Sie sind auf Seite 1von 10

Nutritional status and cognitive functioning in a normally

ag ing sam pie: a 6-y reassessment1

Asenath La Rue, Kathleen M Koehler, Sharon J Wayne, Stephen J Chiulli, Kathleen Y Haaland, and
Philip J Garry

Downloaded from by guest on 13 June 2020

ABSTRACT Associations between nutritional status and cog- tests of mental imagery, verbal fluency, and immediate mem-
nitive performance were examined in 137 elderly (aged 66-90 y) ory. However, the direction of associations was variable, with
community residents. Participants were well-educated, adequately negative correlations predominating. Other recent studies have
nourished, and free of significant cognitive impairment. Perfor- reported associations between selected vitamins and single
mance on cognitive tests in 1986 was related to both past (1980) measures of cognitive performance in healthy aged samples (5,
and concurrent (I 986) nutritional status. Several significant asso- 6), and Riggs et al (7) found that plasma concentrations of
ciations (P < 0.05) were observed between cognition and concur- vitamins B- 12 and B-6, folate, and homocysteine were related
rent vitamin status, including better abstraction performance with to visuospatial copying and memory performance in normal
higher biochemical status and dietary intake of thiamine, ribofla- older adults. Considered together, these investigations raise the
yin, niacin, and folate (r = 0. 19-0.29) and better visuospatial possibility that differences in nutritional status too mild to meet
performance with higher plasma ascorbate (r = 0.22). Concurrent criteria for clinical deficiency may have a subtle influence on
dietary protein in 1986 correlated significantly (r = 0.25-0.26) cognitive performance among older nondemented adults. How-
with memory scores, and serum albumin or transferrin with mem- ever, additional studies are clearly needed to establish the
ory, visuospatial, or abstraction scores (r = 0.18-0.22). Higher reliability of these relations and to explore brain-behavior
past intake of vitamins E, A, B-6, and B-l2 was related to better mechanisms that might underlie the associations.
performance on
visuospatia] recall and/or abstraction tests Data for the present analyses are from a longitudinal fol-
(r = 0. 19-0.28).Use of self-selected vitamin supplements was low-up of the sample of healthy older adults studied by Good-
associated with better performance on a difficult visuospatial test win et al (3). Changes in the cognitive test battery precluded a
and an abstraction test. Although associations were relatively weak direct test-retest analysis. Instead, we focused on cognitive
in this well-nourished and cognitively intact sample, the pattern of performance at retest and asked whether such performance was
outcomes suggests some directions for further research on cogni- related to concurrent or past nutritional status. We had three
tion-nutrition associations in aging. Am J Clin Nutr l997;65: aims in conducting these analyses: 1) to assess the stability of
20-9. associations between cognition and nutrition over time, 2) to
assess the generality of nutrition-cognition relations by looking
KEY WORDS Aging, nutrition, and cognition at an expanded set of nutritional and cognitive measures, and 3)
to examine associations between use of vitamins and other
dietary supplements and cognitive performance.

The possibility that nutritional deficiencies may play a role

in cognitive deficits in old age has been recognized for several
years (1, 2). However, among normally aging older adults, Subjects
there is only scant information available about nutrition-
Participants were members of the New Mexico Aging Pro-
cognition associations. The most widely cited study on this
cess Study, a longitudinal study of nutrition and aging that
topic is that of Goodwin et al (3), who reported that healthy,
began in 1979 [see Garry et al (8) for a complete description of
community-residing older adults with comparatively low di-
etary intakes of protein and selected vitamins (vitamin C,
thiamine, riboflavin, folate, niacin, and vitamin B-6) scored I From the Departments of Psychiatry, Pathology, and Neurology, the
University of New Mexico School of Medicine, and the Veterans Affairs
lower on average than their better-nourished peers on tests of
Medical Center, Albuquerque, NM.
verbal memory and nonverbal abstract reasoning. Significant
2 Supported by grants from the United States Public Health Service, NIH
associations were also noted between blood concentrations of
AG02049, and GCRC DRR5MO100977-l3.
certain nutrients (vitamin C, riboflavin, and vitamin B-l2) and
3 Address reprint requests to KM Koehler, Clinical Nutrition Laboratory,
cognitive performance. More recently, on the basis of a small University of New Mexico, 2701 Frontier Place NE, Surge Building, Room
sample of healthy older adults, Tucker et al (4) reported several 215, Albuquerque,NM 87131-5666.
significant correlations between biochemical indexes of iron Received April 1, 1996.
status, protein, vitamins A and C, and carotene, and laboratory Accepted for publication July 26, 1996.

20 Am J Cliii Nutr l997;65:20-9. Printed in USA. © 1997 American Society for Clinical Nutrition

the sample]. Initially, 304 community-residing individuals (138 were measured in the Clinical Nutrition Laboratory. Plasma
men and 166 women) > 60 y of age were enrolled. Participants ascorbic acid was measured by an automated colorimetric
were basically healthy for their age. For example, individuals procedure (12). Riboflavin status was determined as the eryth-
with diabetes, coronary artery disease, and uncontrolled hyper- rocyte glutathione reductase activity coefficient (EGRAC) by
tension were not enrolled in the study. Entrance was not limited using an automated method ( 13). The normal range for the
to any ethnic group, but all volunteers were white, with 3% EGRAC by this method is 1.00-I .35, with a higher activity
being of Hispanic descent. More than 40% had college degrees coefficient indicating poorer riboflavin status. Thiamine status
and incomes were generally above average. In general, panic- was determined as the erythrocyte transketolase activity coef-
ipants were well-educated, health-conscious, financially se- ficient (ETKAC), also with an automated method, and with a
cure, and highly motivated, and as such, were not a random higher activity coefficient indicating poorer vitamin status.
sample of older adults. At their entrance into the study, they Plasma and erythrocyte folate and plasma vitamin B-12 con-
were given physical examinations, dietary and biochemical centrations were measured simultaneously by using a radioas-
tests to characterize their nutritional status, and a brief battery say kit from Becton Dickinson Immunodiagnostics, Orange-

Downloaded from by guest on 13 June 2020

of cognitive tests. This study was approved by the Human burg, NY (14). Serum transfemn was measured as total iron-
Research Review Committee of the University of New Mexico binding capacity simultaneously with measurement of serum
School of Medicine. We obtained informed consent from each iron by an automated colorimetric method that used sulfonated
participant. bathophenathroline as the chromogen (15). The biochemical
In the report by Goodwin et al (3), nutrition-cognition rela- vitamin status of this study population was reported in detail
tions were examined in 260 of the original study participants (12-16).
(1 19 men and 141 women with mean ages of 71.9 and 71 .6 y,
Cognitive assessment
respectively) who had complete nutritional and cognitive data
from the first round of testing and who were free of disorders All cognitive tests were administered individually by trained
that might directly affect metabolism. The present analyses are research technicians in a single session of ‘2 h. The test
based on data for 137 persons (67 men and 70 women) who battery included the following standardized measures that eval-
completed cognitive testing ‘‘6 y after the initial test round and uate cognitive processes commonly affected by aging or age-
who had nutritional findings available from both 1986 and related disorders.
1980. Their mean age at reassessment was 76.9 y (range: 66-90 The Abstraction Scale from the Shipley-Hartford Intelli-
y); 94% had completed high school and 80% had attended gence Test (17) asks the subject to specify missing items in
some college. Consistent with attrition trends in other longitu- several logically determined series (eg, AB BC CD D?). It
dinal studies (9), these 137 subjects were younger and had provides a measure of abstract reasoning.
better memory test performance initially than the subjects who The Logical Memory and Visual Reproduction subtests from
were not reassessed. the Wechsler Memory Scale (WMS; 18) assess verbal and
nonverbal memory, respectively. For the Logical Memory test,
Procedures the subject listens to two short stories and then attempts to
recall them immediately after presentation and after a delay of
Nutrition evaluation 30 mm. For the Visual Reproduction test, the subject studies
Three-day dietary records were collected annually, together each of four geometric designs and then attempts to draw the
with clinical, anthropometric, and biochemical measures of designs from memory, immediately and after a 30-mm delay.
nutritional status. Subjects received instructions on keeping In the Rey-Ostemeth Complex Figure test ( 19, 20), the
accurate diet records, such as inclusion of all foods and bev- subject is first asked to copy a complex geometric design and
erages consumed, detailed descriptions of items recorded, re- is then asked to draw it from memory, immediately after
cording of home-prepared recipes, and correct estimation of copying it and again after a 30-mm delay. The copying task
portion sizes. Food models and standard utensils were used to provides a measure of visuospatial skills whereas the recall
demonstrate portion sizes, and each subject received a diet tasks measure nonverbal learning and memory. The Rey-
scale and instruction booklet. After subjects recorded their Osterrieth test provides a more difficult test of visuospatial skill
dietary intake for three successive weekdays, project staff and nonverbal memory than does the WMS Visual Reproduc-
visited their homes to collect the completed records and review tion test.
them for completeness and clarity. If vitamin or mineral sup-
Statistical analyses
plements were used by the subject, the brand name, contents,
and amounts of each nutrient were recorded to determine total Because scores on the immediate and delayed versions of the
intakes. Records for this analysis were coded and analyzed by memory tasks were highly intercorrelated, only immediate
using the Highland View Hospital-Case Western Reserve Uni- recall scores were included in statistical analyses. Associations
versity (CWRU) Nutrient Database, release 5, 1980, for 1980 between nutritional variables and cognitive performance were
data and release 6, 1982, for 1986 data (10). The dietary intake tested by Spearman rank-order correlations; all pairings of
of 270 participants in 1980, the initial study year, was reported cognitive scores obtained in 1986 with nutritional values from
in detail (10). 1986 and 1980 were examined. Analyses of variance were used
After an overnight fast, 50 mL blood was obtained with to compare mean scores on cognitive measures for participants
heparin from each participant for biochemical measurements. who took vitamin supplements with those who did not take
Serum albumin was measured by the New Mexico Medical supplements. Wilcoxin nonparametric analyses were used to
Reference Laboratory as part of a routine battery of clinical compare supplement user and nonuser groups if cognitive test
chemistry measures ( 1 1 ). Serum transfemn and vitamin status scores were not normally distributed.

Before conducting these analyses, we examined associations This was a greater proportion than for past nutritional status,
between cognitive and nutritional measures and several vari- for which there were 10 significant correlations of 90 possible.
ables (age, education, sex) that might confound nutrition-
cognition associations. Sex was related to several nutritional Concurrent nutrition-cognition correlations
variables, but there were no significant effects of sex on cog-
There were several significant associations between concur-
nitive performance; by contrast, education was related to cog- rent biochemical or dietary vitamin status and cognitive per-
nitive performance, but generally not to nutritional variables;
formance. Plasma concentrations of vitamin C were positively
therefore, we did not adjust for sex or education in the corre- correlated with Rey-Osterrieth Copy test performance and
lational analyses. Age consistently influenced cognitive perfor- nearly significant for the Visual Reproduction test. Dietary
mance, and was also inversely correlated with three biochem- vitamin C was nearly significantly correlated with the same
ical variables (albumin in 1980, vitamin B-12 in 1980, and cognitive scores. For both thiamine and riboflavin, concurrent
vitamin C in 1986); accordingly, age was covaried in the
dietary intake was positively correlated with Abstraction test
correlational analyses for these nutrients. Age and education
performance, and biochemical activity coefficients were nega-

Downloaded from by guest on 13 June 2020

did not differ for users and nonusers of vitamin supplements, tively correlated with Abstraction test results (with higher
except for an age differential between users and nonusers of coefficients indicating poorer nutrient status). Concurrent di-
supplemental vitamin C in 1986 and a difference in education etary niacin intake correlated significantly with Rey-Osterrieth
for folate supplement users and nonusers in 1986; these factors Recall and Abstraction test performance. All three measures of
were covaried in pertinent comparisons of cognitive test scores. concurrent folate status correlated significantly with Abstrac-
The SAS program (version 6, release 6.04; Cary, NC) was used tion test results.
for statistical analyses. There were also several associations between concurrent
dietary and biochemical measures of protein and cognitive
scores. Serum albumin correlated significantly with Logical
RESULTS Memory scores, serum transferrin with Rey-Osterrieth Copy
scores and Abstraction test scores, and dietary protein with
The mean scores on cognitive tests in 1986 are given in
Rey-Ostemeth Recall and Logical Memory scores. When total
Table 1, biochemical measures in 1980 and 1986 are given in
dietary protein was adjusted for body weight (g protein/kg
Table 2, and the median, 25th percentile, and 75th percentile
body wt), the correlations with Rey-Osterrieth Recall and Log-
scores for dietary intake measures in 1980 and 1986 are pre-
ical Memory scores remained significant (r = 0.19 and 0.20,
sented in Table 3. Also shown for comparison in Table 3 is the
respectively, P < 0.05). Rey-Osterrieth Recall scores, but not
1989 recommended dietary allowance (RDA) for each nutrient
other cognitive scores, were significantly correlated with past
for men and women aged > 50 y (21). In general, cognitive
total energy intake (r = 0.23, P < 0.05) and nearly significant
performance was at or above normative levels expected for
for concurrent energy intake (r = 0.16, P < 0.09). This
age, and average dietary intake of nutrients clearly exceeded
association was no longer present when energy intake was
RDA values (eg, the 25th percentile of dietary intake is at or
corrected for body weight (MJfkg body wt).
above the RDA for all nutrients except vitamin B-6, vitamin E,
and protein).
Past nutrition-cognition associations
The results of correlational analyses relating cognitive and
nutritional measures are presented in Table 4. For simplicity, There was a single significant correlation between past bio-
the pairs of correlations shown are those with a significant chemical nutrition status and cognitive performance. This was
(P < 0.05) or nearly significant (P < 0.10) result for either past between serum transferrin and Rey-Osterrieth Copy score, an
(1980) or concurrent (1986) nutritional status or both. Because association also seen for concurrent nutrition status. Of the nine
the aim of the analyses was to identify patterns of potentially significant correlations between past dietary intake and cogni-
important associations in a sample with above-average nutri- tive status, four were for dietary vitamin E, two each were for
tion and cognition, these levels of significance were considered vitamin A and vitamin B-l2, and one was for vitamin B-6.
appropriate, in contrast with a conservative (eg, Bonferroni) None of these associations for past dietary intake matched
approach of adjusting alpha levels for multiple correlations. Of significant associations for concurrent intake, although there
90 possible intercorrelations with concurrent nutritional status, were nearly significant associations for concurrent dietary vi-
15 were significant, compared with 5 expected by chance. tamin B-6 and B-12 intakes and Abstraction performance.

Vitamin supplements

TABLE 1 At both study times, several subjects reported taking self-

Average values for cognitive measures in 1986 selected supplements of various vitamins. In 1980, 31% of
subjects took folate supplements, 59% took vitamin C supple-
Cognitive test (raw score) ± SD Range ments, and 46-49% took supplements of the remaining seven
Rey-Osterrieth Complex Figure vitamins listed in Table 1. Supplement use was fairly stable
Copy (n = 136) 31.0 ± 4.0 16.5-36 over time, increasing just slightly in 1986, when 43% took
Recall (n = 135) 14.7 ± 6.7 1-33 folate supplements, 62% took vitamin C supplements, and
Wechsler Memory Scale 49-55% took supplements of the remaining vitamins. For each
Visual Reproduction (n 135) 6.0 ± 2.4 0-13 of the six B vitamins (thiamine, riboflavin, niacin, vitamin B-6,
Logical Memory (n = 133) 7.2 ± 2.8 0.25-14.75
folate, and vitamin B- 12), there were no differences between
Shipley-Hartford Abstraction (n 135) 10.4 ± 4.3 0-20
supplement users and nonusers in performance on the Logical

Average values for biochemical measures in 1980 and 1986’

Biochemical measure n 1980 1986

Plasma ascorbate (pmol/L) 133 69.3 ± 19.7 (9.1-134.6) 70.9 ± 20.3 (14.8-154.4)
Thiamine (activity coefficient) 135 1.6 ± 0.23 (0.95-2.15) 1.1 ± 0.03 (1.04-1.18)
Riboflavin (activity coefficient) 135 1 . 1 ± 0.07 (0.97-1 .35) 1 . 1 ± 0.09 (1.00-1.49)
Plasma folate (nmol/L) 133 14.3 ± 6.0 (3.4-33.5) 27.4 ± 14.3 (8.8-94.7)
Erythrocyte (nmol/L)
folate 134 759 ± 347 (249-2184) 1007 ± 408 (435-3172)
Plasma vitamin B-l2 (pmolIL) 133 437 ± 186(111-1033) 343 ± 174(108-1204)
Serum albumin (g/L) 132 41.6 ± 2.3 (36-47) 43.5 ± 2.2 (37-49)
Serum transferrin (g/L) 134 2.3 ± 0.4 (1 .6-3.6) 2. 1 ± 0.3 (1.4-3.1)

‘ i ± SD; range in parentheses.

Downloaded from by guest on 13 June 2020

Memory or Visual Reproduction tests. However, as shown in those of an earlier report from this sample (3), suggest some
Figures 1-3, B vitamin supplement users in 1986 had signif- associations between cognitive performance and dietary intake
icantly higher mean scores than nonusers on the Rey-Osterrieth and/or plasma concentrations of specific nutrients among nor-
Copy (P < 0.05), Rey-Osterrieth Recall (P < 0.05), and mally aging older adults. Generally, participants with the best
Abstraction (P < 0.01) tests, except for folate and the Rey- cognitive performance also had higher nutrient concentrations;
Osterrieth Copy and Recall tests. No comparable associations however, the magnitude of correlations is modest at best,
were seen between cognitive performance and past vitamin accounting for small percentages of the total variation in cog-
use. nitive scores. This is consistent with the outcomes of other
Vitamin C and A supplement use in 1986 had a few associ- recent studies of healthy elderly samples (4-7). The possibility
ations with higher cognitive scores: vitamin C with Visual that findings are due to chance cannot be excluded given the
Reproduction (P < 0.05) (not shown) and Rey-Osterrieth Copy weakness of the intercorrelations; however, the number of
scores (Figures 1-2) and vitamin A with Abstraction scores significant associations exceeds that expected by chance, and
(P < 0.01) (Figure 3). Past users of vitamin C supplements also the consistency of relations with certain cognitive measures
had higher scores than nonusers on the Rey-Osterrieth Copy argues against an entirely spurious set of outcomes.
test (P < 0.05) (not shown). Past users of vitamin E supple- Cognitive performance was not affected in a uniform way by
ments had higher scores than nonusers on four of the cognitive any nutrient. Dietary protein was more closely related to mem-
measures, as shown in Figure 4: Rey-Osterrieth Copy, Rey- ory than to visuospatial or abstraction test performance,
Osterrieth Recall, Visual Reproduction (all P < 0.01), and whereas the opposite pattern was found for associations with
Abstraction tests (P < 0.05). For the Rey-Osterrieth Recall and certain vitamins. However, given the modest size of the sig-
Abstraction tests, mean scores were also higher for concurrent nificant correlations, and the fact that cognitive as well as
(1986) vitamin E supplement users (P < 0.05 and 0.01, nutritional measures tend to be intercorrelated, it seems most
respectively). plausible that our findings suggest a general relation between
nutritional status and cognition, rather than any more specific
link between a particular nutrient and a given mode of cogni-
tive performance.
In light of the growing population of older people and the Among the cognitive tests, the Abstraction test was most
prevalence of cognitive impairment in old age, understanding often associated with nutrition measures. This test is one of the
of any and all factors that contribute to maintenance of cogni- most age-sensitive in the current battery, and although norma-
tive ability is a high priority. Our findings, combined with tive data are limited for very old cohorts, subjects in the present

Average values for dietary intake measures in 1980 and 1986

1980 1986
25th 75th 25th 75th
Dietary intake measure n Median percentile percentile Median percentile percentile RDA’

Vitamin C (mg/d) 122 270 145 649 257 141 661 60

Vitamin A (IU/d) 122 10 496 6519 16 248 10 503 7214 16 789 4000, 5000
Vitamin E (mg/d) 122 16.6 8.2 1 10.0 29.3 6.6 1 10.4 8, 10
Thiamine (mg/d) 122 1.93 1.18 9.34 2.47 1.33 9.00 1.0, 1.2
Riboflavin (mg/d) 122 2.78 1.50 6.61 3.06 1.67 8.04 1.2, 1.4
Niacin (mg/d) 122 27.5 18.3 57.7 30.9 19.4 53.3 13, IS
Vitamin B-6 (mg/d) 122 1.73 0.88 5.30 2.63 1.22 6.25 1.6, 2.0
Folate (g/d) 122 255 190 470 370 250 630 180, 200
Vitamin B-12 (g/d) 122 5.4 2.9 11.7 7.8 3.4 13.9 2.0
Protein (g/d) 122 75 64 85 72 56 87 50, 63

1 Recommended dietary allowances for women and men.


Spearman correlations between cognitive measures in 1986 and nutritional measures in 1980 and 1986’

Rey-Osterrieth WMS Visual WMS Logical Shipley-Hartford

Rey-Ostemeth Copy Recall Reproduction Memory Abstraction

Nutritional measure 1980 1986 1980 1986 1980 1986 1980 1986 1980 1986

Biochemical value
Plasma ascorbate 0.13 0.222 0.l5 0.l7
Thiamine (activity coefficient) 0.l6 -0.08 0.06
Riboflavin (activity coefficient) -0.l7 -0.23
Plasma folate 0.05 0.192
Erythrocyte folate 0.06 0.26
Plasma vitamin B-12 0.11 0.l5

Downloaded from by guest on 13 June 2020

Serum albumin 0.17’ 0.182
Serum transfemn 0.212 0.222 -0.02 0.24
Dietary intake
Vitamin C 0.10 0.l9 0.10 0.l6
Vitamin A 0.222 0.l5 0.24 0.03 0.1 1 0.l7
Vitamin E 0.202 0.06 0.28 0.13 0.192 -0.01 0.24 0.04
Thiamine -0.07 0.l6 0.08 0.l5 0.08 0.29
Riboflavin 0.l6 0.l6 0.1 1 0.222
Niacin 0.06 0.24 0.05 0.222
Vitamin B-6 0.192 0.l5
Folate 0. 1 73 0. 1 3 0. 1 1 0.232
Vitamin B-l2 0.192 0.10 0.202 0.l5
Protein 0.l5 0.25 0.14 0.26
I WMS, Wechsler Memory Scale.
2 p < 0.05.

3P < 0.10.
4P < 0.01.

study appear to have had greater difficulty with this test than 40-49-y olds (18). That is, although many of our subjects
with some other measures. For example, whereas the mean scored as well or better than younger adults on verbal memory,
Abstraction score for the highest-performing quartile in our they scored below what is expected for younger people on the
sample was at the 30th percentile of normative data for 40- Abstraction test. In the present sample, therefore, Abstraction
49-y olds (22), the mean Logical Memory score in the highest- may have been the test most likely to detect any mild, subclin-
performing quartile was above the 60th percentile of norms for ical cognitive impairments.



Vitamin Vitamin Vitamin Thiamine Ribo- Niacin Vitamin Folate Vitamin

C A E 52% flavin 49% B-6 43% B-12
62% 53% 55% 51% 54% 51%

Vitamin supplement and percentage supplemented

FIGURE 1. Mean scores on the Rey-Osterrieth Copy Test for participants with and without vitamin supplementation. Higher scores indicate better
visuospatial performance.







Downloaded from by guest on 13 June 2020

Vitamin Vitamin Vitamin Thiamine Ribo- Niacin Vitamin Folate Vitamin
C A E 52% flavin 49% B-6 43% B-12
62% 53% 55% 51% 54% 51%

Vitamin supplement and percentage supplemented

FIGURE 2. Mean scores on the Rey-Osterrieth Recall test for participants with and without vitamin supplementation. Higher scores indicate better
nonverbal memory performance.

The present results concur in a general way with those ings. Alternatively, the increased age of the sample in 1986
reported by Goodwin et al (3). Both analyses found modest may have lowered abstraction performance overall, thereby
associations between cognitive performance and nutritional increasing the sensitivity of this measure to nutrition effects.
measures, and in both, significant associations were noted with There is a clear trend in the present analyses toward higher
several different nutrients, including protein, folate, niacin, cognitive performance among participants taking vitamin sup-
ascorbate, vitamins B-6 and B-12, thiamine, and riboflavin. We plements compared with those not taking supplements. This
found fewer significant relations between nutrition and verbal was not attributable to higher education or younger age among
memory than did Goodwin et al, perhaps because of selective supplement users. Our original reason for testing the associa-
attrition effects. As noted above, a majority of subjects in the tion of cognition with supplement use was the puzzling result
current analyses had good verbal memory performance, which for dietary vitamin E (Table 2). Past, but not concurrent,
resulted in a more restricted range of scores. By contrast, we vitamin E intake was associated with four of the five cognitive
found more frequent associations between nutrients and ab- measures, a pattern different from that for any of the other
straction performance than had been reported by Goodwin et nutrients. We had no biochemical measure of vitamin E status,
al; different measures of abstraction were used in the 1980 and and we knew that the nutrient database for vitamin E was
1986 testings, which may account for the discrepancy in find- incomplete. Wondering if the result for past vitamin E intake



Vitamin Vitamin Vitamin Thiamine Ribo- Niacin Vitamin Folate Vitamin

C A E 52% flavin 49% B-6 43% 8-12
62% 53% 55% 51% 54% 51%

Vitamin supplement and percentage supplemented

FIGURE 3. Mean scores on the Abstraction test for participants with and without vitamin supplementation.




Downloaded from by guest on 13 June 2020

Cognitive test
FIGURE 4. Mean performance on cognitive tests as a function of concurrent and past use of vitamin E supplements.

was spurious, we used supplementation as a marker for vitamin supplementation might have adverse effects on cognition could
E intake, obtaining results consistent generally with those in not be addressed in these data because of the small number of
Table 2. Additionally, the analysis of supplement intake participants with unusually high supplement intakes.
showed two cognitive measures associated with concurrent The strength of associations in these analyses is likely to
vitamin E intake, associations that were not significant in have been attenuated by the fact that our participants were
correlational analyses (Table 2). generally well nourished and cognitively intact, and interpre-
Concurrent, but not past, supplementation with B vitamins tation of outcomes must be made bearing this in mind. None-
was associated with better performance on the abstraction theless, our findings provide an opportunity to examine some
measure and the more difficult of the two visuospatial tests of the hypotheses that have been raised about brain mecha-
(Rey-Osterrieth Complex Figure as opposed to WMS Visual nisms that might underlie cognition-nutrition relations. Possi-
Reproduction). The B vitamin supplement association was ble models nonspecific
are regarding the time line for interac-
mirrored by significant or near-significant association of ab- tion between nutrition and cognition. One possibility is that
straction with all of the concurrent biochemical and dietary B past nutritional status would influence brain processes over
vitamin measures (Table 2) and of Rey-Osterrieth Recall with time and be demonstrated in concurrent cognitive function, as
several concurrent dietary B vitamin measures (Table 2). This discussed below for vitamins A and E. Alternatively, concur-
similarity of findings for the six B vitamins is reasonable rent nutritional status might influence immediate brain pro-
because most were consumed as part of a multivitamin prep- cesses and be reflected concurrently in cognitive function, as
aration. Supplement use was associated both with better bio- suggested by the data for some of the other nutrients. These
chemical and dietary vitamin status and with better perfor- possiblities are not mutually exclusive and both might coexist.
mance on some cognitive tests. However, whether the For most nutrition variables, there were significant correlations
association is specific to one or to several B vitamins or in the relative rankings of participants from 1980 to 1986 (data
involves all six cannot be distinguished because of the use of not shown), suggesting moderate intraindividual continuity in
multivitamin preparations. The few associations of vitamin A nutritional status.
and C supplement use with cognition involved the same cog-
An antioxidant model
nitive measures correlated with those vitamins in Table 2, but
the association with either past or concurrent nutrition was not In the normal aging brain, there is a sharp increase in
consistent. monoamine oxidase B, important for breakdown of cat-
Prevalence of supplement intake may have provided a crude echolamines, and a corresponding decrease in dopamine in the
indication of vitamin status, identifying individuals with opti- striatum and of norepinephrine in the locus ceruleus, septum,
mal status for the nine vitamins included in the analysis. Thus, and substantia nigra (23). Behaviorally, there are a variety of
supplement intake may have contributed in a positive way to changes consistent with mild impairment of frontal/subcortical
the observed associations between cognition and nutrition. brain systems, including psychomotor slowing, decreased per-
Alternatively, persons with higher cognitive function may be formance on effortful memory tasks, and reduced flexibility in
more apt to take vitamin supplements than those with lower thought and action (24, 25).
cognitive performance. In the present sample, participants were According to Wolters and Calne (26) and others, the nigro-
generally average or above average in educational and occu- striatal system may be particularly vulnerable to lifelong wear
pational attainment and in their performance on cognitive tests; and tear resulting from accumulation of free radicals, neu-
therefore, these findings do not appear attributable to poor romelanin, or other products of dopamine oxidation. Research
nutritional habits resulting from dementia. Whether excessive with Parkinson disease patients suggests that sustained use of

medications with antioxidant properties (eg, monoamine oxi- tions between these nutrients and brain function include inhib-
dase inhibitors) may slow the course of nigrosthatal brain ited methylation of methyl acceptors such as myelin, neuro-
dysfunction (27). A diet rich in nutrients with antioxidant transmitters, and membrane phospholipids (28). Regland et al
properties (eg, vitamins A, E, and C, and carotenes) might yield (37) reported an association between vitamin B-12 supplemen-
similar beneficial effects, and if so, frontallsubcortical behav- tation and a lowering of monoamine oxidase activity in older
ioral functions might also be facilitated. adults with Alzheimer’s disease, suggesting another possible
Our findings provide some limited support for an antioxidant means by which vitamin B-12 concentrations might influence
model. Plasma concentrations of vitamin C and use of vitamin cognition function, as discussed above.
C supplements were related to scores on a visuospatial task We found that plasma, erythrocyte, and dietary folate were
(Rey-Osterrieth Copy) that places demands on frontoparietal all positively correlated with Abstraction scores, when folate
brain function. Other potentially relevant performances (eg, on status and cognition were measured at the same time. Dietary
tests of abstract reasoning or learning and memory) were either intakes of vitamins B-6 and B-12 were also related to Abstrac-
unrelated or only marginally related to vitamin C, although tion performance. Analyses of data obtained subsequently from

Downloaded from by guest on 13 June 2020

cognitive tests that are more specifically sensitive to frontal! this sample show that serum homocysteine was negatively
subcortical functioning would provide a better test of this correlated with serum and erythrocyte folate in 100 elderly
theory. volunteers in 1993, and mean homocysteine concentrations
For the other antioxidant vitamins, our results include were significantly lower in those consuming self-selected sup-
dietary and supplemental intake, but not biochemical mea- plements of folate and vitamins B-6 and B-12 than in the
sures. We lacked biochemical results for vitamin E and nonsupplemented participants (38). These results suggest that
carotene, and did not include serum concentrations of vita- further study of cognitive status is warranted in relation to
mm A in these analyses, because they are known to be vitamin status, supplement use, serum homocysteine, and vas-
insensitive to vitamin A nutritional status except in cases of cular disease.
frank deficiency or toxicity. Associations of cognition with Riggs et al (7) recently reported poorer visuospatial perfor-
dietary and supplemental intake of vitamins A and E were mance among older adults with lower plasma concentrations of
stronger for past than for concurrent intake, a pattern dif- vitamin B- 1 2 and folate and higher concentrations of homo-
ferent from that seen for the other nutrients. These vitamins cysteine; higher concentrations of vitamin B-6 were associated
function as fat-soluble antioxidants, and as such may exhibit with better performance on tests of working memory and
a different time course of interaction with brain function and incidental memory, but not with visuospatial performance.
cognition than do the water-soluble vitamins. Vitamin stores These associations were not explained by clinical diagnoses of
accumulated in the past, such as vitamin A in liver and cerebrovascular disease. Our study showed associations be-
vitamin E in adipose tissue, might be available for antioxi- tween performance on a difficult spatial copying and memory
dant function at a later time, yielding an association between task (the Rey-Osterrieth Complex Figure test) and dietary
cognition and past intake. Alternatively, the antioxidant intake or supplemental use of B vitamins, and in contrast with
protection provided by these vitamins, localized in lipid- Riggs et al, showed relatively consistent associations between
soluble environments in the brain (eg, myelin or cell mem- B vitamin intakes and Abstraction scores. We found a single
branes), may prevent slow oxidative changes that would be association between supplemental use of vitamin B-6 and
manifest by poorer cognition at a later time. memory performance, out of several such possible compari-
At best, our study provides an incomplete test of antioxidant sons. Differences in correlated skills may reflect differences in
effects, due to the specific measurement problems noted, and specific cognitive tests used in the two studies. For example,
because the relatively high intakes of these vitamins for the the test of reasoning included by Riggs et al was a paper-
sample as a whole would be expected to attenuate any possible folding task, which differs substantially from the verbal and
relations. The possibility that prior intake of fat-soluble anti- conceptual measure of abstraction included in the present
oxidants might convey some protection against subsequent research.
age-related neurobiological changes cannot be dismissed, but
more definitive study is clearly needed. A possible role for protein
An intriguing finding in the present analyses was the
association between protein intake and cognitive perfor-
Based on their review of the nutritional literature, Rosenberg mance. Concurrent dietary protein correlated significantly
and Miller (28) suggested that vitamin B-6, vitamin B-12, and with two tests of learning and memory (Rey-Osterrieth
folate may be particularly important for maintenance of cog- Recall and Logical Memory tests), serum albumin concen-
nitive ability in old age. Each of these nutrients is involved in tration also correlated with verbal memory performance
the metabolism of the sulfur amino acid homocysteine, and (Logical Memory test), and serum transferrin concentration
high homocysteine blood concentrations have been associated was related to visuospatial (Rey-Osterrieth Copy test) and
with increased risk for cardiovascular disease in some popula- abstraction performance.
tions (29-32). Moreover, relations between homocysteine con- Why might protein concentrations be important for cogni-
centrations and folate, vitamin B-6, and vitamin B-12 nutriture tion? Tyrosine, the precursor for the catecholamine neurotrans-
have been observed in both middle-aged and older adults mitters dopamine, norepinephrine, and epinephrine, is a corn-
(33-36). It was suggested, therefore, that neurocognitive mon constituent of protein foods. In normal aging, as noted
changes in old age may be mediated by cerebrovascular above, there is a reduction in brain dopamine that may be
changes, which in turn, may be linked to chronically elevated linked to the mild impairments commonly observed in the
homocysteine concentrations. Other mechanisms for associa- cognitive test profiles of older adults (39). Dietary intake of

protein may have an effect on catecholamine production and LM. Nutrition status and brain function in aging. Am J Clin Nutr
release, which in turn, can influence cognition performance. l990;52:93-l02.
Animals studies have reported facilitation of motor activity and 5. Bohnen H, Jolles J, Degenaar CP. Lower blood levels of vitamin Bl2

moderation of the deleterious effects of stress in old mice as a are related to decreased performance of healthy subjects in the Stroop

result of dietary supplementation with tyrosine; however, to Color-Word Test. Neurosci Res Comm l992;l 1:53-6.
6. Lauque 5, Wegner A, Oussett PJ, et al. Nutritional intake and neuro-
our knowledge, there are no parallel studies in humans. Be-
cognitive functions. Age Nutr l995;6:68-72.
cause cognitive testing is often considered mildly stressful,
7. Riggs KM, Spiro A III, Tucker K, Rush D. Relations of vitamin B-l2,
tyrosine concentrations may be of particular importance with
vitamin B-6, folate, and homocysteine to cognitive performance in the
respect to behavior during such testing or during everyday
Normative Aging Study. Am J Clin Nutr l996;63:306-l4.
situations that make demands for high performance. Studies 8. Garry PJ, Goodwin JS, Hunt WC, Hooper EM, Leonard AG. Nutri-
that manipulate tyrosine through acute administration in the tional status in a healthy elderly population: dietary and supplemental
laboratory or through long-term dietary supplementation, using intakes. Am J Clin Nutr l982;36:3l9-3l.
cognitive, motor, and mood outcome measures, would be valu- 9. Schaie KW, ed. Longitudinal studies of adult psychological develop-

Downloaded from by guest on 13 June 2020

able in addressing these issues. ment. New York: Guildford Press, 1983.
There are several reasons to suspect that older individuals 10. Department of Biometry. HVH-CWRU Nutrient Data Base Revision

may be particularly vulnerable to reductions in protein intake. 6. Cleveland, OH: Case Western Reserve University, 1982.
1 1. Garry PJ, Hunt WC, VanderJagt DJ, Rhyne RL. Clinical chemistry
Because of body composition changes in aging, particularly
reference intervals for healthy elderly subjects. Am J Clin Nutr
decreased skeletal mass, there is a shift in the overall pattern of
l989;50: 1219-30.
whole-body protein synthesis and breakdown (39). Thus, mus-
12. Garry PJ, Goodwin JS, Hunt WC, Gilbert BA. Nutritional status in
cle contributes ‘30% to whole-body protein turnover in young
a healthy elderly population: vitamin C. Am J Clin Nutr
adults, but only ‘20% in elderly individuals. Young (39) 1982;36:332-9.
speculated that these changes may reduce the efficiency of 13. Garry PJ, Goodwin JS, Hunt WC. Nutritional status in a healthy
dietary protein use in old age. Additionally, the reduced skel- elderly population: riboflavin. Am J Clin Nutr 1982;36:902-9.
etal muscle protein reservoir may decrease the capacity to 14. Garry PJ, Goodwin JS, Hunt WC. Folate and vitamin B-l2 status in a
adapt successfully to decreased dietary intake or to physiolog- healthy elderly population. J Am Geriatr Soc l984;32:7l9-26.
ical stress requiring protein synthesis. For example, the release 15. Garry PJ, Goodwin JS, Hunt WC, Gilbert BA. lion status and anemia
of the amino acid glutamine from skeletal muscle is important in the elderly: new findings and a review of previous studies. J Am
for immune function and also provides a labile store of nitrogen Geriatr Soc l983;3 1:389-99.

for protein synthesis by other tissues during stress such as 16. Garry PJ, Hunt WC. Biochemical assessment of vitamin status in the
elderly: effects of dietary and supplemental intakes. In: Hutchinson
infection, surgery, or trauma (37). One might further speculate
ML, Munro HN, eds. Nutrition and aging. Bristol Meyers Nutrition
that changes in the pattern of whole-body protein synthesis and
Symposia. Vol 5. Orlando, FL: Academic Press, 1986:117-37.
breakdown might alter the availability in neural tissue of neu-
17. Shipley WC. A self-administering scale for measuring intellectual
rotransmitter precursors, such as tyrosine, or might have some impairment and deterioration. J Psychol l940;9:371-7.
other unforeseen effect on function of neural tissue. 18. Wechsler D. A standardized memory scale for clinical use. J Psychol
1945; 19:87-95.
Conclusions 19. Rey A. L’examen psychologique dans les cas d’encephalopathie trau-
matique. (Psychological examination of traumatic encephalopathy.)
Because our results are correlational, they cannot be used to
Arch Psychol 1941 ;28:286-340(in French).
support any causal associations among nutritional and cogni-
20. Osterrieth PA. Le test de copie d’une figure complexe. (The complex
tive variables, nor can we exclude the possibility that the
figure copy test.) Arch Psychol l944;30:206-365(in French).
associations observed were due to unidentified factors (eg,
21. National Research Council. Recommended dietary allowances. 10th
exercise patterns) affecting both nutrition and cognition. How- ed. Washington, DC: National Academy Press, 1989.
ever, considered together with the earlier findings from this 22. Zachary RA. Shipley Institute of Living, revised manual. Los Angeles:
study and other recent reports, the present outcomes support Western Psychological Corporation, 1986.
the potential value of further research on cognitive effects of 23. Rogers J, Bloom FE. Neurotransmitter metabolism and function in the
nutritional status in normally aging adults. It will be important aging nervous system. In: Finch CE, Schneider EL, eds. Handbook of
to study larger samples at multiple points in time to assess the the biology of aging. New York: Van Nostrand Reinhold,
time course of associations between specific nutrients and 198:64-91.

age-sensitive aspects of cognitive performance. A 24. Albert MS, Kaplan E. Organic implications of neuropsychological
deficits in the elderly. In: Poon LW, Fozard JL, mis. New directions in
memory and aging. Hillsdale, NJ: L Erlbaum, 1980:403-32.
25. Mittenberg W, Seidenberg M, O’Leary DS, DiGiulio DV. Changes in
cerebral functioning associated with normal aging. J Clin Exp Neuro-
1. Bottiglieri T, Crellin RF, Reynolds EH. Folate and neuropsychiatry. psychol 1989;l 1:918-32.
In: Bailey LB. ed. Folate in health and disease. New York: Marcel 26. Wolters EC, Calne DB. Is Parkinson’s disease related to aging? In:
Dekker, 1995:435-62. Calne DB, Crippa G, Comi R, Horowksi R, Trabucchi M, eds. Par-
2. Jacques PJ, Riggs KM. B vitamins as risk factors for age-related kinsonism and aging. New York: Raven Press, 1989:125-32.
disease. In: Rosenberg IH, ed. Nutritional assessment of elderly pop- 27. Tetrud JW, Langston WJ. The effect of Deprenyl (Selegiline) on the
ulations: measure and function. New York: Raven Press, 1995:234-51. natural history of Parkinson’s disease. Science l989;245:5l9-24.
3. Goodwin iS, Goodwin JM, Garry PJ. Association between nutritional 28. Rosenberg IH, Miller JW. Nutritional factors in physical and
status and cognitive functioning in a healthy elderly population. JAMA cognitive functions of elderly people. Am J Clin Nutr 1992;
1983;249:2917-2l. 55(suppl): 12375-435.
4. Tucker DM, Penland JG, Sandstead HH, Milne DB, Heck DO, Klevay 29. Kang 55, Wong PWK, Malinow MR. Hyperhomocyt(e)inemia as a

risk factor for occlusive vascular disease. Annu Rev Nutr status and intake as primary determinants of homocysteinemia in an
l992;12:279-98. elderly population. JAMA 1993;270:2693-8.
30. Pancharuniti N, Lewis CA, Sauberlich HE, et al. Plasma homo- 35. Ubbink JB, Vermaak WJH, van der Merwe A, Becker PJ. Vitamin
cyst(e)ine, folate, and vitamin B-12 concentrations and risk for early- B-12, vitamin B-6, and folate nutritional status in men with hyperho-
onset coronary artery disease. Am J Clin Nutr 1994;59:940-8. mocysteinemia. Am J Clin Nutr 1993;57:47-53.
31. Stampfer Mi, Malinow MR, Willett WC, et al. A prospective study of 36. Newman RP, Weingartner H, Smallberg SA. Effortful and automatic
plasma homocyst(e)ine and risk of myocardial infarction in US phy- memory effects of dopamine. Neurology l984;34:805-7.
sicians. JAMA 1992;268:877-8l. 37. Regland B, Gottfries CG, Oreland L. Vitamin B12-induced reduction
32. Ueland PM, Refsum H, Brattstrom L. Plasma homocysteine and car- of platelet monoamine oxidase activity in patients with dementia and
diovascular disease. In: Francis RB Jr, ed. Atherosclerotic cardiovas- pernicious anaemia. Eur Arch Psychiatry Clin Neurosci
cular disease, hemostasis and endothelial function. New York: Marcel 1991 ;240:288-91.
Dekker, Inc. 1992. 38. Koehler KM, Romero U, Stauber PM, et al. Vitamin supplementation
33. Joosten E, van den Berg A, Riezler R, et al. Metabolic evidence that and other variables affecting serum homocysteine and methylmalonic
deficiencies of vitamin B-l2 (cobalamin), folate, and vitamin B-6 acid concentration in elderly men and women. J Am Coll Nutr

Downloaded from by guest on 13 June 2020

occur commonly in elderly people. Am J Clin Nutr 1996; 15:364-74.
1993;58:468-76. 39. Young VR. Amino acids and proteins in relation to the nutrition of
34. Selhub J, Jacques PF, Wilson PWF, Rush D, Rosenberg IH. Vitamin elderly people. Age Ageing 1990; 19(suppl):S 10-524.