点击显示 收起
1 From the Department of Public Health and Primary Care and the Clinical Gerontology Unit (KTK), University of Cambridge, Cambridge, United Kingdom (AAW and JI); the Medical Research Council Dunn Human Nutrition Unit, Cambridge, United Kingdom (SAB); the Johns Hopkins Bloomberg School of Public Health, Deptartment of Environmental Health Sciences, Baltimore, MD (MDF); the MRC Epidemiology Unit, Elsie Widdowson Laboratory, Cambridge, United Kingdom (NJW); and the IARC, Lyon, France (ER)
2 EPIC-Norfolk is supported by program grants from the Medical Research Council UK and Cancer Research UK and through additional support from the European Union, Stroke Association, British Heart Foundation, the Department of Health, the Food Standards Agency, and the Wellcome Trust. 3 Address reprint requests and correspondence to A Welch, Department of Public Health and Primary Care, University of Cambridge, Strangeways Site, Wort’s Causeway, Cambridge, CB1 8RN, United Kingdom. E-mail: ailsa.welch{at}phpc.cam.ac.uk.
ABSTRACT
Background: The n–3 polyunsaturated fatty acids (n–3 PUFAs) docosahexaenoic acid and eicosapentaenoic acid, found in fish and fish-oil supplements and also formed by conversion of -linolenic acid in soy and rapeseed (canola) oils, are thought to have cardioprotective effects.
Objective: Because the relative feasibility and measurement error of dietary methods varies, this study compared fish and fish-oil intakes obtained from 4 dietary methods with plasma n–3 PUFAs in men and women in a general population.
Design: The study participants were 4949 men and women aged 40–79 y from the European Prospective Investigation into Cancer–Norfolk United Kingdom cohort. Measurements of plasma phospholipid n–3 PUFA concentrations and fish intakes were made with the use of 4 dietary methods (food-frequency questionnaire, health and lifestyle questionnaire, 7-d diary, and first-day recall from the 7-d diary).
Results: Amounts of fish consumed and relations with plasma phospholipid n–3 PUFAs were not substantially different between the 4 dietary methods. Plasma n–3 PUFA concentrations were significantly higher in women than in men, were 20% higher in fish-oil consumers than in non-fish-oil consumers, and were twice as high in fatty fish consumers as in total fish consumers. Only 25% of the variation in plasma n–3 PUFA was explained by fish and fish-oil consumption.
Conclusions: This large study found no substantial differences between dietary methods and observed clear sex differences in plasma n–3 PUFAs. Because variation in n–3 PUFA was only partially determined by fish and fish-oil consumption, this could explain the inconsistent results of observational and intervention studies on coronary artery disease protection.
Key Words: n–3 Polyunsaturated fatty acid • fish • fish oils • diet methods • 7-d diary
INTRODUCTION
Observational studies and intervention trials have suggested that n–3 polyunsaturated fatty acids (n–3 PUFAs) are protective against coronary artery disease (CAD), but more recent studies have provided conflicting evidence (1-8). The n–3 PUFAs docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) are found in fish and fish-oil supplements and are also formed from the conversion of -linolenic (ALA) acid in walnuts, soy, rapeseed (canola), and flaxseed oils (9). Some ALA is also found in fatty fish (10). Although observational and intervention studies have related dietary intake of n–3 PUFAs to measures in plasma, serum, red blood cells, and adipose tissue, there have been few large population studies in general populations with moderate fish intake (11-20).
Many dietary methods exist for estimating intake of foods, ranging from intensive record methods, such as weighed records, 7-d diaries, or 24-h recalls, to structured questionnaires such as food frequency questionnaires (FFQs) or simpler summary questions. Because the relative feasibility, performance, cost, and extent of measurement error of dietary methods varies, the aims of the present study were to compare fish intakes obtained from 4 dietary methods and fish-oil consumption obtained with a health and lifestyle questionnaire (HLQ) with plasma phospholipid n–3 PUFA, as a biomarker of fish intake, and to investigate sex differences in plasma phospholipid n–3 PUFAs.
SUBJECTS AND METHODS
Study population
The participants were part of the European Prospective Investigation into Cancer (EPIC), a nine-country collaboration involving approximately one-half million persons. In Norfolk, United Kingdom, 25 000 men and women aged 40–79 y were recruited for the baseline survey in 1993–1997 (EPIC-Norfolk) as part of this collaboration and a health check was performed (21). For the present study, a subsample of 8000 men and women were selected for analysis of plasma fatty acids, and 4900 men and women with data available from 4 dietary methods were chosen for these analyses. The Norfolk Health District ethics committee granted ethical permission for the study.
Data collection
All participants were asked to complete a self-administered detailed health and lifestyle questionnaire and then attended a health check where measurements and blood samples were obtained by trained nursing staff.
Anthropometric measures
Height and weight were measured to the nearest 0.1 cm and 0.2 kg, respectively, with participants wearing light clothing without shoes at the health check (21). Body mass index (BMI) was calculated as weight (in kg)/height2 (in m).
Blood samples
Blood samples were taken by venepuncture into tubes containing citrate buffer at the health check. After overnight storage in the dark at 4–7 °C, the samples were spun in a centrifuge at 2100 x g for 15 min at 4 °C. Aliquots (450 µL) of plasma were transferred to plastic straws and stored in liquid nitrogen. After addition of 100 µg butylated hydroxytoluene and 20 µg 1,2-dipalmitoyl-D62-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Alabaster, AL) internal standard to 200 µL thawed plasma, lipids were extracted with chloroform-methanol. Plasma phospholipids were isolated by solid-phase extraction chromatography (LC-Si; Supelco, St Louis, MO USA), and fatty acid methyl esters were formed by transmethylation of the phospholipids with METH-Prep (Alltech, Deerfield, IL). Analyses were carried out on an HP 5980 gas chromatograph (Agilent, Palo Alto, CA) equipped with a flame ionization detector. Fatty acid concentrations were calculated in relation to the methyl palmitate-D31 internal standard peak, well-separated chromatographically from methyl palmitate-D0. Quality control was undertaken by using reproducibility analysis of a preprepared batch of plasma samples that was analyzed with each batch of 40 samples. The CVs for these samples for ALA, EPA, DPA, and DHA for concentrations measured in µmol/L were 12%, 11%, 18%, and 15%, respectively, and for concentrations measured in mol% were 8%, 4%, 7%, and 7%. The concentration of each phospholipid fatty acid was expressed as either a concentration (µmol/L plasma) or as a mol% of the 27 fatty acid concentrations measured.
Dietary assessment
Four dietary methods (2 record and 2 frequency methods) were used to estimate intake of total fish, white fish, and fatty fish. White fish included cod, haddock, and white fish products and dishes. Fatty fish included salmon, mackerel, sardines, and trout. The methods were given to the participants in the following order: HLQ, FFQ, first-day recall, and 7-d diary.
Frequency methods
A general HLQ included a question on how often individuals ate particular foods. Fish intake was derived from responses to 2 questions, "How often do you eat the following foods: fatty fish (eg, herring, sprats, pilchards, mackerel) and other fish (eg, cod, tuna, haddock)." Frequencies were converted to amounts eaten by multiplying the responses (never, seldom, once a week, 2–4 times/wk, 5–6 times/wk, once or more daily, and don’t know) by portion sizes of 114 g for white fish and 116 g for fatty fish. Portion sizes were the same as for the FFQ (22). All participants were requested to complete an HLQ, but, because some did not complete both questions, these persons were omitted from the analyses (n = 755).
Data from the 131-item FFQ were obtained by using frequency response for consumption during the previous year to 6 questions: "Fried fish in batter, as in fish and chips," "Fish fingers, fish cakes," "Other white fish, fresh or frozen, eg, cod, haddock, plaice, sole, halibut," "Oily fish, fresh or canned, eg, mackerel, kippers, tuna, salmon, sardines, herring," "Shellfish, eg, crab, prawns, mussels," and "Fish roe, taramasalata" (22). Frequencies ranged from "Never or less than once a month" to "6 or more times a day" and were estimated in relation to average portion sizes or standard units of consumption. The FFQ was a modified version of an FFQ widely used in the United States, and portion sizes were derived from the UK population data and weighed records in study participants aged 40–74 y (20). The FFQ was completed before the health check and checked for completeness of response by nurses at the health check.
Missing responses to questions from the HLQ and FFQ methods were treated as missing data. Data from the FFQ were used after truncation of the top and bottom 0.5% of the ratio of energy intake to estimated basal metabolic rate (BMR) (22).
Record methods
The 7-d food diary with estimated weights of food consumed consisted of an A5-sized booklet with 17 sets of color photographs representing small, medium, or large portions and instructions to guide the detail and type of information to be reported (23). Household measures and standard units were also used to describe amounts of foods consumed. Nurses, trained to standardized protocols, provided instructions on how to complete the 7-d diary and performed an interviewed 24-h recall at the health check of the previous day, from waking to sleeping. This formed the first day of the 7-d diary record (first-day recall). The interviewers were trained in the use of neutral probing questions and did not give dietary advice. The participants were asked to complete the remaining 6 days and then returned the diary to the study center by post. To test for the effect of a single day of intake, the first-day recall was used in analyses.
Data from the diary and first-day recall were entered by using the Data into Nutrients for Epidemiologic Research (DINER) data entry system, which included 699 fish-food list items (479 white fish and 172 fatty fish) (23). Data entry staff received 2 days of training in use of the system, and their work was checked for 3 mo until it was considered satisfactory. A series of checks on entry were performed before data analysis.
Data for intake of total fish, fatty fish, white fish, molluscs, crustacean, and roe were derived by using the Foods Out of Diner System (FOODS) system for creating foods from record dietary data (24). Fish consumed as part of mixed dishes was included in the calculated fish intake. No outliers were removed from the diary data. Because data entry of the diaries takes longer and is more complex than the other methods, the number of persons available for analysis was limited by the availability of entered diaries.
Total fish, white fish, and fatty fish intakes were available from all methods. Because the reported amounts of molluscs, crustacean, and roe were small with the FFQ, 7-d diary, and first-day recall, they were combined with fatty fish for these analyses.
For each dietary method, the percentage of fish consumers was calculated as the number of persons who consumed >0 g fish/d. Validation of the 7-d diary and FFQ for nitrogen, energy, sodium, potassium, and vitamin C has been reported elsewhere (25).
Fish-oil supplement consumption
Consumption of cod liver oil and other fish-oil containing supplements was ascertained by response to the question, "Please list any vitamins, minerals or other food supplements taken regularly during the past year (eg, vitamin C, vitamin D, calcium, fish oils, primrose oil, beta carotene etc) (List brand and daily dose, if known)," and smoking habit was ascertained by response to the question, "Do you smoke cigarettes now?" These questions were part of the HLQ questionnaire. Fish and cod liver oils of all types are referred to throughout this article as "fish oils."
Statistical analyses
Statistical analyses were performed with STATA statistical software version 8.0 (Stata Corp, College Station, TX). Analyses were performed for the whole group and stratified by sex. Means (±SDs) of dietary intakes and plasma n–3 PUFAs were calculated by sex and category of dietary intakes. CVs were calculated for fish intake derived from the different methods. Spearman correlation coefficients were calculated to estimate the relation between plasma fatty acids and the 4 dietary methods.
The whole population was divided into 4 categories of fish intake according to estimates using the different dietary assessment methods. Category 1 comprised those reporting zero intake; those with intakes greater than zero were divided into tertiles (categories 2, 3, and 4). When the HLQ was divided into white and fatty fish, no persons were grouped into category 3. This was because 79% of responses for white fish and 89% of responses for fatty fish resulted in one value (representing the categories "never," "seldom," or "once a week") and so could not be split into 2 parts. Also, because 21% of responses for white fish and 11% of responses for fatty fish represented the remaining categories ("2–4 times a week" or "once a day") and were at the upper end of the distribution, these values also could not be divided. When values for fatty fish and white fish were combined to create the variable total fish, the number of values within the range of distribution of responses was greater and therefore creation of tertiles did not result in a missing category.
Adjusted mean values of total plasma phospholipid n–3 PUFA, EPA, and DHA were calculated by category of intake of total fish, white fish, and fatty fish for each dietary method by using analysis of variance (ANOVA) and the post–estimation command adjust. Two models for adjustment of mean plasma phospholipid fatty acids for covariates were used: 1) by age and 2) by age, weight, height, smoking status, and cod liver oil consumption. Because data from models 1 and 2 were not significantly different, only data from model 2 is presented. Further regression analyses were performed on the whole group using data from the FFQ to quantify the differences in plasma phospholipid fatty acid concentrations according to intake of total fish, fatty fish, and fish-oil consumption and that relating to sex, age, height, and smoking status (models 3 and 4).
To determine whether associations of estimated fish intake with plasma phospholipid fatty acid concentrations could be improved by combining the results of different methods, estimates of fish consumption from the 7d-diary and FFQ, the 7-d diary and HLQ, the FFQ and the HLQ, the first-day recall and the FFQ, and the first-day recall and HLQ were averaged and the analyses repeated. The first-day recall and 7-d diary were not combined because the first-day recall was a component of the 7-d diary.
RESULTS
The men in the sample were slightly older than were the women (Table 1). More women than men consumed fish-oil supplements (32.3% compared with 29.0%; P = 0.001).
View this table:
TABLE 1. Characteristics and n–3 plasma fatty acid profiles in 4949 men and women aged 39–78 y
Plasma phospholipid fatty acids
Of the n–3 fatty acids, DHA was found in the highest concentration in plasma phospholipids and ALA in the lowest (Table 1). The ranking of concentration of fatty acids did not differ significantly whether expressed as absolute or percentage amounts. The concentrations of plasma fatty acids in the women were greater than those in the men, with the exception of DPA expressed as a percentage, which was greater in the men than in the women (Table 1).
Dietary intake
Mean intake of fish in this population estimated by all dietary methods was moderate, with an average of approximately one-third of a portion of total fish and one-tenth of a portion of fatty fish per day (one portion = 120 g). This population consumed more white than fatty fish accounting for between 55% and 68% of total fish intake; the variation was dependent on dietary method (Table 2).
View this table:
TABLE 2. Range of fish intake estimated by 4 dietary methods in 4949 men and women aged 39–78 y classified according to category (C) of fish intake1
The number of persons classified as nonconsumers of total fish was greatest for the first-day recall (77%) and the least for the FFQ (4.4%), which was reflected in the CV (Table 2). Compared with the FFQ, the number of persons classified differently as fish or nonfish consumers by the other dietary methods was 11.2% with the HLQ, 15.1% with the 7-d diary, and 28.1% with the first-day recall.
Correlation analyses
Intercorrelations between the plasma phospholipid fatty acids are shown Table 3. Correlations between plasma fatty acids and fish intake from the dietary methods did not differ substantially (none was >0.27).
View this table:
TABLE 3. Spearman correlation coefficients between individual plasma n–3 fatty acids and total fish, white fish, and fatty fish intakes estimated with 4 different dietary methods in 4949 men and women aged 39–78 y1
Plasma fatty acids by quartile of fish intake
A significant positive relation existed between intake of total fish and total plasma phospholipid n–3 PUFA with all dietary methods in both sexes (Table 4). Relations remained significant after adjustment for age, weight, height, current smoking status, and fish-oil supplement consumption. The dietary method that predicted the lowest concentrations of plasma n–3 PUFA in nonconsumers (category 1) was the FFQ, whereas the first-day recall and the 7-d diary predicted the highest concentrations for category 4. In men, the largest difference between categories 1 and 4 of total adjusted fish consumption was found for the FFQ (74 µmol/L), with the 3 other dietary methods being similar (58–64 µmol/L; Table 4). In women, the differences were greater: 107 µmol/L for the FFQ, 98 µmol/L for the HLQ, 89 µmol/L for the 7-d diary, and 60 µmol/L for the first-day recall. Similar relations between fish intake and plasma concentrations of EPA and DHA were found (Table 5). Adjusted total plasma phospholipid n–3 PUFA for categories 3 and 4 of total fish intake were significantly different from nonconsumers for all dietary methods (Table 4).
View this table:
TABLE 4. Plasma total n–3 fatty acids calculated according to category (C) of dietary total fish intake estimated by 4 dietary methods in 2597 men and 2352 women1
View this table:
TABLE 5. Plasma docosahexaenoic and eicosapentaenoic acid concentrations calculated with model 2 according to category (C) of dietary total fish intake estimated by 4 dietary methods1
Contribution of fish-oil supplements to plasma phospholipid n–3 PUFA concentrations
The contribution of fish-oil supplements to total plasma phospholipids n–3 PUFA concentrations was estimated in nonfish consumers for the FFQ, which identified the greatest percentage of fish consumers. The adjusted mean in non-fish-oil consumers was calculated from the model; it was 340 ± 6 µmol/L (95% CI: 328, 352 µmol/L) in men who did not consume fish oil and 408 ± 11 µmol/L (95% CI: 387, 428 µmol/L) in men who did consume fish oil, and it was 378 ± 7 µmol/L (95% CI: 365, 391 µmol/L) in women who did consume fish oil and 455 ± 10 µmol/L (95% CI: 435, 474 µmol/L) in women who did not consume fish oil.
Relation of total plasma phospholipid n–3 PUFA and type of fish
The relation between plasma phospholipid n–3 PUFA and fatty fish was stronger than that for white fish (Figure 1).
FIGURE 1.. Mean (±SE) total n–3 plasma phospholipid fatty acids (PUFA) according to white fish and fatty fish intakes estimated by 4 dietary methods in men and women (adjusted for age, weight, height, smoking status, and fish-oil consumption with postestimation after ANOVA). Total n–3 fatty acids are the sum of eicosapentaenoic acid, docosapentaenoic acid, and docosahexaenoic acid. n = 2597 men for the 7-d diary, food-frequency questionnaire (FFQ), and first-day recall and 2184 men for the health and lifestyle questionnaire (HLQ); n = 2352 women for the 7-d diary, FFQ, and first-day recall and 2010 women for the HLQ. Bars are categories (C) 1 to 4 of intake: C1 is no consumption of fish, C2-C4 are subsequent tertiles of intake. No persons were grouped into category 3 for the HLQ (see Methods). Interactions between sex and type of fish (within each dietary method) were significant, P < 0.001. P for trend (ANOVA) for fatty fish was <0.001 for the 7-d diary, FFQ, and HLQ and = 0.001 for the men and 0.32 for the women for first-day recall. P for trend for white fish was 0.02 with the 7-d diary, 0.93 with the FFQ, 0.06 with the HLQ, and 0.08 with the first-day recall for the men; for women, the respective P values were <0.001, 0.012, 0.001, and 0.17.
Estimates of plasma phospholipid n–3 PUFA and fish consumption by combined dietary methods
The results for combined dietary methods were not significantly different from those of the individual methods. The greatest discrimination between categories 1 and 4 of estimated intake (adjusted for covariates) was found with the combination of the 7-d diary and the FFQ [mean (±SE) total n–3 PUFA for category 1: 338 ± 20 µmol/L (95% CI: 299, 378 µmol/L) in the men and 320 ± 22 µmol/L (95% CI: 278, 362 µmol/L) in the women; mean (±SE) total n–3 PUFA for category 4: 417 ± 5 µmol/L (95% CI: 404, 425 µmol/L) in the men and 465 ± 6 µmol/L (95% CI: 452, 477 µmol/L) in the women; P for trend < 0.001 for both men and women].
Linear regression coefficients
The relation of total fish with total plasma phospholipid n–3 PUFA was 113 µmol/L per portion of fish (model 3; Table 6). The women had higher concentrations of plasma phospholipid fatty acids than did the men (44 µmol/L), which was equivalent to consumption of about one-half portion of fish. The increment in plasma fatty acids for fatty fish was twice that for total fish. EPA accounted for a one-fourth of the increment in total n–3 PUFA with total fish and DHA accounted for three-fourths (model 3; Table 6). Only fish intake, fish-oil consumption, sex, and age were significant when the model was adjusted for age, height, weight, and current smoking status (model 4, Table 6). The effect of fish-oil supplements on plasma n–3 PUFA amounted to the equivalent of about two-thirds of a portion of total fish or about one-third of a portion of fatty fish (model 4, Table 6).
View this table:
TABLE 6. Linear regression coefficients of plasma n–3 PUFA with total fish and fatty fish intakes estimated by using the food-frequency questionnaire in 4949 men and women aged 39–78 y in the European Prospective Investigation into Cancer-Norfolk study cohort, 1993–19971
DISCUSSION
This population consumed moderate amounts of fish, an average of 2 portions of total fish and three-fourths portion of fatty fish per week, with even those persons in the top category consuming only 4 portions of total fish per week. Similar amounts of fish were consumed with all 4 dietary methods. Fish intakes in the present study were of the same scale as another study conducted in the United Kingdom but less than the consumption in Spain, France, Greece, and Scandinavia and more than that in Germany, the Netherlands, and Italy (26, 27). The wide range of fish intakes in our study enabled estimates of plasma n–3 PUFA concentrations both in nonconsumers and consumers, in contrast with previous studies conducted in populations with relatively high fish consumption (12, 28-30).
Overall, the different dietary methods provided positive and significant relations between plasma n–3 PUFA and total fish that were not substantially different and which remained even after adjustment. However, the first-day recall was less consistent in providing relations and was unable to predict habitual consumption because average consumption, which was determined by other methods, was relatively infrequent (twice per week). Previous estimates of fish and n–3 PUFA biomarkers have ranged from frequency methods to weighed records and diet histories, but we are unaware of other studies that used the 7-d food diary method (12, 28, 29, 31-36). The FFQ categorized the fewest persons as nonconsumers, probably because it captures infrequent consumption better. The simplest, least-cost method is the HLQ, although there was a greater proportion of missing data and restricted distributions of responses with this questionnaire. Because the diary method provides more detailed information and previous validation studies found less measurement error with it, this instrument may have been expected to perform better than other methods (25, 37). If the first-day recall had been collected as a random stratified sample, the results may have differed from those obtained in the present study. The timing of blood collection may have favored the diary and first-day record; however, because the associations with the methods were only slightly different from each other, this seems unlikely. Although the percentage of persons classified as fish consumers varied between dietary methods, estimates of mean consumption were not substantially different, probably because fish was eaten more as an individual food than in mixed dishes and, therefore, the frequency was relatively accurately determined. We found no improvement with estimates of fish intake when methods were combined.
The relation of plasma phospholipid n–3 PUFA with fatty fish was twice as strong as with total fish and was 20% higher in fish-oil consumers. Another study found higher correlations with fatty than total fish (28). Although there were analytic differences between our study and others, correlations between plasma phospholipid n–3 PUFA and total fish were of a similar scale, except for one study conducted in Spain, (12, 28, 28-35). However, in 3 studies in which more fish was consumed, blood concentrations of n–3 PUFA were lower than those of our population (29, 31, 35).
Plasma phospholipid fatty acids were significantly higher in women than in men, in fish and nonfish eaters, both in absolute and percentage terms. From multivariate regression, these sex differences were equivalent to consumption of an extra one-half portion of fish in women. We are unaware of other studies in which fish intake and sex differences in n–3 PUFA have been presented; most studies have been performed only in men or the data not analyzed by sex. Four studies found higher tissue concentrations of n–3 PUFA in women than men, but either fish intake was not estimated or fish was not consumed (11, 13, 38, 39). Because the amounts of fish consumed are similar in men and women, the lower circulating concentrations of n–3 PUFAs in men could be due to increased body size and larger plasma volume (40). Alternatively, the greater concentrations in women may be due to differences in estrogen concentrations, because higher concentrations of n–3 PUFA have been found in women than men who consumed no dietary n–3 PUFA (38). Other studies have reported that DHA changed in subjects who received hormone treatment, increasing in male-to-female transsexual subjects and decreasing in female-to-male transsexual subjects (38).
Our results indicate that fish and fish-oil intake explained only 20% of the variation in plasma phospholipid n–3 PUFA in men and 25% in women. Other studies also found relatively high concentrations of tissue n–3 PUFA in nonfish and nonfish-oil consumers, confirming our findings (30, 36, 41-43). Therefore, despite a significant relation between n–3 PUFAs and intake of fish and fish oils, an underlying residual concentration exists that is not accounted for by these sources, and the reasons for this warrant further investigation. Alternatively, other sources of dietary n–3 PUFA may contribute to circulating concentrations. ALA is converted to EPA and DHA, with an intake of 50 g rapeseed oil/d estimated to be the equivalent of a weekly portion of oily fish, and conversion is greater in women of childbearing age than in men (44-49).
Absolute measures of circulating n–3 PUFA have the advantage over percentage calculations because saturated and monounsaturated fatty acids are not included in the denominator (50). Plasma is more accessible than is adipose tissue, and supplementation studies have found concentrations of n–3 PUFA rise within 1 wk, returning to presupplementation concentrations in 28 d (11, 14, 36). However, unlike recovery biomarkers that translate into absolute intake estimates over a known time period, concentration biomarkers such as n–3 PUFA only provide relative estimates of concentrations in tissues
Despite the known associations between fish, fish oils, and biological measures, studies have shown conflicting evidence for the protective effect of fish on CAD, with many intervention trials and observational studies showing clear benefits of increased fish and fish-oil consumption. However, more recent studies and longer term follow-up of early intervention trials have either found a lack of protection or an increased risk for CAD (1-8). It is possible that other factors affect circulating concentrations of n–3 PUFA, which may explain why intervention studies with fish or fish oils have produced inconsistent results with regard to protection from CAD.
We believe this is the largest study to date that compared several dietary methods with biomarkers of n–3 PUFA in a general population of both men and women. Mean estimates of fish consumption from the 4 dietary methods were not substantially different. Clear sex differences in circulating phospholipid n–3 PUFA concentrations were found in this population. Compared with total fish intake, fatty fish intake had twice the effect on plasma phospholipid n–3 PUFAs, and intake of total fish and fish oils raised concentrations even more than did fish intake. Given that circulating concentrations of n–3 PUFA were lower in the men than the women, these data suggest that men need to eat more fish than do women to obtain the same blood concentrations. Nevertheless, even non-fish eaters had measurable underlying n–3 PUFA plasma concentrations, and those with high concentrations of fish and fish-oil consumption have concentrations only 25% higher. If n–3 PUFA are the main mechanism by which fish and fish oil may influence disease risk, the observation that fish and fish-oil consumption does not explain all the plasma variation in n–3 PUFA may explain the inconsistent results of observational and intervention studies with fish and fish oils.
ACKNOWLEDGMENTS
We thank all the participants in this study and the EPIC-Norfolk study staff at the University of Cambridge, Department of Public Health and Primary Care. The authors thank Beatrice Vozar, Sebastian Chanlon, and Thomas Cler (IARC) for their technical assistance in measuring plasma phospholipid fatty acid concentrations.
AW provided the dietary data, performed the statistical analyses, and wrote the manuscript. MF was responsible for the fatty acid analyses. JI contributed to the dietary data. KTK, SB, and NW are the principal investigators of the EPIC-Norfolk Study. All authors were involved in interpreting the data and contributed to the writing of the manuscript. None of the authors had any conflict of interest.
REFERENCES