Deficiencies in folate and vitamin B12 have long been known to cause various neuropsychiatric symptoms.1 In addition, folic acid, vitamins B6 and B12 are cofactors in the enzymatic pathways of homocysteine metabolism, and high intake of B vitamins reduces homocysteine levels. High homocysteine has direct neurotoxic effects; when injected into rat cerebral cortex, dose-dependent neuronal degeneration is observed.2 Hippocampal cultures incubated in folic acid-deficient medium or in homocysteine demonstrate cell death and increased vulnerability of neurons to amyloid β.3 In addition, elevated homocysteine levels may increase the risk for vascular disease,4 which, in turn, has been implicated in cognitive decline.5
Many epidemiologic studies have directly examined the association of folate and vitamin B12 with cognitive function in older persons; however, the results have been inconsistent. Numerous studies have reported apparent benefits of high blood levels of B vitamins on cognition6–19; however, others have found null results.20–24 Most of these studies have been cross-sectional with generally small sample sizes. A recent large-scale prospective study of dietary folate, with 6 years of follow-up,25 observed worse rates of cognitive decline for those with higher folate intake.
We prospectively examined the association of plasma levels of folate and vitamin B12 with cognitive function measured approximately 10 years after blood draw using data from older women in the Nurses’ Health Study.
Nurses’ Health Study
The Nurses’ Health Study began in 1976, when 121,700 female registered nurses aged 30 to 55 years, living in 11 U.S. states, returned a mailed questionnaire about their lifestyle and health. Since 1976, participants have completed biennial questionnaires to update this information.
From 1989 to 1990, blood samples were collected from one third of participants; details of the blood collection have been published elsewhere.26 Briefly, participants volunteered to send a blood sample by overnight mail, shipped on ice, to our laboratory. Approximately 70% were fasting samples. Ninety-seven percent of the samples were received within 26 hours of being drawn, and the stability of a variety of biomarkers in whole blood for 24 to 48 hours has been previously documented.27 Samples were processed and separated into plasma, red blood cells, and white blood cells and have been stored in liquid nitrogen freezers. We have used these samples to study the relation of many plasma biomarkers, including plasma folate and vitamin B12, to chronic diseases in various nested case–control studies.28,29
Health and lifestyle characteristics were similar between the whole Nurses’ Health Study cohort and those who returned blood samples. For example, 43% of the entire cohort never smoked versus 46% of those who provided blood; mean alcohol intake was 5 g per day for both groups.
Cognitive Function Subcohort
From 1995 through 2001, we selected 22,715 Nurses’ Health Study participants age 70 years and older, free of diagnosed stroke, for a substudy of cognitive function. Of those whom we contacted, 92% completed the interview (n = 19,514). We have been administering follow-up cognitive assessments at approximately 2-year intervals with very high follow-up rates. At the second cognitive assessment, we maintained 91% participation (5% declined and 4% died); a third assessment is ongoing (86% complete to date, also with 91% participation to date).
Telephone Cognitive Assessment
All interviews were administered by trained nurses; before any testing was done, the interviewers confirmed that participants had no hearing difficulties and, if necessary, adjusted their speaking volume to minimize difficulties.
When we began cognitive testing, we used only the Telephone Interview for Cognitive Status30 (TICS; mean score in this population = 34, standard deviation [SD] = 3, range = 8–41); this is a telephone adaptation of the Mini-Mental State Examination. We gradually added 5 other cognitive tests to the initial testing, and so the sample size differs somewhat for each test. The additional tests are: immediate and delayed recalls of the East Boston Memory Test31 (EBMT; immediate recall: mean = 9, SD = 2, range =0–12; delayed recall: mean = 9, SD = 2, range = 0–12); a test of category fluency in which women name animals during 1 minute (mean = 17, SD = 5, range = 0–38); a delayed recall of the TICS 10-word list (mean = 2, SD = 2, range = 0–10); and digit span backward, in which women repeat backward increasingly long series of digits (mean = 7, SD = 2, range = 0–12).
Our telephone method showed high validity; we found a correlation of 0.81 comparing overall performance on our brief telephone interview with overall performance measured from an in-person interview in a similar group of well-educated women.32 In tests of instrument reliability, we administered the TICS twice to 61 Nurses’ Health Study participants at an interval of 1 month and found a correlation of 0.70.33
We focused on 2 main outcomes: overall performance on our test battery and verbal memory, a strong predictor of Alzheimer disease development.34,35 The primary outcome was a global composite score combining our 6 cognitive tests by averaging the z-scores for each test. The verbal memory score was calculated by averaging the z-scores of 4 measures: immediate and delayed recalls of both the EBMT and TICS 10-word list. The global score and the verbal memory score were calculated only for participants who completed all the component tests. Such composite scores are commonly used in cognitive research32 because they integrate information from a variety of sources and provide a more stable representation of overall cognition than a single test.
Population for Analysis
Figure 1 is a flow-chart depicting the selection of participants for analysis. We first included only women who met the following criteria: 1) completed the first cognitive interview (1995–2001); 2) provided a blood sample (1989–1990); and 3) had plasma levels of folate and vitamin B12 measured as part of nested case–control studies of heart disease, breast cancer, colon cancer, or colon polyps. From these 938 women, we excluded those who were selected as cases in the nested case–control studies of heart disease, breast cancer, or colon cancer, because their disease status may be associated with both B vitamin levels and cognition; however, we did not exclude cases of colon polyps, because having a history of polyps is unlikely to have substantial influence on cognition. Thus, 635 women were included in the primary analyses.
Characteristics of the 635 women in the primary analyses were quite similar to those in the entire cognitive study. For example, mean age at cognitive assessment was identical (74 years); mean TICS score at the first interview was 34 in both the subset and entire group; and prevalence of current multivitamin use at blood draw was 41% in the subset and 40% in the entire group. Thus, despite using a convenience sample, the analytic sample appeared representative of the cognitive cohort.
In preliminary analyses of cognitive decline over 4 years, we included the subset of 391 women who had completed all follow-up cognitive assessments to date (ie, a total of 3 interviews); repeated cognitive testing is ongoing for the remainder of the women. The 391 women in this subset were similar to the entire analytic population of 635; their mean age was 74.0 years (vs 74.0 years overall), mean folate level was 9.9 ng/mL (vs 10.0 ng/mL overall), mean vitamin B12 level was 449 ng/mL (vs 459 ng/mL overall), and mean TICS score was 34 (vs 34 overall).
Ascertainment of Plasma Folate and Vitamin B12
All assays were conducted at the Jean Mayer USDA Human Nutrition Research Center at Tufts University. Levels of folate and vitamin B12 were determined by a radioassay kit (Bio-Rad, Richmond, CA).29 From each of the 9 batches of different nested case–control studies, blinded replicate samples were included for quality control; the coefficients of variation for folate ranged from 4.8% to 12.0% (median = 9.8%) and for vitamin B12 from 3.6% to 13.7% (median = 8.3%).
Approximately 70% of blood samples were fasting samples. The plasma nutrient levels from nonfasting samples were similar but slightly higher than fasting samples; for folate, nonfasting samples were approximately 10% higher on average, and for vitamin B12, levels were approximately 1% higher. Because alternate analyses excluding the nonfasting samples yielded qualitatively similar results, we present analyses presented based on the data from both fasting and nonfasting samples.
For the main analysis of performance in the initial cognitive interview, we used linear regression to estimate age- and education-adjusted and multivariable-adjusted mean differences in performance across plasma nutrient quartiles. There was little batch-to-batch variation, and the median values for both nutrients were comparable across batches; thus, we analyzed quartiles created from raw values of folate and vitamin B12 to maximize interpretability of results. In an alternate analysis in which we analyzed quartiles created with batch-specific cut points, we confirmed that the results were nearly identical.
For longitudinal analysis using data from a subset of 391 participants who completed all follow-up interviews to date, we used repeated-measures models incorporating random effects for intercepts and slopes.36 This approach permits description of individual paths of decline and provides explicit tests regarding the relation of exposures to rates of cognitive change.
In multivariable models, we included a wide array of potential confounding variables selected a priori from the published literature as risk factors for poor cognition. The covariates included: assay batch (categories of 1–9), time between blood draw and cognitive interview (years), age at interview (years), highest attained education (registered nurse, Bachelor's, Master's or above), history of diabetes, history of high blood pressure, history of high cholesterol, postmenopausal hormone use (current, past, never), age at menopause (<50, 50–52, 53+ years), body mass index (<22, 22–24, 25–29, 30+ kg/m2), cigarette smoking (current, past, never), antidepressant use, aspirin use (<3/wk, 3+/wk), alcohol intake (0, <5 g/d, 5–14 g/d, 15+ g/d), physical activity (quintiles of metabolic-equivalent-hours/wk), mental health index (0–51, 52–100), and energy–fatigue index (0–49, 50–100) from the Medical Outcomes Short Form-36 (a questionnaire to assess self-perceived health status)37 and use of vitamin E supplements. Information on potential confounding variables was determined from the biennial questionnaire responses immediately before the blood collection. In main analyses, we did not include current multivitamin use (at blood draw) as a covariate, because this would be equivalent to partially adjusting for our exposures of interest; however, in alternate analyses adjusting for current multivitamin use, findings remained quite similar to those reported here. To determine dietary intake of folate and vitamin B12 from food sources, we used the Willett food frequency questionnaire and USDA data on nutrient contents of various foods.38
To take into account possible changes in health and lifestyle between the time of blood draw and time of cognitive interview, we performed additional analyses using information on potential confounding variables obtained from the questionnaire immediately before the initial cognitive interview rather than at the time of the blood draw. However, there were minimal changes from the results presented here.
We performed several secondary analyses. First, to evaluate the effect of overall vitamin B status, we examined women with high levels of both folate and vitamin B12 (highest 20th percentile of each nutrient; >14.3 ng/mL for folate and >575.0 pg/mL for vitamin B12) compared with women with low levels of both folate and vitamin B12 (lowest 20th percentile of each nutrient; <5.1 ng/mL for folate and <319.2 pg/mL for vitamin B12). Very few people (n = 14) had deficient levels of both folate and vitamin B12 using the common definition of <4 ng/mL for folate and <250 pg/mL for vitamin B12; however, our cut points to define “low” were fairly close to deficient levels. We also conducted analyses stratified by current multivitamin use as of the biennial questionnaire immediately preceding the blood draw. Finally, because alcohol can interfere with folate metabolism and vitamin B12 absorption,1 we also examined effect modification by alcohol intake at blood draw.
We observed a wide range in the distribution of plasma nutrient levels (Table 1) with over a 4-fold difference in the median levels of plasma folate between the top and bottom quartiles and over a 2-fold difference for vitamin B12. Mean age at blood draw was identical across the nutrient quartiles (63 years), although women in the higher quartiles had higher educational attainment, had less obesity, and somewhat more hypercholesterolemia, especially for vitamin B12; they also smoked cigarettes less, took more aspirin and, as expected, used vitamin supplements substantially more than women with lower B vitamin levels. Cognitive performance on the first assessment was generally similar across plasma nutrient levels.
Initial Cognitive Function
In general, we observed no differences in mean test scores across quartiles of plasma folate or vitamin B12 (Table 2). For example, for the top versus bottom quartiles of plasma folate, the multivariable-adjusted mean difference on the global score was 0.06 standard units (95% confidence interval [CI] = −0.10 to 0.22). Similarly, for plasma vitamin B12, for the top versus bottom quartiles, the mean difference on the global score was 0.15 (0.00 to 0.31). When we included both nutrients together in models, the estimates for each nutrient changed very little (data not shown). Alternatively, when we modeled plasma levels of both folate and vitamin B12 as continuous variables, we did not observe evidence of dose–response trends.
When we evaluated the combined effects of especially high versus low levels of folate and vitamin B12 (Table 3), women whose plasma levels were high for both nutrients performed markedly better on the global score than women whose levels were low for both nutrients (mean difference = 0.34; 0.05 to 0.62). This mean difference in the global score was equivalent to the mean difference we observed between subjects 6 years apart in age. For the TICS and the verbal score, cognitive scores were also generally higher in those with high levels of both folate and vitamin B12.
When we examined associations by strata of current multivitamin use immediately before blood draw (Table 4), we observed no interactions for plasma folate, but we found an apparent interaction between vitamin B12 and multivitamin use. Specifically, for women who did not use multivitamins, there was no clear evidence of a relation between vitamin B12 and cognition. Compared with the reference group (women who did not use multivitamins and were in the lowest quartile of vitamin B12), the women who used multivitamins and still had low levels of plasma vitamin B12 appeared to have worse cognitive performance (mean difference = −0.18; −0.43 to 0.07), whereas those with multivitamin use and high levels of plasma vitamin B12 had better cognitive scores (mean difference = 0.19; −0.01 to 0.40). These findings should be interpreted cautiously because there were relatively small numbers of subjects within strata (an average of approximately 80 subjects in each strata). We found no suggestion of interactions with alcohol intake.
We generally did not observe associations with decline in cognition for high plasma levels of folate, vitamin B12, or the combination of both (Table 5). In addition, women who had high levels of both nutrients did not have different rates of decline compared with women who were low in both nutrients (mean difference in rate of decline in global score = −0.05; −0.14 to 0.05).
We found no relation of plasma folate and vitamin B12 levels to cognitive function or cognitive decline over 4 years. There was a suggestion that high plasma levels of both nutrients may be associated with better cognitive performance than for low levels of both nutrients, similar to reports from other studies.6,7 However, we did not observe this association for cognitive decline.
We observed a possible relation between vitamin B12 levels and cognition among multivitamin users. This result deserves further research. Interestingly, we also found particularly poor cognition among women with low plasma vitamin B12 despite multivitamin use. It seems possible that underlying illnesses that contribute to malabsorption of vitamin B12 from supplements could also be associated with poor cognition. Another possibility is the better absorption of vitamin B12 from supplements compared with the natural vitamin B12 in foods. Nonetheless, these associations could be due to chance and were based on relatively small samples within the strata and thus should be interpreted with caution.
The unique strengths of this study include the prospective design with folate and vitamin B12 levels assessed in relation to cognitive function approximately 10 years later. In particular, because diminished cognition develops over many years, exposures at these younger ages may have special significance. In addition, our folate measurements predate folate supplementation of the dietary grain supply that started in January 1998 and thus may more accurately reflect long-term folate status than do more recent measurements.39 Our plasma measurements of nutrients also have key advantages over self-reported intake, because they are not subject to errors due to recall and, more importantly, better capture bioavailable nutrient levels. Finally, there was a wide range in the levels of nutrients within the subjects, allowing a strong contrast between high and low levels of folate and vitamin B12; indeed, although these subjects are a homogeneous group of health professionals, we have previously found positive relations between low plasma folate and other diseases,29,40 attesting to our ability to detect such relations. Nonetheless, few subjects were deficient in either folate or vitamin B12 and thus we could not adequately explore cognitive effects of true deficiency.
Our study may be limited by several factors. The participants included in these analyses were a select subset of those who had plasma vitamin levels measured for other reasons; thus, it is possible that this group may not be representative of the entire cognitive cohort. However, various demographic and lifestyle characteristics were quite comparable in the analytic sample and the overall population. In addition, it is possible that we may have missed short-term cognitive effects resulting from increased folate intake because most of the initial cognitive assessments were made after folate fortification began. However, we believe this bias is minimal given that cognitive impairment takes many years to develop, and it seems unlikely that folate changes due to fortification would manifest as sudden, longlasting changes in cognition.
In this observational study, confounding may also be an issue. However, we conducted numerous analyses to explore confounding; we adjusted for potential confounding factors at blood draw as well as at cognitive interview, and we stratified analyses by multivitamin use. Moreover, in this cohort of health professionals, with similar access to health care and health knowledge, many opportunities for confounding are inherently reduced. Finally, our analyses of cognitive decline were limited to a subset of subjects, and the follow-up duration may have been relatively short to observe potential associations. Thus, we will reexamine this issue in the coming years with additional follow-up.
Many other studies have examined blood levels of folate and vitamin B12 in relation to cognition; however, the majority of studies have been cross-sectional and the results have been largely inconsistent. For example, among cross-sectional studies, some have shown low levels of either nutrient were related to poorer cognition8,12,16; others have reported that low folate, but not vitamin B12, was associated with worse cognition,7,10,11,13,17,41 whereas some found that low vitamin B12, not folate, was associated with cognitive impairment.14,15,19 Moreover, other studies found no association with either nutrient.23,24
Among the limited prospective studies, the literature is also unclear. In the Veterans Affairs Normative Aging Study (n = 321),42 high baseline plasma folate levels predicted better scores in the figure-copying test taken after 3 years of follow-up, whereas no association was observed with vitamin B12. In the MacArthur Study (n = 370), those in the lowest quartile of plasma folate had a 1.6-fold increase in the risk of cognitive decline over a 7-year follow-up; there was no association with vitamin B12.18 In contrast, in a recent prospective study (n = 3718) of dietary intake of B vitamins, Morris et al25 found that low levels of folate intake were actually related to better cognitive health, and high vitamin B12 predicted better scores, particularly among the oldest participants. Nonetheless, the majority of prospective investigations,9,21,43 including our own, have reported no association of either nutrient with cognitive function or decline.
In summary, among generally healthy well-nourished aging women, we found no clear association of plasma folate or vitamin B12 levels with cognitive function or cognitive decline. Although preventing deficiency of folate and vitamin B12 remains an important health concern for older persons, especially combined deficiency of both these vitamins, our data do not support the hypothesis that high folate or vitamin B12 levels provide clear cognitive benefits.
1. Selhub J, Bagley LC, Miller J, et al. B vitamins, homocysteine, and neurocognitive function in the elderly. Am J Clin Nutr
2. Beal MF, Swartz KJ, Finn SF, et al. Neurochemical characterization of excitotoxin lesions in the cerebral cortex. J Neurosci
3. Kruman II, Kumaravel TS, Lohani A, et al. Folic acid deficiency and homocysteine impair DNA repair in hippocampal neurons and sensitize them to amyloid toxicity in experimental models of Alzheimer's disease. J Neurosci
4. Lentz SR, Sadler JE. Inhibition of thrombomodulin surface expression and protein C activation by the thrombogenic agent homocysteine. J Clin Invest
5. Launer LJ. Demonstrating the case that AD is a vascular disease: epidemiologic evidence. Ageing Res Rev
6. Bell IR, Edman JS, Marby DW, et al. Vitamin B12
and folate status in acute geropsychiatric inpatients: affective and cognitive characteristics of a vitamin nondeficient population. Biol Psychiatry
7. Wahlin A, Hill RD, Winblad B, et al. Effects of serum vitamin B12
and folate status on episodic memory performance in very old age: a population-based study. Psychol Aging
8. Ortega RM, Manas LR, Andres P, et al. Functional and psychic deterioration in elderly people may be aggravated by folate deficiency. J Nutr
9. La Rue A, Koehler KM, Wayne SJ, et al. Nutritional status and cognitive functioning in a normally aging sample: a 6-y reassessment. Am J Clin Nutr
10. Hassing L, Wahlin A, Winblad B, et al. Further evidence on the effects of vitamin B12
and folate levels on episodic memory functioning: a population-based study of healthy very old adults. Biol Psychiatry
11. Morris MS, Jacques PF, Rosenberg IH, et al. Hyperhomocysteinemia associated with poor recall in the third National Health and Nutrition Examination Survey. Am J Clin Nutr
12. Duthie SJ, Whalley LJ, Collins AR, et al. Homocysteine, B vitamin status, and cognitive function in the elderly. Am J Clin Nutr
13. Quadri P, Fragiacomo C, Pezzati R, et al. Homocysteine, folate, and vitamin B-12 in mild cognitive impairment, Alzheimer disease, and vascular dementia. Am J Clin Nutr
14. Lewerin C, Matousek M, Steen G, et al. Significant correlations of plasma homocysteine and serum methylmalonic acid with movement and cognitive performance in elderly subjects but no improvement from short-term vitamin therapy: a placebo-controlled randomized study. Am J Clin Nutr
15. Bunce D, Kivipelto M, Wahlin A. Apolipoprotein E, B vitamins, and cognitive function in older adults. J Gerontol B Psychol Sci Soc Sci
16. Lewis MS, Miller LS, Johnson MA, et al. Elevated methylmalonic acid is related to cognitive impairment in older adults enrolled in an elderly nutrition program. J Nutr Elder
17. Campbell AK, Jagust WJ, Mungas DM, et al. Low erythrocyte folate, but not plasma vitamin B-12 or homocysteine, is associated with dementia in elderly Latinos. J Nutr Health Aging
18. Kado DM, Karlamangla AS, Huang MH, et al. Homocysteine versus the vitamins folate, B6
, and B12
as predictors of cognitive function and decline in older high-functioning adults: MacArthur Studies of Successful Aging. Am J Med
19. Jelicic M, Jonker C, Deeg DJ. Effect of low levels of serum vitamin B12
and folic acid on cognitive performance in old age: a population-based study. Dev Neuropsychol
20. Bernard MA, Nakonezny PA, Kashner TM. The effect of vitamin B12
deficiency on older veterans and its relationship to health. J Am Geriatr Soc
21. Teunissen CE, Blom AH, Van Boxtel MP, et al. Homocysteine: a marker for cognitive performance? A longitudinal follow-up study. J Nutr Health Aging
22. Luchsinger JA, Tang MX, Shea S, et al. Plasma homocysteine levels and risk of Alzheimer disease. Neurology
23. Ravaglia G, Forti P, Maioli F, et al. Homocysteine and cognitive performance in healthy elderly subjects. Arch Gerontol Geriatr Suppl
24. Ariogul S, Cankurtaran M, Dagli N, et al. Vitamin B12
, folate, homocysteine and dementia: are they really related? Arch Gerontol Geriatr
25. Morris MC, Evans DA, Bienias JL, et al. Dietary folate and vitamin B12
intake and cognitive decline among community-dwelling older persons. Arch Neurol
26. Hankinson SE, Willett WC, Manson JE, et al. Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst
27. Hankinson SE, London SJ, Chute CG, et al. Effect of transport conditions on the stability of biochemical markers in blood. Clin Chem
28. Platz EA, Hankinson SE, Hollis BW, et al. Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and adenomatous polyps of the distal colorectum. Cancer Epidemiol Biomarkers Prev
29. Zhang SM, Willett WC, Selhub J, et al. Plasma folate, vitamin B6
, vitamin B12
, homocysteine, and risk of breast cancer. J Natl Cancer Inst
30. Brandt J, Folstein MF. Telephone Interview for Cognitive Status: Professional Manual
. Lutz, FL: Psychological Assessment Resources, Inc; 2003.
31. Albert M, Smith LA, Scherr PA, et al. Use of brief cognitive tests to identify individuals in the community with clinically diagnosed Alzheimer's disease. Int J Neurosci
32. Wilson RS, Schneider JA, Barnes LL, et al. The apolipoprotein E e4 allele and decline in different cognitive systems during a 6-year period. Arch Neurol
33. Weuve J, Kang JH, Manson JE, et al. Physical activity, including walking, and cognitive function in older women. JAMA
34. Elias MF, Beiser A, Wolf PA, et al. The preclinical phase of Alzheimer disease: a 22-year prospective study of the Framingham Cohort. Arch Neurol
35. Small BJ, Fratiglioni L, Viitanen M, et al. The course of cognitive impairment in preclinical Alzheimer disease: three- and 6-year follow-up of a population based sample. Arch Neurol
36. Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics
37. Ware J, Snow K, Kosinski M, et al. SF-36 Health Survey: Manual and Interpretation Guide
. Boston: The Health Institute, New England Medical Center; 1993.
38. Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol
39. Jacques PF, Selhub J, Bostom AG, et al. The effect of folic acid fortification on plasma folate and total homocysteine concentrations. N Engl J Med
40. Han J, Hankinson SE, Zhang SM, et al. Interaction between genetic variations in DNA repair genes and plasma folate on breast cancer risk. Cancer Epidemiol Biomarkers Prev
41. Irizarry MC, Gurol ME, Raju S, et al. Association of homocysteine with plasma amyloid β protein in aging and neurodegenerative disease. Neurology
42. Tucker KL, Qiao N, Scott T, et al. High homocysteine and low B vitamins predict cognitive decline in aging men: the Veterans Affairs Normative Aging Study. Am J Clin Nutr
43. Crystal HA, Ortof E, Frishman WH, et al. Serum vitamin B12
levels and incidence of dementia in a healthy elderly population: a report from the Bronx Longitudinal Aging Study. J Am Geriatr Soc