time of recruitment. Analyses on subpopulations were not reported in either study.

Covariates. The effects of covariates on risk associated with omega-3 FA were not reported.

Effects of dose, source, and exposure duration

Dose: Both studies assessed the risk of developing non-Hodgkin's lymphoma given different levels of fish or omega-3 fat consumption and found no dose effect (p for trend > 0.40 for all comparisons).

Source: The source of omega-3 fatty acid was fish in one study³⁵ and marine omega-3 FA in the other.⁴⁷

Exposure duration: Each of the studies assessed fish consumption at baseline only; the follow-up period in these studies ranged from 6 to 14 years. Neither study assessed the effect of exposure duration to omega-3 FA on risk of non-Hodgkin's lymphoma.

Sustainment of effect. Neither of the studies specifically assessed sustainment of effect.

Table 3.14 Relationship between methodologic quality (more...)

Table 3.14 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of non-Hodgkin's lymphoma.*

Cohort	Applicability	Quality Parameters
Author, Year		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Iowa Women's Health Study	II	Yes	NR	Yes	Yes	Yes
Chiu, 199635
Nurses' Health Study	II	Yes	Yes	Yes	Yes	Yes
Zhang, 199947

*

NR = Not Reported.

Quality and applicability. See Table 3.14.

Ovarian Cancer

Table 3.15 Risk of ovarian cancer for different categories (more...)

Table 3.15 Risk of ovarian cancer for different categories of consumption of omega-3 FA, by category.*

Cohort	Study arm (quartile, quintile or dose group)	n†	Median intake	Estimates of effect
Author, Year				Age adjusted RR (95% CI)		Multivariate RR (95% CI)		Multivariate Adjustors
ALA
Nurses' Health Study	1	NR	NR	1.0		1.0		Age, parity, age at menarche, oral contraceptive use and duration, menopausal status/postmenopausal hormone use, smoking status.
Bertone, 200248	2	NR	NR	0.74	NR	0.95	(0.68, 1.33)
	3	NR	NR	0.62	NR	0.8	(0.56, 1.14)
	4	NR	NR	0.86	NR	0.82	(0.58, 1.15)
	5	NR	NR	0.98	NR	0.88	(0.62, 1.24)
	Total 80,258						p = 0.27‡
EPA
Nurses' Health Study	1	NR	NR	1		1		Age, parity, age at menarche, oral contraceptive use and duration, menopausal status/postmenopausal hormone use, smoking status.
Bertone, 200248	2	NR	NR	1.01	NR	1.04	(0.68, 1.59)
	3	NR	NR	0.73	NR	0.75	(0.47, 1.17)
	4	NR	NR	0.96	NR	1.00	(0.66, 1.52)
	5	NR	NR	0.96	NR	0.97	(0.64, 1.48)
	Total 80,258						p = 0.80‡
DHA
Nurses' Health Study	1	NR	NR	1		1		Age, parity, age at menarche, oral contraceptive use and duration, menopausal status/postmenopausal hormone use, smoking status.
Bertone, 200248	2	NR	NR	1.06	NR	1.06	(0.70, 1.61)
	3	NR	NR	0.67	NR	0.67	(0.42, 1.08)
	4	NR	NR	1.05	NR	1.07	(0.71, 1.63)
	5	NR	NR	0.88	NR	0.86	(0.55, 1.33)
	Total 80,258						p = 0.52‡

*

NR = Not Reported;

† Number of people included in analysis;

‡ = test for trend.

Overall Effect. We identified one report⁴⁸ that evaluated the effect of different kinds of fat, including the omega-3 FA DHA, EPA, and ALA, on the incidence of ovarian cancer among women enrolled in the Nurses Health Study. This study found no evidence of an association between intake of any type of fat, including DHA, EPA, and ALA, and the incidence of ovarian cancer (Table 3.15). Secondary analyses showed that total fat intake (i.e., different levels of total fat intake) had no effect on the development of specific subtypes of ovarian cancer (serous, mucinous, and endometrial tumors). However, these analyses were not conducted for omega-3 FA specifically.

Sub-populations. The subjects in this study were all female registered nurses in the US. The effect of total fat intake, but not omega-3 FA intake was assessed for several different subpopulations. The relation between fat intake and ovarian cancer risk (i.e., no association) did not differ substantially by age or menopausal status.

Covariates. The effects of several covariates on the effect of total fat intake but not omega-3 fat were assessed. Neither body mass index, oral contraceptive use, smoking status, nor physical activity level had an effect on the relation between fat intake and ovarian cancer.

Effects of dose, source, and exposure duration

Dose: No dose effect was observed for ALA, DHA, or EPA consumption (Table 3.15).

Source: The effects of source were not specifically assessed.

Exposure duration: This study assessed dietary intake at four time points. Analyses that excluded cases diagnoses during the first 2 and 4 years of follow-up did not differ in their findings from analyses including all cases.

Sustainment of effect. Sustainment of effect was not assessed.

Table 3.16 Relationship between methodologic quality (more...)

Table 3.16 Relationship between methodologic quality and applicability for estimates of effect of omega-3 fatty acid consumption on risk of ovarian cancer

Cohort Author, Year	Applicability	Quality Parameters
		Adjustment for confounders	Blinding	Valid ascertainment, cases	Valid ascertainment, exposure	Withdrawals and dropouts described
Nurses' Health Study	II	Yes	Yes	Yes	Yes	Yes
Bertone, 200248

Quality and applicability. See Table 3.16.