点击显示 收起
1 From the University of Aberdeen, Department of Mental Health, Royal Cornhill Hospital, Aberdeen, United Kingdom (LJW and HCF); the Robert Gordon University, Department of Life Sciences, Aberdeen, United Kingom (KWW); the University of Edinburgh, Department of Geriatric Medicine, Royal Victoria Hospital, Edinburgh, United Kingdom (JMS); and the University of Edinburgh, Department of Psychology, Edinburgh, United Kingdom (IJD)
2 Supported by the Wellcome Trust, the Medical Research Council (UK), the Biotechnology and Biological Sciences Research Council (UK), and The Alzheimer's Research Trust. LJW holds a Wellcome Trust Career Development Award. JMS holds a Health Foundation Leading Practice Through Research Award. IJD is a recipient of a Royal Society of LondonWolfson Research Merit Award. 3 Reprints not available. Address correspondence to LJ Whalley, Department of Mental Health, Clinical Research Center, Royal Cornhill Hospital, Cornhill Road, Aberdeen, AB25 2ZD, United Kingdom. E-mail: l.j.whalley{at}abdn.ac.uk.
ABSTRACT
Background: Food supplement use is widely promoted, but little is known about the cognitive effects of food supplements.
Objective: We examined the effects of food supplement use on cognitive aging.
Design: This was an observational study of subjects born in 1936 whose mental ability was tested in 1947 and who were followed up in 20002001, at which time cognition, diet, food supplement use, and risk factors for vascular disease were assessed. In a nested case-control study, fish-oil users were matched with nonusers, and cognitive function was related to erythrocyte n3 fatty acid composition.
Results: Childhood intelligence quotient (IQ) did not differ significantly by category of food supplement use (ie, none, fish oil, vitamins, and other). At the age of 64 y, cognitive function was higher in food supplement users than in nonusers before adjustment for childhood IQ. After adjustment for childhood IQ, digit symbol (mental speed) test scores were higher in food supplement users. Fish-oil supplement users consumed more vitamin C and vegetable and cereal fiber than did non-supplement-users. In a nested case-control study, erythrocyte membrane n3 content was higher in fish-oil supplement users than in nonusers, but cognitive function did not differ significantly between groups. Total erythrocyte n3 fatty acids and the ratio of docosahexaenoic acid to arachidonic acid was associated with better cognitive function in late life before and after adjustment for childhood IQ.
Conclusions: Food supplement use and erythrocyte n3 content are associated with better cognitive aging. If associations with n3 content are causal, optimization of n3 and n6 fatty acid intakes could improve retention of cognitive function in old age.
Key Words: Childhood intelligence aging cognition food supplement diet erythrocyte n3 polyunsaturated fatty acids
INTRODUCTION
Mental performance in later life is affected by environmental factors, some of which may prove relevant to dementia. Nutrient supplements of micronutrients [eg, antioxidant vitamins (1)] or macronutrients [eg, n3 polyunsaturated fatty acids (PUFAs) (2)] are widely accepted personal choices intended to improve diets and contribute to the maintenance of good health across the life span. Assessments of the cognitive benefits of food supplement use have been limited by a lack of information on the characteristics of food supplement users and the paradox that those who seem more healthy also seem more likely to use supplements (3), including fish oil (4). The suggested association, for example, between better retention of cognitive function in old age and supplement use could be explained by better lifelong cognitive function informing health choices in late life. Few studies of the cognitive benefits of food supplement use have adjusted the results for the possible contribution of mental ability from earlier in life, largely because this information is not available and reliance is instead placed on years of formal education as a proxy measure of early mental ability.
The mechanisms of food supplement action are poorly understood. There are strong grounds for the belief that supplementation with antioxidant vitamins, for example, can have direct effects on central neuronal metabolism (5). Similar views are held about the role of oxidative stress in atherosclerosis (6). These findings suggest that the use of food supplements may have direct beneficial effects on brain function and indirect effects through slowing of progressive cerebral atherosclerosis.
The Scottish Council for Research in Education's national surveys of the mental ability of Scottish schoolchildren are the only cognitive studies with near-complete ascertainment of an entire year-of-birth cohort (7). These surveys of children born in 1921 and 1936 were conducted on 1 June 1932 and 4 June 1947, respectively, and provide valid childhood mental ability data for tens of thousands of persons now in old age. Here, we compare cognition at the age of 64 y with mental ability test scores from the same subjects at the age of 11 y.
The erythrocyte content of n3 PUFAs is known to increase after nutrient supplementation with fish oils (8) and is a reliable marker for total fish-oil consumption. Our aims, therefore, were 2-fold: to investigate the relations between cognitive aging and the use of food supplements and to examine the associations between erythrocyte n3 PUFA content and cognitive aging in subjects who did or did not use fish-oil supplements.
SUBJECTS AND METHODS
Study setting and design
This was a follow-up, observational study of a cohort living independently in the community, born in 1936 and at school in Aberdeen on 4 June 1947 when their cognitive ability (intelligence quotient, or IQ) was measured and the results archived (7). A nested case-control design was also used to examine associations between erythrocyte membrane n3 fatty acids and cognitive aging. All study procedures followed were in accordance with the ethical standards of the Grampian Research Ethics Committee, who approved the study protocol.
Sampling
Archived IQ scores were made available by the Scottish Council for Research in Education. With the consent of the Grampian Ethics of Research Committee, in 19992001 we matched the 2617 children who took part in the Scottish Mental Survey in Aberdeen City with a local health register (99% coverage of the local population) and identified 957 local men and women who could be exactly matched by birth date and birth name. Of this eligible sample of 957 subjects, 567 were randomly selected and invited by letter, of whom 423 (75%) men and women volunteered for a prospective longitudinal study of brain aging and health. Two subjects were excluded because they scored <24 points on the Mini-Mental State Examination (MMSE) and met the clinical criteria for dementia. Of the remaining 421 subjects, the present study describes 350 (83%) who completed all assessments of cognitive function, dietary habit, exposure to vascular risk factors, physical examination, and blood sampling; other subjects (n = 71) completed some, but not all, assessments. The study sample comprised 171 women and 179 men.
To detect cognitive benefits of fish-oil supplement use, we measured erythrocyte membrane fatty acid content in a nested case-control comparison of 60 (of 72) fish-oil users with 60 non-supplement-users matched by sex and IQ score at age 11 y (within 5 IQ points, or one-third of an SD). Power calculations determined that this would provide 80% power to detect a difference of 0.5 SD in IQ between groups ( = 0.05, two-tailed). Fish-oil supplement users reported nonseasonal use of fish-oil supplements most days for more than the preceding 6 mo, and the other group reported no use of food supplements. Food supplement use was categorized as follows: 1) none, 2) fish oil with or without antioxidant vitamins, 3) vitamins but no fish oil, and 4) other. Fish-oil supplements available in the United Kingdom contain antioxidant vitamins (A, D, and E) to retard peroxidation; some subjects consumed additional antioxidant vitamins. Vitamin supplement users most often took multivitamin preparations providing 100% of the recommended daily allowance of up to 11 vitamins.
Examinations
Research procedures were completed during one visit (lasting 3 h) to the Clinical Research Center. Written informed consent was obtained by a trained research nurse at the interview. By using data obtained at this interview, usual occupations were categorized by job title (9). Usual consumption of alcohol was recorded on a standardized questionnaire as units of alcohol (equivalent to a glass of wine) consumed in a typical week. Smoking history was classified as never smoked, former smoker, or current smoker. History of hypertension and blood pressure (both sitting and standing) on each of 3 occasions at intervals of 5 min were recorded. Hypertension was identified by a mean systolic pressure >144 mm Hg, a mean diastolic pressure >94 mm Hg, or both. Subjects with treated hypertension were included in the hypertensive group. Blood samples were collected after blood pressure was measured. Height and weight were measured on the same apparatus for all subjects.
Childhood intelligence at age 11 y
On 4 June 1947, The Scottish Council for Research in Education surveyed the mental ability of almost all children in Scottish schools who were born in 1936 (7). All children took a version of the Moray House Test no. 12, which is an omnibus group-administered mental test containing 71 numbered questions (75 items in total) with a maximum raw score of 76. The test was administered in school classrooms across Scotland after a schoolteacher read out a standard set of instructions. Before the test proper, 8 practice items were set. The results of this national survey were retained in the archives of the Scottish Council for Research in Education. In 1998 the Scottish Council for Research in Education gave access to their archived mental ability records (10). These detail birth name, school, and results of the Moray House Test of childhood mental ability.
Cognition at age 64 y
The MMSE was used to detect dementia (11). Cognitive tests covered a comprehensive range of cognitive functions. Raven's Standard Progressive Matrices (12) was used to measure nonverbal reasoning. The Rey Auditory Verbal Learning Test (13) was used to test verbal memory. The uses of common objects test (14) was used as a test of executive function or purposive action. The digit symbol subtest of the Wechsler Adult Intelligence Scale-revised (15) was used as an indicator of speed of information processing and as a test of psychomotor performance. The block design subtest of the Wechsler Adult Intelligence Scale-revised (15) was used to examine constructional ability.
Dietary assessments
A food-frequency questionnaire (16) was self-completed and returned by post. These data were analyzed at the Rowett Research Institute by using computer-based algorithms based on the known nutrient content of common UK foods. Use of food supplements was recorded at the interview and was defined as daily use, which was not seasonal. Specific prompts were made to distinguish between food supplements and use of herbal or homeopathic remedies. A small number subjects used only herbal or homeopathic preparations and were classified as non-food-supplement-users.
Fatty acid determination
The fatty acid content of erythrocyte membranes was measured within 1224 mo of blood sampling after the erythrocytes had been separated from whole blood by centrifugation and stored at 70 °C. Total lipids were extracted from the erythrocyte membranes according to the method of Bligh and Dyer (17) as previously described (18), and methyl esters of the component fatty acids were prepared by trans-esterification of a portion of the extracted total lipid in methanolicHCl as described previously (19). Methyl esters were separated in a Hewlett-Packard (Palo Alto, CA) 5890 gas chromatograph by using a CP Sil 88, 50-m wall-coated glass capillary column with split-splitless injection and flame ionization detection. The chromatographic temperature program was initially 80 °C and then rose by 10 °C/min up to 180 °C followed by 2 °C/min to 220 °C; the total run time was 40 min. Peak areas were calculated on a Hewlett-Packard integrator as relative percentage concentrations. Carbon-19 chain length fatty acid was used as an internal standard. Values are given as means and SDs for total saturated fatty acids, total n9 PUFAs, total n6 PUFAs, total n3 PUFAs, and the specific PUFAs cis-linoleic acid, arachidonic acid (AA), eicosapentaenoic acid (EPA), docosapentaenoic acid, and docosahexaenoic acid (DHA).
Statistical analysis
The statistical analysis was performed by using SPSS version 11 (SPSS Inc, Chicago). Scores from the Moray House Test in 1947 were transformed into conventional-type IQ test results (
RESULTS
Description of the samples
As shown in Table 1, supplement use differed by sex, with women being more likely to use supplements than were men (chi-square = 13.2, df = 3, P < 0.005), but did not differ by occupational category; smoking history; alcohol use; presence of hypertension, heart disease, or diabetes; use of psychotropic agents; or use of statins. There were no significant differences in blood pressure, years of education, body mass index (BMI), forced expiratory volume, or the plasma lipid profile by type of supplement use except that plasma triacylglycerols were higher in vitamin users than in users of other supplements (P < 0.05, analysis of variance; Table 2).
View this table:
TABLE 1. Demographic and vascular health characteristics of 350 community residents born in 1936 and examined at the age of 64 y in 19992001 according to use of food supplements
View this table:
TABLE 2. Education and vascular health characteristics of 350 community residents born in 1936 and examined at the age of 64 y in 19992001 according to use of food supplements1
The food-frequency data were incomplete and differed by category of supplement use (Table 3). Multivariate analysis of variance was significant (Pillai's trace, P = 0.031). Post hoc examination by Scheffe's test showed that those subjects who consumed fish-oil supplements also consumed more vitamin C (P < 0.01) and more vegetable fiber (P < 0.05) than did those who did not use food supplements. When consumption of carbohydrates, protein, fats, and sugars was recalculated as percentages of daily energy intake, there were no significant differences between categories of food supplement use.
View this table:
TABLE 3. Consumption of foodstuffs by 350 community residents born in 1936 according to use of food supplements1
Cognitive function by type of food supplement use is shown in Table 4. In the original sample of 421 subjects, data were missing for the Auditory Verbal Learning Test (18 subjects), block design subtest (34), digit symbol subtest (33), and uses of objects test (38). Missing data were not randomly distributed; those with missing data were more likely to have a lower childhood IQ and to score less well on the cognitive tests they completed. For 350 subjects with complete cognitive data, childhood IQ did not differ significantly between groups. MMSE scores were in the nondemented range (2430, median: 29 points); 2 subjects scored 24, 10 subjects scored 25, and 15 subjects scored 26 points. Other subjects scored 2730 points.
View this table:
TABLE 4. Cognitive test scores, according to use of food supplements in 266 community residents who were both cognitively tested at the age of 64 y and intelligence quotient (IQ) tested in 1947 at the age of 11 y1
Multivariate analysis of variance by food supplement category and sex showed an overall difference by category of supplement use in cognitive test results at age 64 y (Pillai's trace, P < 0.05). There was also an overall difference by sex (Pillai's trace, P < 0.001) but no significant interaction between supplement use and sex (Pillai's trace, NS). Estimated marginal means were significantly higher in fish-oil, vitamin, and other supplement users than in nonusers for IQ at age 64 y (P < 0.01) and the digit symbol subtest (P < 0.01). When cognitive scores were reexamined with childhood IQ as a covariate, there was an overall difference between categories of supplement use (Pillai's trace, P < 0.02). Estimated marginal means were significantly higher in fish-oil, vitamin, and other supplement users than in non-supplement-users for the digit symbol subtest (P < 0.02).
Nested case-control comparison between fish-oil supplement users and food supplement nonusers
The 60 matched fish-oil users and 60 nonusers did not differ significantly in exposure to vascular disease risk factors; frequencies of hypertension, smoking, and heart disease did not differ significantly between the groups. Plasma concentrations of cholesterol, total triacylglycerols, HDL cholesterol, and LDL cholesterol also did not differ significantly between the groups. Dietary variables did not differ, except that fish-oil supplement users consumed more sugars (
View this table:
TABLE 5. Scores on cognitive tests in users and nonusers of fish oil supplements1
Erythrocyte membrane n3 fatty acid content in fish-oil supplement users and nonusers is shown in Table 6. n3 Fatty acids were higher (P < 0.01) in fish-oil supplement users than in nonusers, and n6 fatty acids were lower (P < 0.02) in fish-oil users than in nonusers. The ratio of n6 to n3 fatty acids was lower (P < 0.001) and the ratio of DHA to AA ratio was higher (P < 0.001) in fish-oil users than in nonusers.
View this table:
TABLE 6. Erythrocyte membrane fatty acid contents expressed as mean percentages of total membrane fatty acid content in regular fish-oil supplement users and nonusers1
In the combined nested case-control study sample (n = 120), there were significant correlations between erythrocyte n3 fatty acids and risk factors for vascular disease. EPA correlated negatively with BMI (r = 0.226, P < 0.02), HDL cholesterol (r = 0.260, P < 0.005), vitamin C intake (r = 0.243, P < 0.01), and vegetable and cereal fiber intake (r = 0.203, P < 0.05). Total n3 fatty acids correlated negatively with BMI (r = 0.233, P < 0.01) and HDL cholesterol (r = 0.220, P < 0.02),and the ratio of DHA to AA correlated with intakes of vitamin C (r = 0.247, P < 0.01) and vegetable cereal fiber (r = 0.269, P < 0.005).
Relation between cognitive aging and erythrocyte PUFA content
The correlations between childhood IQ, IQ at age 64 y, and log-transformed n3 fatty acids are shown in Table 7 for all subjects (n = 120). In this combined sample, total n3 fatty acids, EPA, DHA, and the ratio of n6 to n3 PUFAs correlated significantly with both childhood IQ and IQ at age 64 y. The ratio of DHA to AA correlated significantly with IQ at age 64 y but not with childhood IQ. Total n3 PUFAs, the ratio of n6 to n3 PUFAs, and the ratio of DHA to AA correlated significantly with block design and Raven's Progressive Matrices scores. Digit symbol scores correlated significantly with DHA and the ratio of DHA to AA. There were no significant differences between correlations estimated separately for the samples of fish-oil users and nonusers. Partial correlations, controlled for childhood IQ, between cognitive scores and log-transformed n3 PUFAs identified 2 significant correlations: between the ratio of DHA to AA and IQ at age 64 y (r = 0.227, P < 0.02) and between the ratio of DHA to AA and the block design subtest (r = 0.250, P < 0.01).
View this table:
TABLE 7. Pearson's correlation coefficients between cognitive test scores and log-transformed erythrocyte polyunsaturated fatty acid (PUFA) content in 120 subjects born in 19361
DISCUSSION
Use of food supplements in late adulthood is associated with cognitive performance. This association does not depend on differences in cognitive ability present in childhood. Specific cognitive advantages at the age of 64 y were found in users of food supplements compared with food supplement nonusers (Table 4). These cognitive advantages of vitamins are consistent with earlier reports that vitamins may reduce dementia risk (2024).
Cognitive advantages of food supplements were found on the digit symbol subtest and these remained after adjustment for IQ at age 11 y. This test is sensitive to cognitive aging and Alzheimer disease: the digit symbol subtest has discriminatory value in dementia (25) and in mild cognitive impairment (26). The digit symbol and block design tests together predict functional capacity in dementia (27).
Data from the Scottish Heart Health study (28) show that vascular risk factors (hypertension, cigarette smoking, and poor diet) are interrelated within the Scottish population. Vitamin C intake explained the difference in cerebrovascular disease incidence between Scotsmen and women (29). The presence of vascular risk factors that predict stroke is strongly associated with a wide range of cognitive deficits around the age of 60 y (30). Lower childhood IQ is also associated with an increased incidence of cardiovascular disease (31). Use of food supplements was not included in these studies but may link their findings. Higher consumption of vegetables and fruit and foods rich in vitamin C was reported in the present study by fish-oil supplement users than by non-supplement-users. We interpret these observations as suggestive of dietary habits associated with consumption of more fruit and vegetables in fish-oil supplement users than in nonusers. These differences confound the distinction between the contribution made to cognitive aging by the use of food supplements and that made by usual dietary habits. Among those who chose to use fish-oil supplements, their greater consumption of fruit and vegetables and more frequent meals of oily fish may reflect heightened awareness of the value of specific foods. The same preferences for fruit and vegetables are less frequent among vitamin supplement users whose higher plasma triacylglycerols suggest a poorer diet to which extra vitamins have been added.
We asked whether exposure to risk factors for vascular disease differed between categories of food supplement use and whether these differences were associated with cognitive advantages. Except for a higher mean plasma triacylglycerol concentration in vitamin supplement users, however, there were no significant differences in exposure to established risk factors for vascular disease. Thus, such differences do not explain differences in cognitive function.
The inclusion of selected biomarkers can help avoid total reliance on self-reported food frequencies. Previously, in an older population-based sample (22), we included fasting measures of plasma vitamins in a study of food supplement use. We found that users of vitamin supplements had higher plasma vitamin and lower plasma homocysteine concentrations, consistent with regular vitamin use. Here, we examined erythrocyte membrane fatty acid content as a marker of fish-oil consumption over previous weeks. Erythrocyte membrane total n3 PUFA content was greater in users of fish-oil supplements, but this group did not perform better than fish-oil nonusers on cognitive tests. We interpret this observation as evidence that self-report of use of fish-oil supplements is a less informative indicator of fish-oil consumption than are measures of erythrocyte n3 PUFA content. When the samples of fish-oil users and nonusers were examined separately, there were significant correlations between childhood IQ and erythrocyte n3 PUFAs, specifically EPA. Higher childhood IQ is probably associated with higher fish-oil consumption (including fish-oil supplements) in later adulthood. In the total sample, there was a weak but significant inverse relation between childhood IQ and the ratio of n6 to n3 PUFAs. Higher childhood IQ appears linked in later adulthood with greater consumption of n3 PUFAs and lower n6 PUFAs.
IQ at age 64 y was significantly correlated in the total sample with erythrocyte membrane n3 PUFA content and with the ratios of DHA to AA and of n6 to n3 fatty acids. These data are consistent with reports that blood and erythrocyte total n3 PUFA content is lower in Alzheimer disease (32, 33) and in age-related cognitive decline (34). Dietary surveys [except for a report by Laurin et al (35)] suggest that habitual consumption of marine oils is associated with higher cognitive function in late life. For example, a cross-sectional population-based study (36) showed that dietary n3 PUFA consumption was inversely related to cognitive performance in subjects somewhat younger than the present sample (aged 4570 y). Like some (37) earlier reports, the authors of this last study (36) sought to control for bias toward better (ie, "healthier") dietary choices by better educated, higher-earning subjects by adjusting for many possible confounders and excluding subjects taking cholesterol-lowering drugs. The sample size and methods used in the present study differed from these approaches but yielded essentially the same result: greater dietary fish-oil intake is related to higher cognitive function in late adulthood.
As in all observational studies, however, the direction remains uncertain of associations between cognitive scores in late life, self-reported supplement use, and blood or erythrocyte n3 PUFAs. Subjects of higher current or original mental ability may be better informed about healthy dietary habits than are subjects of lower ability and may more often choose to include food supplements in their diet. Likewise, higher cognitive performance in late life is associated with more years of education and higher occupational status. These variables could ensure higher income and make affordable a healthier diet and the regular use of food supplements (3, 22, 38).
There are well-established benefits of fish-oil consumption for vascular health (39), and these advantages may explain better cognitive performance in those with higher erythrocyte n3 PUFA content. Cerebrovascular disease is associated with cognitive decline and progression to dementia (40), and this association could be weakened if fish-oil consumption effectively slowed or even prevented progressive age-related cerebrovascular pathology (41).
We interpret these data as supporting the role of n3 fatty acids, especially the ratio of DHA to AA, independent of childhood IQ, in the retention of cognitive function in late life. This role may contribute to the maintenance of healthy neuronal membrane function in the aging brain (42) through maintaining the balance between the production of proinflammatory eicosanoids (derived from n6 precursors) and antioxidative eicosanoids that are also derived from n3 precursors (42). Equally plausible is that the ratio of DHA to AA indicates an overall healthy diet containing an appropriate balance between n6 and n3 PUFAs (43).
Food supplement use contributes to the nutrient intake of a large and increasing proportion of the elderly population. The main nutritional consequence is that a significant proportion of that population now has high, sometimes exceptionally so, intake values. Constant vigilance is required to maintain an optimum balance between possible benefits and the potential for harm of high consumption of specific nutrients. With this caveat in mind, it is noteworthy that among this sample of food supplement nonusers, there were many who were in good health, with no evidence of cognitive impairment and who derived sufficient n3 PUFAs from their usual diet.
ACKNOWLEDGMENTS
We thank Charlotte Appleton for her assistance in the preparation of this manuscript and the staff at the Rowett Research Institute, Aberdeen, for their assistance in the analysis of the food-frequency questionnaires.
All authors were involved in planning the study design. Overall study direction was by LJW, data collection was by HCF, and cognitive testing was supervised by IJD. Erythrocyte fatty acid content was determined by KWW. The data analysis was performed by LJW, as discussed with JMS and IJD, who also wrote the first draft of the manuscript. All authors were involved in revisions and final approval. None of the authors had any financial or personal interest in the organizations that sponsored the research.
REFERENCES