Homocysteine, B Vitamins, and Cognitive Impairment

Abstract

Moderately elevated plasma total homocysteine (tHcy) is a strong modifiable risk factor for vascular dementia and Alzheimer's disease. Prospectively, elevated tHcy is associated with cognitive decline, white matter damage, brain atrophy, neurofibrillary tangles, and dementia. Most homocysteine-lowering trials with folate and vitamins B6 and/or B12 tested as protective agents against cognitive decline were poorly designed by including subjects unlikely to benefit during the trial period. In contrast, trials in high-risk subjects, which have taken into account the baseline B vitamin status, show a slowing of cognitive decline and of atrophy in critical brain regions, results that are consistent with modification of the Alzheimer's disease process. Homocysteine may interact with both risk factors and protective factors, thereby identifying people at risk but also providing potential strategies for early intervention. Public health steps to slow cognitive decline should be promoted in individuals who are at risk of dementia, and more trials are needed to see if simple interventions with nutrients can prevent progression to dementia.

Full text

NU36CH09-Smith ARI 10 June 2016 11:7
Homocysteine, B Vitamins,
and Cognitive Impairment
A. David Smith1and Helga Refsum1,2
1OPTIMA, Department of Pharmacology, University of Oxford, Oxford OX1 3QT,
United Kingdom; email: david.smith@pharm.ox.ac.uk
2Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo,
0316 Oslo, Norway; email: helga.refsum@medisin.uio.no
Annu. Rev. Nutr. 2016. 36:211–39
The Annual Review of Nutrition is online at
nutr.annualreviews.org
This article’s doi:
10.1146/annurev-nutr-071715-050947
Copyright c
2016 by Annual Reviews.
All rights reserved
Keywords
folate, cobalamin (vitamin B12), Alzheimer’s disease, dementia, cognition,
clinical trial
Abstract
Moderately elevated plasma total homocysteine (tHcy) is a strong modifi-
able risk factor for vascular dementia and Alzheimer’s disease. Prospectively,
elevated tHcy is associated with cognitive decline, white matter damage,
brain atrophy, neurofibrillary tangles, and dementia. Most homocysteine-
lowering trials with folate and vitamins B6 and/or B12 tested as protective
agents against cognitive decline were poorly designed by including subjects
unlikely to benefit during the trial period. In contrast, trials in high-risk sub-
jects, which have taken into account the baseline B vitamin status, show a
slowing of cognitive decline and of atrophy in critical brain regions, results
that are consistent with modification of the Alzheimer’s disease process. Ho-
mocysteine may interact with both risk factors and protective factors, thereby
identifying people at risk but also providing potential strategies for early in-
tervention. Public health steps to slow cognitive decline should be promoted
in individuals who are at risk of dementia, and more trials are needed to see
if simple interventions with nutrients can prevent progression to dementia.
211
Click here to view this article's
online features:
• Download figures as PPT slides
• Navigate linked references
• Download citations
• Explore related articles
• Search keywords
ANNUAL
REVIEWS
Further
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
tHcy: total
homocysteine
Contents
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
A Cardinal Principle in Nutrition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
HOMOCYSTEINE,COGNITIVE DECLINE,AND DEMENTIA............... 214
Initiationof CognitiveImpairment inAging ..................................... 215
Conversionfrom CognitiveImpairment toDementia............................. 215
Incidence of Dementia and Alzheimer’s Disease in a Population . . . . . . . . . . . . . . . . . . . 216
IncreasedRate ofCognitive DeclineinAlzheimer’sDisease....................... 217
IncreasedWhite MatterDamage intheBrain.................................... 218
Density of Neurofibrillary Tangles in the Brain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
IncreasedRate ofAtrophy intheBrain........................................... 218
HOMOCYSTEINE:MARKER ORCULPRIT?................................... 219
Results from Genetic Variations and Mendelian Randomization . . . . . . . . . . . . . . . . . . 220
POSSIBLE MECHANISMS THROUGH WHICH HOMOCYSTEINE
CANHARM THEBRAIN..................................................... 221
CerebrovascularEffectsofHomocysteine........................................ 223
Activationof TauKinases....................................................... 223
Inhibition of Methylation Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
WHAT HAVE CLINICAL TRIALS INVOLVING B VITAMIN
SUPPLEMENTATIONTOLD US?........................................... 226
TheFACIT Trial.............................................................. 228
The VITACOG Trial and the Importance of Identifying the Groups
thatAre Likelyto Respond................................................... 228
POSSIBLEPUBLIC HEALTHIMPLICATIONS................................. 231
INTRODUCTION
In this review, we consider homocysteine and the related B vitamins, i.e., folate and vitamins
B6 and B12 (10, 115), in relation to cognitive impairment and dementia in the elderly. The
association of B vitamin status with normal function of the nervous system was already observed
in 1849 when Addison reported on the “wandering mind” of patients with pernicious anemia, a
syndrome later shown to be due to vitamin B12 deficiency (74, 127). In the 1960s, the harmful
effects of excess homocysteine on the brain were suggested by mental disturbances in children with
homocystinuria (74). These early studies have been reviewed (74, 116, 121). We focus here on
more recent evidence that moderately elevated plasma total homocysteine (tHcy) concentrations
or a low status of relevant B vitamins can cause cognitive impairment and dementia in the elderly.
The establishment of a causal relationship will have important implications for public health.
In a simplified scheme (Figure 1), there are several pathways by which low B vitamin status
can lead to cognitive impairment. Low folate or vitamin B12 status (pathways 1a or 1b) will lead
to elevated homocysteine, which in turn leads to cognitive impairment (pathway 2). Alternatively,
low folate or B12 status (pathways 3a and 3b) may lead to cognitive impairment, independent of
homocysteine. Third, pathways 1–3 may all be operative. Finally, the outcome of cognitive im-
pairment refers to impairment in several different cognitive domains or dementia with potentially
very different causes. A key question, therefore, is does elevated homocysteine, usually measured
as tHcy, serve as a marker for low B vitamin status or reflect another causal mechanism, or does
homocysteine itself have direct effects on the brain? A different but related question of scientific
212 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
Low folate status Low B12 status
Raised homocysteine status
Cognitive impairment
1a 1b
23b3a
Figure 1
Different pathways relating B vitamins to cognitive impairment. The distinct pathways to cognitive
impairment are identified by different numbers.
and potential public health relevance, whether high folate status impairs cognition, is reviewed in
the Supplemental Material (follow the Supplemental Material link from the Annual Reviews
home page at http://www.annualreviews.org).
A Cardinal Principle in Nutrition
The relationship between the nutrient status and a given outcome usually follows a sigmoidal curve,
which, if the nutrient becomes toxic at higher levels, becomes an inverted U shape (Figure 2). Very
simply, at low status, additional nutrient intake is beneficial; at high intake, it could be harmful; and
when at the plateau, it will have no effect. This concentration–response relationship is a cardinal
principle in nutrition, but it is often overlooked (6, 53, 82). Furthermore, it has critical implications
both for population studies and for clinical trials: In population studies, no association will be found
if most subjects’ nutrient values fall on the horizontal part of the curve. Such studies can completely
miss important effects on the ascending or descending part of the curve. Likewise, in clinical trials,
a beneficial effect of the intervention is unlikely if the nutrient status is already at optimal levels.
As a consequence, a clinical trial result might be reported as negative in a population with optimal
nutrient levels, even though a population with low nutrient status might have shown benefit.
The curve in Figure 2 is subject to important qualifications. First, the nutrient concentration
at which the plateau (maximum, or ceiling, effect) is reached may differ according to the response
or outcome being measured. Second, the curve may show a parallel shift to the left or to the
right, or the curve may have a different slope. For example, the optimum level of vitamin B12
that prevents macrocytosis may be lower than the level that prevents neuropsychiatric symptoms.
Third, different subgroups of the population may also show a parallel shift, or change in slope,
of the curve, with the implication that an adequate nutrient level for some people may not be
adequate for others. These differences reflect what Archibald Garrod in 1902 called “chemical
individuality” (40) and what Roger Williams in 1953 called “biochemical individuality” (155). Such
differences range from the extreme of Garrod’s “inborn errors of metabolism” caused by mutations
www.annualreviews.org •Homocysteine and Cognitive Impairment 213
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
100
80
60
40
20
0
Biological response
Nutrient intake or concentration (log scale)
Toxic
eect
Maximum
response
Figure 2
A cardinal principle of nutrition. The curves show the typical relationship between a nutrient and biological
outcome(s) that are influenced by the nutrient. Higher concentrations of the nutrient may be required for
some outcomes (parallel shift, red versus blue), or the slope of the relationship may change according to
interaction with genetic or environmental factors ( yellow). Much higher concentrations of the nutrient may
lead to toxic effects (dashed line).
in key genes, through common genetic polymorphisms, to the distinct genetic composition of each
individual. Garrod described this clearly:
If it be, indeed, the case that in alkaptonuria and the other conditions mentioned we are dealing with
individualities of metabolism and not with the results of morbid processes the thought naturally presents
itself that these are merely extreme examples of variations of chemical behavior which are probably
everywhere present in minor degrees and that just as no two individuals of a species are absolutely
identical in bodily structure neither are their chemical processes carried out on exactly the same lines.
(40, p. 1620)
Although it follows that the relationship between a nutrient status and an outcome will not be
identical for each person, it is common practice to use the average for a population under study.
However, the average for a population may hide important subgroups. Such subgroups may be
related to common genetic polymorphisms (2, 99, 111), race, gender, age, health status, lifestyle
factors including nutrients, and exposure to drugs. The importance of not ignoring subgroups in
relation to B vitamins is exemplified in our review.
HOMOCYSTEINE, COGNITIVE DECLINE, AND DEMENTIA
The early cross-sectional studies relating elevated tHcy or low B vitamin status to cognitive
impairment (74, 116, 121) generated the hypothesis that there is a causal link. For better evidence
of causality, data from prospective studies are essential and should include the association of tHcy
with the following outcomes:
Cognitive impairment in aging,
Conversion from cognitive impairment to dementia,
214 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
Incidence of dementia and Alzheimer’s disease (AD) in the elderly,
Increased rate of cognitive decline in AD,
Increased white matter damage in the brain,
Neurofibrillary tangles in the brain, and
The rate of whole and regional brain atrophy.
We survey some of these studies, with the main focus on prospective studies, but we do not attempt
a systematic review.
Initiation of Cognitive Impairment in Aging
In 23 cognitively normal elderly individuals, McCaddon et al. (75) found that the baseline tHcy
independently predicted final cognitive scores and the rates of cognitive decline over 5 years. In a
cohort of 1,241 French elderly individuals, there was a concentration-related association of tHcy
with declines in global cognition and attention over 4 years (27): the higher the concentration, the
faster the decline. A similar association between tHcy and declines in constructional praxis and
recall memory scores was found over 3 years in 321 elderly American men; a significant decline
was observed in those with tHcy >11 μmol/L (136). In a large Norwegian cohort (n=2,189),
tHcy was measured at baseline and after 6 years (88). A decline in tHcy over 6 years was associated
with a higher memory score assessed at follow-up, whereas an increase in tHcy was associated with
a lower memory score (Supplemental Figure 1). This study can be compared with a natural trial.
Lifestyle changes over the six years likely led to the changes in tHcy (105), a suggestion consistent
with inverse changes in serum folate levels in the same subjects (Supplemental Figure 1).
A community study in the United Kingdom assessed cognitive decline over 10 years in 691
elderly individuals (mean age 71.9 years) using a global measure, i.e., the Mini-Mental State Exam-
ination score (MMSE). The mean MMSE declined by only 1.3 points over 10 years. Nevertheless,
a doubling of the tHcy from 10 to 20 μmol/L was associated with an 88% increased rate of cogni-
tive decline (13). Another study on 889 community-dwelling elderly individuals in Sydney showed
that elevated tHcy was associated with a decline in executive function, but not in other cognitive
domains over 3 years; the estimated population attributable risk for tHcy was 9.7% (66). The
Cardiovascular Risk Factors, Aging and Dementia study in Finland on 274 dementia-free sub-
jects found that elevated baseline tHcy was associated with an increased risk of cognitive decline
in several domains 7 years later (50). The same study found a protective effect of high baseline
holotranscobalamin, which extended over the entire range of holotranscobalamin concentrations.
These results are consistent with an increased risk of cognitive impairment in elderly people
with elevated tHcy. In contrast, a study (n=516) of a multiethnic cohort in Manhattan showed
no association between incident mild cognitive impairment (MCI) and tHcy (108). However, the
authors did not consider whether the introduction of folic acid fortification, and thereby marked
improvement in folate status, during the project period could have confounded the results. More
prospective studies on the conversion from normal elderly status to MCI are needed, taking into
account baseline risk and likelihood of response to treatment.
Conversion from Cognitive Impairment to Dementia
In a memory clinic study from Sweden, the mean tHcy concentration was higher in 61 MCI
subjects who converted to dementia compared with 32 who did not convert over a 6-year period;
the odds ratio (OR) per μmol/L increase of tHcy was 1.3 for those aged 60–65 years and 1.1 for
those aged ≥65 years (4). A memory clinic study in Poland reported higher tHcy in MCI patients
who converted to AD over 3 years than in those who remained stable (38). Another Swedish
www.annualreviews.org •Homocysteine and Cognitive Impairment 215
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
memory clinic study found that elevated tHcy was among the few non-CSF (cerebrospinal fluid)
factors that were associated with conversion from MCI to AD: Out of 157 MCI patients, 57
developed AD over 4 years, and the OR per μmol/L increase in tHcy was 1.08 (P<0.005) (44). A
community study in 77 people aged >75 years with MCI found that those with stable MCI over
a 2.5-year period showed a slight fall in tHcy, whereas those who converted to AD showed an
increase in mean tHcy of almost 2 μmol/L (9).
Incidence of Dementia and Alzheimer’s Disease in a Population
A landmark study was published by Seshadri et al. (119) in 2002. This study included 1,092 elderly
participants of the Framingham cohort who were free from cognitive impairment at baseline.
There was a strong concentration-related effect of baseline tHcy, with no obvious threshold, with
the risk of incident dementia up to 11 years later. The relative risk (RR) [95% confidence interval
(CI)] for dementia after 8 years was 1.4 (1.1, 1.9) per 1 standard deviation (SD) increase in log
tHcy concentration, and the RR was even higher when confined to incident AD [RR =1.8 (1.3,
2.5)]. The estimated population attributable risk for AD of tHcy was 16%, which, if causal, would
have important public health implications.
A striking concentration-related association of plasma tHcy and of serum folate with risk of
incident dementia was also demonstrated by Ravaglia et al. (104) in a study from Italy (Figure 3).
In Mexican Americans, elevated tHcy 4.5 years before diagnosis was associated with an almost 60%
increased risk of dementia or cognitive impairment (41). In a study from New York, performed
in the period during which folic acid fortification was introduced, the risk of incident AD [hazard
ratio (HR) 2.0 (1.2, 3.5)] for elderly individuals in the top quartile of tHcy doubled, but this was no
longer significant after adjusting for age, gender, education, and APOE4 [HR 1.4 (0.8, 2.4)] (71).
043215
25
20
15
10
5
0
Cumulative incidence of dementia (%)
Duration of follow-up (y)
043215
25
20
15
10
5
0
Duration of follow-up (y)
1: <8.9
2: 8.9–11.8
3: 11.9–15.2
4: >15.2
Serum folate, quartiles: nmol L–1
4: >15.0
3: 12.6–15.0
2: 10.1–12.5
1: <10.1
Plasma tHcy, quartiles: μmol L–1
Figure 3
Cumulative incidence of dementia in 816 community-dwelling subjects (mean age 73.6 years) over 5 years,
according to quartiles of plasma total homocysteine (tHcy) or serum folate at baseline. Data are for 816
subjects; the number of incident dementia cases was 112. For tHcy from the top to the bottom quartiles,
n = 55, 21, 23, and 13 with dementia; for folate from the top to the bottom quartiles, n = 18, 18, 32, and 44
with dementia. Redrawn from Reference 104.
216 Smith ·Refsum
Erratum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
However, in the same cohort, dietary folate intake was associated with protection from incident
AD, with an HR of 0.5 (0.3, 0.9) after multiple adjustments (70).
In a Swedish study on 190 elderly individuals, those in the top tHcy tertile had double the risk
of incident AD [OR 2.64 (1.37, 5.11)] over 6.7 years compared with those in the bottom tertile
(3). A study from Finland on 271 elderly individuals found an adjusted OR for AD after 7 years of
1.19 (1.01, 1.39) per 1 μmol/L increase in baseline tHcy (49). In this latter study, baseline holo-
transcobalamin was protective against incident AD, with an adjusted OR of 0.977 (0.958–0.997)
per 1 pmol/L increase. A Japanese study with a mean follow-up of 7.3 years showed that elevated
tHcy at baseline was associated with incident dementia with an adjusted RR in the top tertile of
tHcy of 3.01 (1.33, 7.47), compared with the bottom tertile. The RR remained significant [2.50
(1.01, 6.63)] even after additional adjustments for vascular risk factors and small vessel disease (81).
Two studies are notable for their long duration. In the Framingham cohort, elevated tHcy
as long as 16 years before diagnosis was predictive of dementia incidence (119). An even more
dramatic result was found in the Prospective Population Study of Women in Gothenburg (163),
where tHcy concentrations in 1,368 women in midlife were related to incident dementia up to
35 years later. Those in the top tertile of tHcy at baseline had an HR for dementia of 1.7 (1.1, 2.6)
and an HR of 2.2 (1.2, 3.7) for AD. The Kaplan–Meier plots showed a divergence for dementia
22 years after the blood sample had been taken. Notably, the divergence became apparent at the
time when dementia prevalence increased markedly.
Overall, these prospective studies, especially the last two, make a strong case for elevated
tHcy being a likely causal risk factor for incident dementia worldwide. Three recent reviews
and meta-analyses of prospective cohort studies have examined the relation between tHcy and
incident dementia (7, 86, 145). Each of these reviews reached the conclusion that elevated tHcy is
a significant risk factor for incident dementia or cognitive deficit. In the report by Wald et al. (145),
including 8 studies with a collective 8,669 participants, the OR for a 5-μmol/L increase in tHcy
was 1.35 [95% CI 1.02, 1.79] or 1.50 (1.13, 2.00) after adjusting for regression dilution bias. The
authors estimated that a 3-μmol/L reduction in tHcy, achievable by B vitamin treatment, may lead
to a 22% risk reduction. Beydoun et al. (7) analyzed 5 studies (4 included in Wald’s meta-analysis)
on a total of 4,578 participants and obtained a pooled RR of 1.93 (1.5, 2.49) for high versus low
tHcy (different cutoffs were used) (Supplemental Figure 2). The population attributable risk for
tHcy was estimated as 21.7%. The meta-analysis by Nie et al. (86) included 14 studies with a total
of 15,908 participants with end points of incident dementia or cognitive deficit. The overall pooled
RR for hyperhomocysteinemia (variously defined) was 1.53 (1.23, 1.90), P=0.0002. Subgroup
analyses were done to check for effects of exposure period and region. The pooled RR of high
tHcy for studies from American and European countries was 1.60 (1.21, 2.13), whereas for the
rest of the world it was 1.27 (1.02, 1.59).
Based on the overall evidence, Beydoun et al. (7) concluded from their meta-analysis that
tHcy was one of the three strongest risk factors for AD: “Higher Hcy levels, lower educational
attainment, and decreased physical activity were particularly strong predictors of incident AD”
(p. 643).
Increased Rate of Cognitive Decline in Alzheimer’s Disease
To analyze the rate of cognitive decline, the use of nonlinear models may be critical because a
linear model may not reveal any effect (128), as shown for ApoE ε4(73). A nonlinear model using
a logistic S-shaped curve in 97 patients with AD in the OPTIMA (Oxford Project to Investigate
Memory and Ageing) cohort revealed a striking association between baseline tHcy concentrations
and the rate of cognitive decline [using the Cambridge Cognition Examination (CAMCOG)
www.annualreviews.org •Homocysteine and Cognitive Impairment 217
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
score]: the higher the tHcy, the faster the decline (93). The association was found only in patients
aged <75 years and was more marked at lower ages. For patients aged 65 years, with 16 years of
schooling and no previous stroke, a reduction in tHcy of 4 μmol/L was associated with a 19-month
lengthening of the time to reach half the baseline CAMCOG score (Supplemental Figure 3).
Increased White Matter Damage in the Brain
Lesions in the white matter, recognized as leukoaraiosis on CT scans and as hyperintensities on
MRI scans, are associated with an increased risk of neurological and cognitive impairment (21).
Since two early studies reported that high tHcy was associated with white matter lesions (46, 140),
there have been many confirmatory cross-sectional studies [e.g., a large study from Korea (95)] but
relatively few prospective studies. Dufouil et al. (27) found a tendency for a concentration-related
effect of tHcy over 2 years. However, the Gothenburg Women’s Study found no association over
24 years (164). A longitudinal MRI study in 663 atherosclerosis patients from the Netherlands
found a strong association over 3.9 years between hyperhomocysteinemia (>16.2 μmol/L) and
increasing volume of white matter lesions, with an OR of 2.4 (1.4, 3.9) (60). These findings were
confirmed in a clinicopathological study of >100 patients in Finland, where blood was obtained
up to 10 years before postmortem MRI scans were performed: The OR for periventricular white
matter lesions was 4.69 (1.14, 19.33) for patients in the top tertile of tHcy versus the lowest tertile
(48). A prospective study in Finland showed that the progression of white matter hyperintensities
over 6 years was related to elevated baseline tHcy, but only in those with systolic blood pressure
>150 mm Hg (47). It seems likely, therefore, that elevated tHcy is associated with damage to the
white matter, a kind of lesion that is usually considered to reflect small vessel disease, an effect
that may be limited to certain subgroups.
Density of Neurofibrillary Tangles in the Brain
Neurofibrillary tangles in the brain are a hallmark of AD. A landmark study from Finland, pub-
lished in 2013, showed that tHcy in 265 subjects aged ≥85 years was related to histopathology and
brain MRI findings after death up to 10 years later (48). A strong relationship was found between
tHcy and the density of neocortical neurofibrillary tangles: The OR for having a high tangle count
was 2.60 (1.28, 5.28) for those in the top quartile of tHcy versus the bottom quartile; likewise, the
OR for being in a more severe Braak stage was 1.96 (1.05, 3.68). The association between tHcy
and tangles was even stronger for those who had dementia (OR 3.46) or those who had cerebral
infarcts (OR 3.98). In contrast, there was no significant association of tHcy with amyloid load
or with CERAD (Consortium to Establish a Registry for Alzheimer’s Disease) score. This study
reveals some of the strongest evidence for a role of tHcy in the disease process.
Increased Rate of Atrophy in the Brain
The medial temporal lobe is atrophied in patients with AD (56), and rapid atrophy occurs as the
disease progresses (55). The first prospective study linking tHcy to an increased rate of brain
atrophy in AD was reported in the OPTIMA study in 1998 (14). The thickness of the medial
temporal lobe was measured from temporal lobe–oriented CT scans in AD patients: There was
no significant atrophy after 3 years in those individuals in the bottom baseline tertile of tHcy, but in
patients in the middle (>11.1 μmol/L) and top tertiles of tHcy, the average medial temporal lobe
thickness had decreased by 19% and 32%, respectively. Since then, several cross-sectional studies
have shown associations between tHcy and regional gray matter atrophy. These studies include
nondemented elderly individuals (24, 112, 132, 151, 154), the Framingham Offspring cohort (120),
218 Smith ·Refsum
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
poststroke patients (158), and subjects from the Alzheimer’s Disease Neuroimaging Initiative
cohort (72). In the Framingham offspring cohort with a mean age of 58 years, one SD increase in
tHcy levels was associated with the same degree of whole-brain atrophy as one year of aging (100).
Prospective studies have also shown associations of baseline tHcy with increased global or
regional atrophy rate in subjects with MCI (26, 123), in those with hypertension (35, 85), or in those
with atherosclerotic disease (57). In the Oxford-based VITACOG trial, the rate of whole-brain
atrophy in the placebo group was 0.84% per year in the bottom baseline tHcy quartile and rose to
1.52% per year in the top quartile (123). In the Swedish clinic-pathological study by Hooshmand
et al. (48), tHcy measured up to 10 years before death was strongly associated with medial temporal
lobe atrophy assessed by MRI in postmortem brains. In a study on normal elderly individuals from
the United Kingdom, investigators found that the rate of whole-brain atrophy over 5 years was
related to the entire range of baseline vitamin B12 and holotranscobalamin concentrations (141);
serum folate was not related to brain atrophy in this study. The same result for vitamin B12 and
holotranscobalamin was obtained in the Swedish normal aging cohort mentioned above (47). The
authors also reported that red blood cell folate was not related to atrophy, whereas elevated baseline
tHcy and low B12 status were associated with an increased rate of whole-brain atrophy over 6 years.
Atrophy of the white matter has been associated with tHcy in cross-sectional (33, 103) and
prospective studies (35). In the cross-sectional study by Feng et al., elevated tHcy was associated
with slower speed of central processing, but this association was lost when white matter volume
was taken into account (33). On the basis of data from the same cohort, they concluded that
elevated tHcy is associated with small vessel disease more strongly than with large vessel disease
(32). Finally, regional brain atrophy is a hallmark of AD (109); in the prospective studies where
regional atrophy was assessed (26, 35), subjects with elevated tHcy suffered from a greater rate of
regional atrophy, as previously mentioned for the medial temporal lobe (14).
HOMOCYSTEINE: MARKER OR CULPRIT?
The prospective studies reviewed above are consistent with and strongly support the view that
elevated tHcy is a risk factor for the development of cognitive impairment, brain atrophy, demen-
tia, and AD. After early cross-sectional reports, it was suggested by several commentators that
cognitively impaired people often suffered from inadequate diet (an example of reverse causality)
so that they had a poor B vitamin status. However, the prospective studies show that this is an un-
likely explanation. Elevated concentrations of tHcy are associated with dementia diagnosis, white
matter damage, brain atrophy, or neurofibrillary tangle deposition at least 5, and up to 35, years
later. Thus elevated tHcy is at least a good predictor of cognitive decline, neurodegeneration, and
dementia. At best, it is a causal factor that can easily be identified and treated using inexpensive
B vitamins. McCaddon & Miller (76) have published an excellent critical review, based on the cri-
teria of Bradford Hill, of the evidence that homocysteine is causally linked to cognitive impairment
and dementia.
Three further challenging questions remain: First, does elevated homocysteine cause cognitive
impairment, or is it just a marker for another causative factor, such as low B vitamin status, poor
lifestyle, or impaired renal function? Second, does systemic homocysteine reach high enough
concentrations in the brain to cause harm? Third, does elevated homocysteine merely potentiate
the effects of a true causative agent, or is it critically dependent on other factors?
As we consider the first question, one way to explore whether homocysteine itself is just a marker
is to include in the analysis the status of its major determinants, i.e., B vitamin status (folate, B6,
B12) and renal function (e.g., creatinine). In most, but not all, reports where this was done, the
association between tHcy and the outcome measure remains significant (see, for example, 14, 29,
www.annualreviews.org •Homocysteine and Cognitive Impairment 219
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
30, 31, 75, 77). These studies strongly suggest that elevated tHcy is an independent risk factor for
the measured outcomes and not just a marker of impaired B vitamin status or poor renal function.
The second question is related to the low concentration of tHcy in human CSF, which is about
1% of that in the blood plasma (91). The bulk of plasma tHcy being covalently bound to albumin
may partly explain the difference because the albumin concentration in CSF is about 1/200th of
that in plasma. Notably, the CSF concentration of S-adenosylhomocysteine (SAH) is similar to
that in plasma (91). Evidence that elevated tHcy in plasma can, in fact, reach the brain and, in turn,
increase brain or CSF SAH levels has been reported in the important study by Blaise et al. (8). In
this study of rats, pups from dams who were made B vitamin deficient showed a doubling of the
tHcy concentration in plasma, a 34% increase in brain tHcy, and a 120% increase in brain SAH
concentration but showed no change in S-adenosylmethionine (SAM). These changes were ac-
companied by a dramatic increase in immunoreactivity for homocysteine in several brain areas: in
the hippocampus, the proportion of CA1 pyramidal neurons positive for homocysteine increased
from 3.5% in controls to 32.7% in the offspring of mothers fed a B vitamin–deficient diet. The
immunoreactive homocysteine was localized to specific neurons (CA1 pyramidal cells) and to as-
trocytes. The hippocampal neurons positive for homocysteine immunoreactivity showed markers
of apoptosis. These results are important because the CA1 neurons in the hippocampus play a
crucial role in spatial memory. The authors suggested that the behavioral deficits in the hyperho-
mocysteinemic pups could be due to impaired spatial memory (8). This study strongly supports
the idea that peripheral hyperhomocysteinemia can directly influence neurons in the brain.
The third question, i.e., whether elevated homocysteine potentiates the effects of a causative
agent or is critically dependent on other factors, is difficult to answer because it involves interactions
between risk factors or exposure variables. For example, is the association of tHcy with cognitive
function influenced by the ApoE genotype (30, 114)? Most reports that have assessed the ApoE
genotype did not find an interaction with tHcy. However, evidence indicates that the association
of vitamin B12 with cognition is dependent on the ApoE genotype (34, 142). Also, diabetes is a
strong risk factor for AD, and, interestingly, in the Maine–Syracuse Study, the inverse relationship
between tHcy and MMSE scores was stronger in those with diabetes (110). Much further work
needs to be done to look for this type of interaction with other risk factors or exposure variables. We
suggest that some of the studies where no apparent association between tHcy and cognition was
found may be because the population did not have one or more unidentified interacting risk factors.
Results from Genetic Variations and Mendelian Randomization
Mendelian randomization uses genetic variants of known function to examine the potential effect
of a particular variable on disease outcomes. It is considered an unbiased estimate that is not con-
founded by traditional risk factors. The design markedly reduces the problem of reverse causation,
which is otherwise a severe problem in epidemiological studies. Although there are methodolog-
ical shortcomings (124), the approach represents an alternative to intervention studies. The most
recognized polymorphism in the homocysteine field is the MTHFR 677 C >T variant, which has
consistently been associated with a modest elevation of plasma tHcy and low circulating folate
(59, 65, 105). In VITACOG, there was a nonsignificant association of MTHFR 677 C >T with
brain atrophy that increased with number of T alleles (123). Many studies have been done on
this genotype and its relation to cognitive decline or dementia, but the results are inconsistent. As
originally reported for the association of MTHFR 677 C >T polymorphism with cardiovascular
disease (137), most of these studies are seriously underpowered. A large meta-analysis recently
investigated the association of the MTHFR 677 C >T genotype with dementia. It included 40
case-control studies with 4,503 AD cases and 5,767 controls (96). The authors found a significant
220 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
increased risk of AD in subjects with the CT/TT genotypes compared with the CC genotype
(summary OR =1.27; 95% CI 1.16, 1.38). Subgroup analyses suggested that the MTHFR poly-
morphism is a significant risk factor for AD in Asians and in APOE4 carriers and for late-onset
AD. Thus, this analysis adds support that lifelong exposure to modest elevation of tHcy or to low
plasma folate is causally related to AD.
POSSIBLE MECHANISMS THROUGH WHICH HOMOCYSTEINE
CAN HARM THE BRAIN
The biological plausibility of homocysteine as a risk factor for dementia has no shortage of can-
didate mechanisms. Some of the hypotheses are given in Table 1 and Figure 4.Webriefly
consider three of the postulated mechanisms: cerebrovascular effects, activation of tau kinases,
and inhibition of methylation reactions (Figure 4).
Table 1 Proposed mechanisms for the deleterious effect of homocysteine on cognitive function.
See reviews by Obeid & Herrmann (89), Smith (121), Zhuo et al. (162), McCaddon & Miller (76),
and Liu et al. (68) for additional original references
Proposed mechanism References, if applicable
Causes excitotoxic cell death by directly activating the neuronal
NMDA receptor or after formation of homocysteic acid
Generates reactive oxygen radicals (oxidative stress)
Causes formation/deposition of β-amyloid 63, 94
Potentiates the neurotoxic effect of β-amyloid on its own or via
homocysteic acid
Causes cerebral amyloid angiopathy 64
Causes NO-mediated dysfunction of the endothelium in the cerebral
vasculature
62
Causes cerebrovascular ischemia, leading to neuronal cell death
and/or tangle formation
Causes cerebrovascular ischemia, leading to small vessel disease and
white matter damage
Activates tau kinases, such as Cdk5, leading to tangle deposition
Triggers the cell cycle in neurons, leading to tangle formation and/or
cell death
84, 159
Causes DNA damage and limits DNA repair, leading to apoptosis
Increases SAH, which inhibits crucial methylation reactions, such as
DNA cytosine methylation in promotors for amyloid genes and
other epigenetic effects; inhibits PP2A activity, leading to tangle
deposition; inhibits methylation of phosphatidylethanolamine
Stimulates the endoplasmic reticulum stress response, leading to
amyloid formation
Causes a leaky blood-brain barrier
Activates the immune system
Abbreviations: NMDA, N-methyl-D-aspartate; NO, nitric oxide; PP2A, protein phosphatase-2A; SAH,
S-adenosylhomocysteine.
www.annualreviews.org •Homocysteine and Cognitive Impairment 221
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
Ischemia/reperfusion
Excitotoxicity/oxidative damage
Calcium inux
Calpain activation
P35 cleavage/activation
Cdk5 activation
Tau hyperphosphorylation
Neurobrillary tangles
Regional brain atrophy
Raised
plasma
tHcy
Raised
plasma
tHcy
β
-amyloid
formation
Increased
amyloid
gene
expression
Inhibition of DNA
methylation
Inhibition
of PP2A
Inhibition of
PP2A methylation
Inhibition of PEMT
Impaired synaptic
function
Activation of
cell cycle
Apoptosis
Lower
omega-3-rich PC
Raised
plasma
tHcy
Cognitive
impairment
NO-mediated
endothelial dysfunction
Figure 4
Some pathways
through which
homocysteine might
lead to neuronal cell
death, Alzheimer-type
pathology, and
cognitive impairment.
The blue boxes
indicate important
intermediate outcomes
that, in combination,
lead to the final
outcome of cognitive
impairment or
dementia. For
references, see Table
1and text.
Abbreviations: NO,
nitric oxide; PC,
phosphatidylcholine;
PEMT, phos-
phatidylethanolamine
methyltransferase;
PP2A, protein
phosphatase-2A; tHcy,
homocysteine.
222 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
Cerebrovascular Effects of Homocysteine
Elevated tHcy has long been known to be a risk factor for cardiovascular disease (107), so it was nat-
ural that the early reports often explained the association of tHcy with dementia as due to damage
to the cerebral vasculature rather than due to AD pathology (87). However, the first autopsy-
confirmed study showed that elevated tHcy was a risk factor not only for pathologically defined
vascular dementia [adjusted OR for top tertile versus bottom tertile: 4.5 (1.6, 12.8)], but also for
histopathologically diagnosed AD [adjusted OR for top tertile versus bottom tertile: 4.5 (2.2, 9.2)]
(14). Because AD pathology and vascular pathology often coexist in patients with dementia, it is not
straightforward to determine causal mechanisms. Almost every vascular mechanism linked to vas-
cular dementia (52, 125) has also been proposed as one of the mechanisms by which homocysteine
causes cognitive decline and dementia (42, 130). We can distinguish mechanisms that are purely
vascular from those that might, via vascular pathways, lead to brain atrophy and the histopathol-
ogy characteristics of AD. Damage to the white matter, revealed as low density (leukoaraiosis)
on CT scans or as hyperintensity on MRI, is commonly considered to reflect small vessel disease
and is associated with elevated tHcy (see section above). However, white matter damage as seen
on MRI is also associated with low-normal folate status (20) or low-normal vitamin B12 status
(19). At least for B12, the leukoaraiosis is partially reversible (143) and therefore unlikely to be
due to small vessel disease. Homocysteine may cause small vessel disease by directly damaging the
endothelium, but many other possible actions of homocysteine on vascular cells might impair the
functioning of the small vessels (42). For example, elevated tHcy in AD is associated with higher
levels of the nitric oxide (NO) synthesis inhibitor asymmetric dimethylarginine and lower levels of
NO in plasma (117). Homocysteine impairs NO-mediated endothelial vasodilatation (62), which
may lead to cerebral hypoperfusion and increased risk of dementia (133).
Animal models have been useful in the study of the effects of elevated tHcy on the cerebral vas-
culature and on brain function. These studies include animals with genetic defects in cystathionine
β-synthase or MTHFR and animals that had dietary B vitamin deficiency and/or were exposed to
diets enriched in methionine (42). A particularly elegant study is that described by Troen et al.
(135), who found that after 10 weeks on a B vitamin–deficient diet, mice displayed impairments in
spatial learning and memory that were correlated with rarefaction of the capillary vasculature in
the hippocampus and with plasma tHcy concentrations. The authors found no evidence of gliosis,
which suggests that the memory deficits were likely to be due to the vascular changes rather than
to neurodegeneration.
One way that homocysteine-induced vascular changes could lead to an AD-type pathology
is by causing transient ischemia, for instance by impairing endothelial function (25, 62). Animal
studies have shown that transient ischemia can lead to deposition of hyperphosphorylation of tau
in hippocampal neurons (149). The hyperphosphorylated tau forms intracellular filaments similar
to those found in AD, and it is striking that almost all the neurons with hyperphosphorylated tau
show markers of apoptosis (150). Activation of the tau kinase Cdk5 mediated the effect of transient
ischemia (150) (Figure 4), which is interesting in view of evidence that homocysteine can also
activate Cdk5 (see below).
Activation of Tau Kinases
The protein tau contains more than 80 putative phosphorylation sites, but the hyperphosphory-
lated tau that is deposited in neurons in AD has only certain sites occupied by phosphate residues
(146). One of the kinases that phosphorylates most of the AD-related sites in tau is Cdk5, so it
is of special interest that this kinase is activated both in vivo and in vitro by homocysteine (63).
www.annualreviews.org •Homocysteine and Cognitive Impairment 223
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
The latter authors used triple transgenic mice (3xTg) carrying genes for mutant APP, mutant
presenilin, and tau and made them deficient in dietary folate and vitamins B6 and B12. The defi-
cient mice (a) showed an 85% increase in the brain tHcy concentration compared with controls,
(b) suffered from deficits in learning and memory performance, and (c) demonstrated ∼50% in-
crease in phosphorylation of tau at two typical AD sites. The levels of Cdk5 activators were also
increased, but the levels of other known tau kinases did not change. To show that the effect of
homocysteine was direct (and not mediated by ischemia; see above), the authors exposed neuronal
cell cultures to homocysteine and found an increase in P-tau and in Cdk5 activation (63). The
above in vivo effects of homocysteine on tau phosphorylation were independent of the changes
in β-amyloid formation that were also caused by the deficient diet. It is notable that the tau ki-
nase Cdk5 (68) appears to fall on the final common pathway for a number of factors that might
link homocysteine to AD (Figure 4), including transient ischemia (see above), direct activation
by homocysteine (see above), activation of the cell cycle (84, 159), neuronal apoptosis (68), and
formation of β-amyloid (68).
Inhibition of Methylation Reactions
A particularly attractive hypothesis is that elevated tHcy leads, via elevated SAH concentrations, to
inhibition of methylation reactions (22) in which SAM is the methyl donor. This mechanism may
have wide implications in view of the large number of methylation reactions in the body. Some
methylation reactions are likely more vulnerable than others to inhibition by SAH, depending on
the relative affinities of substrate and SAH for a particular methylase. We briefly consider three
examples of inhibition of methylation that may provide mechanisms for the link with dementia.
Epigenetic regulation of gene expression in the β-amyloid pathway. Fuso and colleagues
(37) have shown that specific CpG residues in the promoter region of the presenilin gene in mice
become demethylated when the animals are hyperhomocysteinemic owing to B vitamin deficiency,
leading to increased expression of the gene. Both the demethylation and the increased expression
could be reversed by administering SAM. These experiments led the authors to propose that
elevated SAH, as a result of hyperhomocysteinemia, and/or low SAM due to reduced homocysteine
remethylation, inhibits methylation of the cytosine residues. These observations could, in part,
explain reports that hyperhomocysteinemia can lead to increased deposition of β-amyloid in
experimental animals (63, 94). It is notable that reduced levels of cytosine methylation in DNA
have been found in the hippocampus in AD and that lower levels correlated with increased amyloid
plaque load (11). Epigenetic modifications involving methylation of either DNA or histones are
widespread, and the likelihood that they can be regulated by SAH, and therefore indirectly by
homocysteine, could have major significance for the pathobiology of AD (79).
Reduced activity of protein phosphatase-2A. Protein phosphatase-2A (PP2A) is the main
enzyme responsible for dephosphorylation of phosphorylated-tau, which is the main component
of neurofibrillary tangles, a pathological hallmark of AD. Methylation of the carboxy-terminal
leucine residue of a subunit of PP2A is required to activate the enzyme. The proposed pathways
are shown in Supplemental Figure 4. As reviewed by Sontag & Sontag (126), human and animal
studies are consistent with the following scheme:
Elevated tHcy →Inhibition of leucine carboxymethyltransferase →Lower activity
of PP2A →Less dephosphorylation of P-tau →Build-up of P-tau →Neurofibrillary
tangle deposition →Regional brain atrophy →Domain-specific cognitive impairment.
224 Smith ·Refsum
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
In rats, elevated plasma tHcy leads to deposition of P-tau in the brain and to memory deficits,
both of which can be prevented by administration of folate and vitamin B12 (148, 161). In man,
elevated plasma tHcy is associated with an increase in density of neurofibrillary tangles in the
brain (48). There is a correlation between the CSF levels of tHcy and SAH, and the level of
SAH is correlated with that of CSF P-tau in humans (90, 101). Thus, the accelerated formation
of neurofibrillary tangles in the brains of those with hyperhomocysteinemia, due to activation of
the Cdk5 kinase (see above) or to reduced activity of PP2A, is one possible mechanism to link
homocysteine to AD. The increased rate of brain atrophy in hyperhomocysteinemia could also
be a consequence of tangle deposition. A link between neurofibrillary tangle density and regional
brain atrophy has been known for some time (83, 153), and the association of regional atrophy with
domain-specific cognitive deficits is well known. According to this hypothesis, lowering tHcy by
B vitamin treatment will decrease activation of the tau kinase Cdk5 and/or will relieve the inhibition
of PP2A. The overall effect will lead to reduced phosphorylation of tau and therefore prevent
deposition of tangles and the subsequent regional atrophy and cognitive impairment (26).
Impaired formation of phosphatidylcholine enriched in omega-3 fatty acids. The brain is
remarkably rich in the dietary essential polyunsaturated fatty acids, such as the omega-3 fatty
acids. These fatty acids play crucial roles not only in membrane structure, as components of
phospholipids, but also in synaptic function and signaling pathways (67, 160). Concentrations of a
key long-chain omega-3 fatty acid, docosahexaenoic acid (DHA), are lower in the brain of patients
with AD than in age-matched controls (5). Furthermore, in AD there is a relative depletion of the
molecular species of phosphatidylcholine (PC) that contain DHA and related omega-3 fatty acids in
the brain (160), erythrocytes (118), and plasma (152). These species of PC are formed mainly from
phosphatidylethanolamine (PE) by serial methylations catalyzed by phosphatidylethanolamine
methyltransferase (PEMT) (23, 102), and they are especially important in synaptic membranes
(131). Selley (118) found that elevated plasma tHcy and elevated SAH in AD patients were both
correlated with decreased erythrocyte concentrations of PC and with increased concentrations of
PE; he therefore suggested that elevated homocysteine inhibits PEMT. Furthermore, in AD, the
increases in plasma tHcy and SAH were correlated with a decline in the erythrocyte PC content of
an important omega-3 fatty acid, DHA (118). The PEMT pathway plays a key role in mobilizing
DHA from the liver into the blood and thus to the brain (5, 147, 157). Selley suggested that the
inhibition of liver PEMT by SAH is one of the causes of AD. In the VITACOG trial (see below),
we found an interaction between omega-3 fatty acid status and B vitamin treatment in relation to
brain atrophy and cognitive decline, and we suggested that lowering tHcy by B vitamin treatment
relieves the inhibition of PEMT by SAH and so permits the formation of PC enriched in omega-3
fatty acids from PE (92) (see Figure 5). Animal studies are consistent with this hypothesis. In rats
that were made hyperhomocysteinemic by folate deficiency, the brain concentration of PC was
halved and the concentration of PE tripled, implying inhibition of PEMT and thereby preventing
methylation of PE (134). The deficient animals also showed impaired learning ability. Notably,
these changes in membrane phospholipids and in behavior could be reversed by administering
methionine. The fatty acid composition of the phospholipids was not determined in this study
(134), but a study on chick embryos found that exposure to homocysteine likewise led to a decrease
in brain PC content and a rise in PE content; notably, exposure to homocysteine also led to
a decrease in the amounts of polyunsaturated fatty acids (including DHA) in the PC of brain
membranes and an increase in the saturated fatty acids (80).
The large number of plausible mechanisms through which homocysteine could cause harm
to the brain (Table 1;Figure 4) is consistent with the concept that there are multiple causal
pathways. This may be one reason why elevated tHcy is such a strong risk factor for dementia.
www.annualreviews.org •Homocysteine and Cognitive Impairment 225
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
Cystathionine
Cysteine
SAM
SAH
Hcy
Met
PEMT
B6
B6
B12
CH3THF
T
HF
Phosphatidylcholine
Phosphatidylethanolamine (rich in omega-3)
Figure 5
Interface between homocysteine (Hcy), B vitamins, and omega-3 fatty acids. Elevated total Hcy (tHcy) leads
to elevated S-adenosylhomocysteine (SAH), which inhibits the methylation of phosphatidylethanolamine to
phosphatidylcholine by phosphatidylethanolamine N-methyltransferase (PEMT). Lowering tHcy by B
vitamins releases this inhibition and allows phosphatidylcholine enriched in omega-3 fatty acids to be
synthesized. Abbreviations: Met, methionine; SAM, S-adenosylmethionine; THF, tetrahydrofolate.
WHAT HAVE CLINICAL TRIALS INVOLVING B VITAMIN
SUPPLEMENTATION TOLD US?
If homocysteine really is one of the causes of cognitive decline and dementia, then lowering tHcy
concentrations using B vitamin treatment should slow cognitive decline and eventually prevent
dementia. Many clinical trials have studied cognitive outcomes after administration of one or more
B vitamins (folic acid, vitamins B6 and B12). Furthermore, there are several meta-analyses of these
intervention trials. Thus, we do not review individual trials but focus on three larger meta-analyses
(12, 36, 144), all of which concluded that lowering homocysteine does not prevent cognitive decline
or dementia. Several reviews on the use of clinical trials and meta-analyses in nutritional research
have highlighted some of the problems we discuss (15, 17, 53, 82, 113). Following, we summarize
some critical points related to trial design:
1. What is the hypothesis being tested? Is it to see if the treatment slows cognitive decline or
prevents dementia? If so, the placebo group must show cognitive decline or have sufficient
numbers that develop dementia. Unless the placebo group declines or experiences events
(dementia), the trial can show only whether the treatment enhances cognition or causes side
effects.
2. Is the age appropriate for intervention? If cognitive decline is the outcome, the age of the
subjects should be in the range where cognitive decline and dementia occur.
3. Is the trial of sufficient duration to observe cognitive decline or dementia? In cognitively
healthy elderly individuals, MMSE declines by only 0.1 points per year, thus requiring many
years with intervention if decline in MMSE is used as an outcome measure.
226 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
4. Is the cognitive instrument used to assess a treatment effect sensitive enough to detect change
over the period of the trial? The instrument should not show ceiling or floor effects.
5. Is the intervention appropriate, i.e., does it lead to sufficient lowering of tHcy?
6. Is the group likely to respond to B vitamin intervention? One cannot expect a trial to succeed
if all the participants have a more than adequate B vitamin status or are unlikely to respond
(see A Cardinal Principle in Nutrition and Figure 2, above). Prespecified subgroup analyses
should be done to stratify for baseline values of B vitamins and tHcy.
The points above will determine if the trial has sufficient power to detect an effect on cognitive
decline or risk of dementia.
Unfortunately, most of the 26 trials analyzed in the 3 meta-analyses did not fulfill these require-
ments. Furthermore, by combining data from different trials (e.g., by using standardized mean
differences), the meta-analyses have neutralized some of these requirements that were fulfilled
in individual trials. For example, in the analysis by Wald et al. (144), 7 of the 9 trials were of
short duration (≤12 months; the two shortest lasted 1 month), with an overall median duration
of 6 months. In the trial with the largest number of participants (910), which was given a strong
weighting accordingly in the Wald meta-analysis, the daily doses of folic acid (0.2 mg) and vitamin
B12 (1 μg) were probably too low to have had much effect on tHcy. In a trial where the doses
of vitamins were adequate (78), there was no significant cognitive decline in the placebo group:
The mean MMSE at start was 29.17 and was 29.32 after 2 years. Thus, we do not consider that
the authors are justified in concluding that there was “no effect of folic acid on the prevention of
age-related cognitive decline,” given that the population did not experience decline.
A larger meta-analysis of 19 trials (including 5,398 participants), of which 8 were in the analysis
by Wald et al., was published by Ford & Almeida (36). This meta-analysis has many of the same
limitations as that described above, some of which were acknowledged by the authors in their
discussion. For example, they stated,
Given that the majority of participants did not have any pre-existing cognitive impairment and the
relative brevity of some trials, it would seem unlikely that this meta-analysis would have been adequately
powered to detect differences associated with small effect sizes (<0.2) between the supplemented and
placebo groups. (p. 146)
Another limitation is that the analysis combined data from 62 different cognitive tests; this was
achieved by converting the individual test scores to standard deviation scores to arrive at the stan-
dardized mean difference. Thus, there was considerable clinical heterogeneity in the combined
data. Furthermore, the authors assumed zero correlation between the assessments at baseline and
the study end point, which is probably a serious limitation because baseline cognitive test perfor-
mance is a major determinant of later test performance (156). Thus, in any analysis of cognitive
decline, it is imperative to use a model that controls for the baseline value; this is not possible with
the widely used standardized mean difference method. A further limitation of this method is that
it depends on the ttest, so the analysis cannot be adjusted for important covariates, such as gender
and education, which are known to influence cognitive change. Furthermore, the authors must
investigate relevant outcomes, i.e., cognitive decline or dementia. We therefore cannot agree
that these meta-analyses have shown that “homocysteine-lowering treatment does not change
the cognitive outcome of older people with and without cognitive impairment” (36, p. 146).
The most recent meta-analysis, by Clarke et al. (12), set out to review the effect of lowering
tHcy on “cognitive aging,” although the latter was not defined. The analysis considered 11 trials
on almost 22,000 subjects; of the 11 trials, 5 had been included in one or both of the above meta-
analyses. The authors specifically excluded trials on subjects with a prior diagnosis of cognitive
www.annualreviews.org •Homocysteine and Cognitive Impairment 227
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
impairment or dementia, and they did not examine these diagnoses as outcomes but considered
only cognitive status. Their main conclusion was that “homocysteine lowering by using B vitamins
had no significant effect on individual cognitive domains or global cognition or on cognitive aging”
(p. 657). This conclusion has been used by Clarke as the definitive evidence that B vitamins cannot
prevent cognitive decline or dementia (138).
This meta-analysis had several of the weaknesses described above. Furthermore, in most
subjects in the included trials, cognitive decline could not be measured because baseline global
cognitive measures were not available for 76% of 20,431 subjects. It is difficult to see what can
be concluded from such trials, apart from the lack of a cognitive-enhancing effect or side effects
of B vitamins. Two of the trials (2,825 subjects) reported cognitive scores at baseline, and both
trials showed significant effects of B vitamins on cognitive change in subjects with high tHcy (28)
or with poor B vitamin status (58); however, the results were either ignored or considered by
the authors of the meta-analyses to be “due to chance” (12). We think this attitude verges on bad
science: Instead of dismissing results as “due to chance,” one should ask, “Why are the results
different in these trials?” Further criticisms of this meta-analysis can be found in letters to The
American Journal of Clinical Nutrition (39, 122) and in the review by McCaddon & Miller (76).
Two further trials have been published since the most recent meta-analysis, one of folic acid and
vitamin B12 in 3,027 elderly individuals (139) and the other of vitamin B12 in 201 elderly individ-
uals (16). Both of these trials suffered from several of the weaknesses described above, in particular
the flaw that the placebo groups did not experience cognitive decline during the project period.
The above meta-analyses have not been helpful in answering the question of whether lowering
tHcy will slow cognitive decline, but two trials have shown clear beneficial effects of B vitamin
treatment.
The FACIT Trial
The FACIT trial was carried out on 818 subjects from the Netherlands with a mean age of 60 years
(28) who were recruited on the basis that tHcy was high, in the range of 13–26 μmol/L, but with
adequate vitamin B12 status. Folic acid (0.8 mg/day) was administered for 3 years, and at the end
of the trial tHcy concentration was 26% lower in the B vitamin intervention group compared
with the placebo group. The trial assessed several cognitive domains and found that performance
on sensorimotor speed, information processing speed, and complex speed had all declined in the
placebo group after 3 years and that the decline was slowed by folic acid treatment. The placebo
group showed improvement in memory performance (learning effect), but a greater degree of
improvement occurred in the folic acid group. Global cognitive performance also improved more
in the folic acid group. Notably, there was a larger beneficial effect of folic acid on information
processing speed in those whose baseline tHcy was above the median than in those below the
median. The authors estimated that folic acid treatment gives an individual the performance of
someone who is 4.7 years younger for memory, 1.7 years younger for sensorimotor speed, 2.1 years
younger for information processing speed, and 1.5 years younger for global cognitive function.
Overall, this well-designed trial is fully consistent with the view that lowering tHcy may slow
decline in those cognitive domains assessed by appropriate tests that are sensitive to aging.
The VITACOG Trial and the Importance of Identifying the Groups
that Are Likely to Respond
The VITACOG trial, from the University of Oxford, was designed to see if lowering tHcy would
slow the accelerated rate of brain atrophy that occurs in MCI (123). Volunteers with MCI were
228 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
B vitamins
Placebo
RL L
Figure 6
VITACOG trial in subjects with mild cognitive impairment. Regions with significant loss of gray matter
over a two-year period are shown in yellow. The volume of gray matter loss was sevenfold lower in subjects
treated with B vitamins compared with placebo. From Reference 26.
randomized to placebo or a daily dose of a combination of folic acid (0.8 mg), vitamin B12 (0.5 mg),
and vitamin B6 (20 mg) for 2 years. Volumetric MRI scans were performed at the start and end of
the trial (n = 168). The B vitamin treatment led to a highly significant 30% slowing in the rate
of global brain atrophy. Notably, the effect of B vitamins depended on the baseline tHcy, with a
53% slowing of atrophy in those in the top quartile (>13 μmol/L), but no significant effect was
shown in those in the bottom 2 quartiles (≤11. 3 μmol/L), as shown in Supplemental Figure 5.
The VITACOG trial was not powered to detect any effect of homocysteine lowering on cog-
nition. Nevertheless, a prespecified analysis stratifying for tHcy, using models that controlled for
baseline cognition values, showed that cognitive decline was virtually prevented by B vitamins in
those whose baseline tHcy was >11.3 μmol/L in the following domains: episodic memory, seman-
tic memory, and global cognition (MMSE) (18). The results for MMSE and semantic memory
are shown in Supplemental Figure 6. In addition, the B vitamin treatment slowed decline in
executive function independent of tHcy. A striking effect on clinical measures was also found:
For those in the top quartile of tHcy (>13 μmol/L), B vitamin treatment had beneficial effects
on the clinical dementia rating (CDR) and the IQCode scores. For example, the proportion of
subjects who reverted to a global CDR score of zero (i.e., normal) doubled in the B vitamin group
compared with placebo.
Further analysis of the VITACOG scans revealed strong effects of B vitamin treatment on
the regional brain atrophy in MCI (26). Sevenfold less regional brain atrophy was noted in the
B vitamin group compared with the placebo group (Figure 6). This effect of B vitamins was
significant only in those with tHcy above the median.
The gray matter regions of the brain protected by B vitamin treatment included structures
of the medial temporal lobe, the precuneus, the angular gyrus, and the supramarginal gyrus, i.e.,
regions that are vulnerable to the AD disease process. The authors concluded that the B vitamin
treatment had greatly slowed the atrophy of AD-related regions of the brain. Causal relationships
were suggested (26) using a Bayesian network analysis consistent with the following causal chain
www.annualreviews.org •Homocysteine and Cognitive Impairment 229
Supplemental Material
Erratum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
of events:
B vitamins →lower tHcy →slows atrophy rate →slows cognitive decline.
Thus, it appears that B vitamin treatment slows the disease process of AD in people with elevated
tHcy. The VITACOG trial provides some of the best evidence for a causal role of elevated
homocysteine or low B vitamins in brain atrophy and cognitive decline in people with MCI.
Further trials are needed to see if B vitamin treatment will also slow the conversion to dementia.
Importance of the baseline homocysteine concentration. The VITACOG trial implies a
threshold effect of tHcy on biological outcomes such as brain atrophy and cognition: Slowing of
brain atrophy and of cognitive decline occurred only at tHcy >11 μmol/L, whereas improvement
of clinical measures by B vitamin intervention was found only at tHcy >13 μmol/L. A similar
result was obtained in a Chinese trial in mild to moderate AD, where only subjects with tHcy
>13 μmol/L showed cognitive benefit from B vitamin treatment (61). These findings are also
consistent with previous observations. For example, as shown in Figure 3 above, there is a
concentration-related (>10 μmol/L) association of tHcy with the incidence of dementia. In
the OPTIMA cohort, only concentrations >11 μmol/L were associated with an increased rate
of atrophy of the medial temporal lobe (14); likewise, at concentrations >10 μmol/L, there
was a concentration-dependent increase in the rate of cognitive decline in AD patients under
75 years (93) (see also Supplemental Figure 3). If the threshold for effects of tHcy is ∼10–
11 μmol/L, then it could explain why some studies, for example in countries with mandatory folic
acid fortification, do not find associations of tHcy with cognitive or brain outcome measures.
Thus, an 18-month trial of high-dose B vitamins in patients with AD in the United States
did not find an overall effect of treatment on cognitive decline (1), but the baseline tHcy was
9.16 μmol/L. However, in this trial, patients with a CDR score of 0.5 at baseline, i.e., with mild
AD, experienced a protective effect of B vitamins (Figure 7), which suggests that timing of the
intervention is critical. The lack of clinical or cognitive effect in more severe dementia cases is
not surprising because the brain is already extensively and irreversibly damaged.
Interactions with other risk factors. In the Homocysteine: Marker or Culprit? section above,
we have mentioned some other risk factors that might interact with homocysteine. A post hoc
analysis of the VITACOG trial data has revealed an interaction with omega-3 fatty acids: The
protective effect of B vitamin treatment on both brain atrophy and cognitive decline in MCI
occurred only in those with a good status of long-chain omega-3 fatty acids (54, 92). A possible
explanation for the interaction between homocysteine and omega-3 fatty acids has been described
above in the Mechanisms section (Figure 5). We also observed that the beneficial effect of B
vitamins on brain atrophy was absent in subjects taking aspirin (123), consistent with results in
a subgroup analysis according to antiplatelet use in the cardiovascular trials (43). In addition to
those with elevated tHcy, one other group in the VITACOG trial responded particularly well to
B vitamins, namely those with a history of stroke or transient ischemic attacks, which supports
the potential importance of vascular mechanisms. Although high creatinine was associated with
a faster brain atrophy rate, a beneficial effect of B vitamins was observed in those with creatinine
above and those below the median and, if anything, was stronger in those with low creatinine.
This observation strongly suggests that the B vitamin effect is not due to interaction with renal
impairment.
Overall, these results emphasize the importance of taking exposure variables and risk factors
into account. It is also a likely reason why some trials of B vitamins have failed, e.g., in populations
230 Smith ·Refsum
Supplemental Material
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
2
1
0
–1
–2
–3
–4
Change in MMSE score
0 3 6 9 12 15 18
Months since start of trial
B vitamins
Placebo
Figure 7
Effect of B vitamin treatment on patients with mild Alzheimer’s disease (AD) in the VITAL trial. In the
patients who had a clinical dementia rating (CDR) score of 0.5 at baseline (mild AD), B vitamin treatment
(folic acid, vitamins B6 and B12) significantly slowed cognitive decline over the 18-month period of the trial
(general estimating equations, P=0.017, unadjusted for multiple comparisons). No significant effect was
found in the total trial participants, which included those with moderate as well as mild AD (see Reference 1).
Unpublished results kindly provided by R. Diaz-Arrastia and P. Aisen. Abbreviation: MMSE, Mini-Mental
State Examination.
with already low tHcy or in those taking aspirin, as in many trials included in the described
meta-analysis (12). Furthermore, populations may have had a poor omega-3 fatty acid status.
On the basis of the evidence from the VITACOG trial, long-term trials of a combination of
B vitamins and omega-3 fatty acids in older people at risk would be well worthwhile. These trials
should focus on groups who are likely to respond during the projected trial period. As a side note,
a recent small pilot trial of a nutrient mixture that included omega-3 fatty acids, 1 mg folic acid,
and 20 μg B12 was reported to show cognitive benefits in 29 elderly women (129).
POSSIBLE PUBLIC HEALTH IMPLICATIONS
The estimate that ∼20% of dementia cases may be related to elevated tHcy concentrations (7),
that tHcy is a marker of poor lifestyle (105) and is readily lowered by B vitamin supplementation,
has potential public health implications. tHcy levels likely have to exceed a threshold of ∼10–
11 μmol/L before they are associated with harm (14, 104, 123). However, 10–11 μmol/L is below
reference ranges (106), which suggests that a large part of the population may be at risk, even in
countries that have introduced folic acid fortification. The implication is that tHcy concentrations
in the elderly population should ideally be kept at <10 μmol/L. The potential benefits will depend
on tHcy population levels, strongly determined by environmental factors, including mandatory
folic acid fortification. The NHANES studies in the United States show that fortification of flour
with folic acid markedly reduced the proportion of the population with tHcy >13 μmol/L from
about 25% of the population to 10%, but there was only a modest decrease of 10% in the median
concentration to 7.7 μmol/L (98). The sector of the population at risk for dementia is people
aged >60 years. From the percentile distributions, we can conclude that ∼30% of those aged
>60 years have tHcy concentrations >10 μmol/L (97); this corresponds to ∼17 million people
in the United States who may benefit from lowering their homocysteine levels. In countries that
have not fortified with folic acid, a greater proportion of the elderly population will have tHcy
www.annualreviews.org •Homocysteine and Cognitive Impairment 231
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
>10 μmol/L. Thus, in Oxford, for example, a relatively wealthy and healthy part of the United
Kingdom, the median tHcy was 11.3 μmol/L (123), whereas in nearby Banbury, the median was
14 μmol/L, with >75% having higher than 11 μmol/L (45). In Ireland, which has a liberal voluntary
fortification policy, the median in people aged >65 years was 13.2 μmol/L (51).
Overall, the evidence strongly suggests that elevated tHcy is associated with cognitive decline
and dementia owing to neurodegeneration. There is ample evidence that homocysteine may in-
terfere with critical pathways associated with neuronal death and cognitive decline. Some trials
strongly support the view that B vitamins may be beneficial against cognitive decline; however,
most trials and all the meta-analyses have failed to address the question adequately by not includ-
ing subjects likely to respond during the trial period. Furthermore, they have not used relevant
outcome measures, i.e., cognitive decline or dementia. Thus, we need additional intervention trials
to test whether lowering tHcy will slow or prevent the conversion from cognitively healthy or
MCI to dementia in high-risk groups. In the meantime, a policy practiced in Swedish memory
clinics should be considered (69): All patients with memory problems are routinely screened for
tHcy and, in those with elevated tHcy concentrations, B vitamin supplements are recommended.
Given the high risk of dementia in susceptible populations, combined with the low prevalence of
adverse effects of B vitamins, supplementation is a pragmatic solution while waiting for additional
trial evidence.
DISCLOSURE STATEMENT
A.D.S. is named as inventor on four patents held by the University of Oxford on the use of B
vitamins to treat Alzheimer’s disease or mild cognitive impairment (US6008221, US6127370,
PCT/GB2010/051557, and WO 2015/140545). H.R. is named as inventor on two patents
held by the University of Oxford on the use of B vitamins to treat mild cognitive impairment
(PCT/GB2010/051557 and WO 2015/140545).
LITERATURE CITED
1. Aisen PS, Schneider LS, Sano M, Diaz-Arrastia R, van Dyck CH, et al. 2008. High-dose B vitamin supple-
mentation and cognitive decline in Alzheimer disease: a randomized controlled trial. JAMA 300:1774–83
2. Ames BN, Elson-Schwab I, Silver EA. 2002. High-dose vitamin therapy stimulates variant enzymes with
decreased coenzyme binding affinity (increased Km): relevance to genetic disease and polymorphisms.
Am. J. Clin. Nutr. 75:616–58
3. Annerbo S, Kivipelto M, L¨
okk J. 2009. A prospective study on the development of Alzheimer’s disease
with regard to thyroid-stimulating hormone and homocysteine. Dement. Geriatr. Cogn. Disord. 28:275–80
4. Annerbo S, Wahlund LO, L¨
okk J. 2006. The significance of thyroid-stimulating hormone and homo-
cysteine in the development of Alzheimer’s disease in mild cognitive impairment: a 6-year follow-up
study. Am. J. Alzheimers Dis. Other Demen. 21:182–88
5. Astarita G, Jung KM, Berchtold NC, Nguyen VQ, Gillen DL, et al. 2010. Deficient liver biosynthesis of
docosahexaenoic acid correlates with cognitive impairment in Alzheimer’s disease. PLOS ONE 5:e12538
6. Barnes JL, Tian M, Edens NK, Morris MC. 2014. Consideration of nutrient levels in studies of cognitive
decline. Nutr. Rev. 72:707–19
7. Beydoun MA, Beydoun HA, Gamaldo AA, Teel A, Zonderman AB, Wang Y. 2014. Epidemiologic studies
of modifiable factors associated with cognition and dementia: systematic review and meta-analysis. BMC
Public Health 14:643
8. Blaise SA, N´
ed´
elec E, Schroeder H, Alberto JM, Bossenmeyer-Pouri´
e C, et al. 2007. Gestational vitamin
B deficiency leads to homocysteine-associated brain apoptosis and alters neurobehavioral development
in rats. Am. J. Pathol. 170:667–79
232 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
9. Blasko I, Jellinger K, Kemmler G, Krampla W, Jungwirth S, et al. 2008. Conversion from cognitive
health to mild cognitive impairment and Alzheimer’s disease: prediction by plasma amyloid beta 42,
medial temporal lobe atrophy and homocysteine. Neurobiol. Aging 29:1–11
10. Carmel R, Jacobsen DW. 2001. Homocysteine in Health and Disease. Cambridge, UK: Cambridge Univ.
Press
11. Chouliaras L, Mastroeni D, Delvaux E, Grover A, Kenis G, et al. 2013. Consistent decrease in global
DNA methylation and hydroxymethylation in the hippocampus of Alzheimer’s disease patients. Neurobiol.
Aging 34:2091–99
12. Clarke R, Bennett D, Parish S, Lewington S, Skeaff M, et al. 2014. Effects of homocysteine lowering
with B vitamins on cognitive aging: meta-analysis of 11 trials with cognitive data on 22,000 individuals.
Am. J. Clin. Nutr. 100:657–66
13. Clarke R, Birks J, Nexo E, Ueland PM, Schneede J, et al. 2007. Low vitamin B-12 status and risk of
cognitive decline in older adults. Am. J. Clin. Nutr. 86:1384–91
14. Clarke R, Smith AD, Jobst KA, Refsum H, Sutton L, Ueland PM. 1998. Folate, vitamin B12, and serum
total homocysteine levels in confirmed Alzheimer disease. Arch. Neurol. 55:1449–55
15. Dacks PA, Bennett DA, Fillit HM. 2014. Evidence needs to be translated, whether or not it is complete.
JAMA Neurol. 71:137–38
16. Dangour AD, Allen E, Clarke R, Elbourne D, Fletcher AE, et al. 2015. Effects of vitamin B-12 supple-
mentation on neurologic and cognitive function in older people: a randomized controlled trial. Am. J.
Clin. Nutr. 102:639–47
17. de Jager CA, Dye L, de Bruin EA, Butler L, Fletcher J, et al. 2014. Criteria for validation and selection
of cognitive tests for investigating the effects of foods and nutrients. Nutr. Rev. 72:162–79
18. de Jager CA, Oulhaj A, Jacoby R, Refsum H, Smith AD. 2012. Cognitive and clinical outcomes of
homocysteine-lowering B-vitamin treatment in mild cognitive impairment: a randomized controlled
trial. Int. J. Geriatr. Psychiatry 27:592–600
19. de Lau L, Smith AD, Refsum H, Johnston C, Breteler MM. 2009. Plasma vitamin B12 status and cerebral
white-matter lesions. J. Neurol. Neurosurg. Psychiatry 80:149–57
20. de Lau LML, Refsum H, Smith AD, Johnston C, Breteler MB. 2007. Plasma folate concentration and
cognitive performance: Rotterdam Scan Study. Am. J. Clin. Nutr. 86:728–34
21. Debette S, Markus HS. 2010. The clinical importance of white matter hyperintensities on brain magnetic
resonance imaging: systematic review and meta-analysis. BMJ 341:c3666
22. Deguchi T, Barchas J. 1971. Inhibition of transmethylations of biogenic amines by S-adenosylhomo-
cysteine. Enhancement of transmethylation by adenosylhomocysteinase. J. Biol. Chem. 246:3175–81
23. DeLong CJ, Shen YJ, Thomas MJ, Cui Z. 1999. Molecular distinction of phosphatidylcholine synthesis
between the CDP-choline pathway and phosphatidylethanolamine methylation pathway. J. Biol. Chem.
274:29683–88
24. den Heijer T, Vermeer SE, Clarke R, Oudkerk M, Koudstaal PJ, et al. 2003. Homocysteine and brain
atrophy on MRI of non-demented elderly. Brain 126:170–75
25. Di Marco LY, Venneri A, Farkas E, Evans PC, Marzo A, Frangi AF. 2015. Vascular dysfunction in
the pathogenesis of Alzheimer’s disease—a review of endothelium-mediated mechanisms and ensuing
vicious circles. Neurobiol. Dis. 82:593–606
26. Douaud G, Refsum H, de Jager CA, Jacoby R, Nichols TE, et al. 2013. Preventing Alzheimer’s disease-
related gray matter atrophy by B-vitamin treatment. PNAS 110:9523–28
27. Dufouil C, Alperovitch A, Ducros V, Tzourio C. 2003. Homocysteine, white matter hyperintensities,
and cognition in healthy elderly people. Ann. Neurol. 53:214–21
28. Durga J, van Boxtel MP, Schouten EG, Kok FJ, Jolles J, et al. 2007. Effect of 3-year folic acid supplemen-
tation on cognitive function in older adults in the FACIT trial: a randomised, double blind, controlled
trial. Lancet 369:208–16
29. Elias MF, Robbins MA, Budge MM, Elias PK, Brennan SL, et al. 2006. Homocysteine, folate, and
vitamins B6and B12 blood levels in relation to cognitive performance: The Maine-Syracuse study.
Psychosom. Med. 68:547–54
30. Elias MF, Robbins MA, Budge MM, Elias PK, Dore GA, et al. 2008. Homocysteine and cognitive
performance: modification by the ApoE genotype. Neurosci. Lett. 430:64–69
www.annualreviews.org •Homocysteine and Cognitive Impairment 233
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
31. Elias MF, Sullivan LM, D’Agostino RB, Elias PK, Jacques PF, et al. 2005. Homocysteine and cognitive
performance in the Framingham offspring study: age is important. Am. J. Epidemiol. 162:644–53
32. Feng C, Bai X, Xu Y, Hua T, Huang J, Liu X-Y. 2013. Hyperhomocysteinemia associates with small
vessel disease more closely than large vessel disease. Int. J. Med. Sci. 10:408–12
33. Feng L, Isaac V, Sim S, Ng TP, Krishnan KR, Chee MW. 2013. Associations between elevated homo-
cysteine, cognitive impairment, and reduced white matter volume in healthy old adults. Am.J.Geriatr.
Psychiatry 21:164–72
34. Feng L, Li J, Yap KB, Kua EH, Ng TP. 2009. Vitamin B-12, apolipoprotein E genotype, and cognitive
performance in community-living older adults: evidence of a gene-micronutrient interaction. Am. J.
Clin. Nutr. 89:1263–68
35. Firbank MJ, Narayan SK, Saxby BK, Ford GA, O’Brien JT. 2010. Homocysteine is associated with hip-
pocampal and white matter atrophy in older subjects with mild hypertension. Int. Psychogeriatr. 22:804–11
36. Ford AH, Almeida OP. 2012. Effect of homocysteine lowering treatment on cognitive function: a sys-
tematic review and meta-analysis of randomized controlled trials. J. Alzheimers Dis. 29:133–49
37. Fuso A, Nicolia V, Pasqualato A, Fiorenza MT, Cavallaro RA, Scarpa S. 2011. Changes in Presenilin 1
gene methylation pattern in diet-induced B vitamin deficiency. Neurobiol. Aging 32:187–99
38. Gabryelewicz T, Styczynska M, Luczywek E, Barczak A, Pfeffer A, et al. 2007. The rate of conversion of
mild cognitive impairment to dementia: predictive role of depression. Int. J. Geriatr. Psychiatry 22:563–67
39. Garrard P, Jacoby R. 2015. B-vitamin trials meta-analysis: less than meets the eye. Am. J. Clin. Nutr.
101:414–15
40. Garrod AE. 1902. The incidence of alkaptonuria: a study in chemical individuality. Lancet 160:1616–20
41. Haan MN, Miller JW, Aiello AE, Whitmer RA, Jagust WJ, et al. 2007. Homocysteine, B vitamins, and
the incidence of dementia and cognitive impairment: results from the Sacramento Area Latino Study on
Aging. Am. J. Clin. Nutr. 85:511–17
42. Hainsworth A, Yeo NE, Weekman EM, Wilcock DM. 2016. Homocysteine, hyperhomocysteinemia and
vascular contributions to cognitive impairment and dementia (VCID). Biochem. Biophys. Acta 1862:1008–
17
43. Hankey GJ, Eikelboom JW, Yi Q, Lees KR, Chen C, et al. 2012. Antiplatelet therapy and the effects
of B vitamins in patients with previous stroke or transient ischaemic attack: a post-hoc subanalysis of
VITATOPS, a randomised, placebo-controlled trial. Lancet Neurol. 11:512–20
44. Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. 2006. Association between
CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-
up study. Lancet Neurol. 5:228–34
45. Hin H, Clarke R, Sherliker P, Atoyebi W, Emmens K, et al. 2006. Clinical relevance of low serum
vitamin B12 concentrations in older people: the Banbury B12 study. Age Ageing 35:416–22
46. Hogervorst E, Ribeiro HM, Molyneux A, Budge M, Smith AD. 2002. Plasma homocysteine levels,
cerebrovascular risk factors, and cerebral white matter changes (leukoaraiosis) in patients with Alzheimer
disease. Arch. Neurol. 59:787–93
47. Hooshmand B, Mangialasche F, Kalpouzos G, Solomon A, Ka˚reholt A, et al. 2016. Vitamin B12, folate
and sulphur amino acids in relation to markers of brain aging: a longitudinal population-based study.
JAMA Psychiatry. In press
48. Hooshmand B, Polvikoski T, Kivipelto M, Tanskanen M, Myllykangas L, et al. 2013. Plasma homocys-
teine, Alzheimer and cerebrovascular pathology: a population-based autopsy study. Brain 136:2707–16
49. Hooshmand B, Solomon A, Ka˚ reholt I, Leivisk¨
a J, Rusanen M, et al. 2010. Homocysteine and holo-
transcobalamin and the risk of Alzheimer disease: a longitudinal study. Neurology 75:1408–14
50. Hooshmand B, Solomon A, Ka˚reholt I, Rusanen M, H ¨
anninen T, et al. 2012. Associations between
serum homocysteine, holotranscobalamin, folate and cognition in the elderly: a longitudinal study.
J. Intern. Med. 271:204–12
51. Hopkins SM, Gibney MJ, Nugent AP, McNulty H, Molloy AM, et al. 2015. Impact of voluntary forti-
fication and supplement use on dietary intakes and biomarker status of folate and vitamin B-12 in Irish
adults. Am. J. Clin. Nutr. 101:1163–72
52. Iadecola C. 2013. The pathobiology of vascular dementia. Neuron 80:844–66
234 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
53. James MJ, Sullivan TR, Metcalf RG, Cleland LG. 2014. Pitfalls in the use of randomised controlled
trials for fish oil studies with cardiac patients. Br.J.Nutr.112:812–20
54. Jerner´
en F, Elshorbagy AK, Oulhaj A, Smith SM, Refsum H, Smith AD. 2015. Brain atrophy in cogni-
tively impaired elderly: the importance of long-chain ω-3 fatty acids and B vitamin status in a randomized
controlled trial. Am. J. Clin. Nutr. 102:215–21
55. Jobst KA, Smith AD, Szatmari M, Esiri MM, Jaskowski A, et al. 1994. Rapidly progressing atrophy of
medial temporal lobe in Alzheimer’s disease. Lancet 343:829–30
56. Jobst KA, Smith AD, Szatmari M, Molyneux A, Esiri MM, et al. 1992. Detection in life of confirmed
Alzheimer’s disease using a simple measurement of medial temporal lobe atrophy by computed tomog-
raphy. Lancet 340:1179–83
57. Jochemsen HM, Kloppenborg RP, de Groot LC, Kampman E, Mali WP, et al. 2013. Homocysteine,
progression of ventricular enlargement, and cognitive decline: the Second Manifestations of ARTerial
disease-Magnetic Resonance study. Alzheimers Dement. 9:302–9
58. Kang JH, Cook N, Manson J, Buring JE, Albert CM, Grodstein F. 2008. A trial of B vitamins and
cognitive function among women at high risk of cardiovascular disease. Am. J. Clin. Nutr. 88:1602–10
59. Klerk M, Verhoef P, Clarke R, Blom HJ, Kok FJ, Schouten EG. 2002. MTHFR 677C→T polymorphism
and risk of coronary heart disease: a meta-analysis. JAMA 288:2023–31
60. Kloppenborg RP, Geerlings MI, Visseren FL, Mali WP, Vermeulen M, et al. 2014. Homocysteine and
progression of generalized small-vessel disease: the SMART-MR Study. Neurology 82:777–83
61. Kwok T, Lee J, Law CB, Pan PC, Yung CY, et al. 2011. A randomized placebo controlled trial of
homocysteine lowering to reduce cognitive decline in older demented people. Clin. Nutr. 30:297–302
62. Lai WK, Kan MY. 2015. Homocysteine induced endothelial dysfunction. Ann. Nutr. Metab. 67:1–12
63. Li J-G, Chu J, Barrero C, Merali S, Pratic`
o D. 2014. Homocysteine exacerbates β-amyloid,tau pathology,
and cognitive deficit in a mouse model of Alzheimer’s disease with plaques and tangles. Ann. Neurol.
75:851–63
64. Li J-G, Pratic `
o D. 2015. High levels of homocysteine results in cerebral amyloid angiopathy in mice.
J. Alzheimers Dis. 43:29–35
65. Liew SC, Gupta ED. 2015. Methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism:
epidemiology, metabolism and the associated diseases. Eur. J. Med. Genet. 58:1–10
66. Lipnicki DM, Sachdev PS, Crawford J, Reppermund S, Kochan NA, et al. 2013. Risk factors for late-life
cognitive decline and variation with age and sex in the Sydney Memory and Ageing Study. PLOS ONE
8:e65841
67. Liu JJ, Green P, Mann JJ, Rapoport SI, Sublette ME. 2015. Pathways of polyunsaturated fatty
acid utilization: implications for brain function in neuropsychiatric health and disease. Brain Res.
1597:220–46
68. Liu SL, Wang C, Jiang T, Tan L, Xing A, Yu JT. 2016. The role of Cdk5 in Alzheimer’s disease. Mol.
Neurobiol. In press
69. L¨
okk J. 2013. B-vitaminer kan pr ¨
ovas vid kognitiv svikt. Lakartidningen 110:1528
70. Luchsinger JA, Tang MX, Miller J, Green R, Mayeux R. 2007. Relation of higher folate intake to lower
risk of Alzheimer disease in the elderly. Arch. Neurol. 64:86–92
71. Luchsinger JA, Tang MX, Shea S, Miller J, Green R, Mayeux R. 2004. Plasma homocysteine levels and
risk of Alzheimer disease. Neurology 62:1972–76
72. Madsen SK, Rajagopalan P, Joshi SH, Toga AW, Thompson PM, Alzheimer’s Disease Neuroimaging
I. 2015. Higher homocysteine associated with thinner cortical gray matter in 803 participants from the
Alzheimer’s Disease Neuroimaging Initiative. Neurobiol. Aging 36(Suppl. 1):S203–10
73. Martins CA, Oulhaj A, de Jager CA, Williams JH. 2005. APOE alleles predict the rate of cognitive
decline in Alzheimer disease: a nonlinear model. Neurology 65:1888–93
74. McCaddon A. 2006. Homocysteine and cognition—a historical perspective. J. Alzheimers Dis. 9:361–80
75. McCaddon A, Hudson P, Davies G, Hughes A, Williams JH, Wilkinson C. 2001. Homocysteine and
cognitive decline in healthy elderly. Dement. Geriatr. Cogn. Disord. 12:309–13
76. McCaddon A, Miller JW. 2015. Assessing the association between homocysteine and cognition: reflec-
tions on Bradford Hill, meta-analyses and causality. Nutr. Rev. 73:723–35
www.annualreviews.org •Homocysteine and Cognitive Impairment 235
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
77. McIlroy SP, Dynan KB, Lawson JT, Patterson CC, Passmore AP. 2002. Moderately elevated plasma
homocysteine, methylenetetrahydrofolate reductase genotype, and risk for stroke, vascular dementia,
and Alzheimer disease in Northern Ireland. Stroke 33:2351–56
78. McMahon JA, Green TJ, Skeaff CM, Knight RG, Mann JI, Williams SM. 2006. A controlled trial of
homocysteine lowering and cognitive performance. N. Engl. J. Med. 354:2764–72
79. Millan MJ. 2014. The epigenetic dimension of Alzheimer’s disease: causal, consequence, or curiosity?
Dialogues Clin. Neurosci. 16:373–93
80. Miller RR, Leanza CM, Phillips EE, Blacquire KD. 2003. Homocysteine-induced changes in brain
membrane composition correlate with increased brain caspase-3 activities and reduced chick embryo
viability. Comp. Biochem. Physiol. B Biochem. Mol. Biol. 136:521–32
81. Miwa K, Tanaka M, Okazaki S, Yagita Y, Sakaguchi M, et al. 2015. Increased total homocysteine levels
predict the risk of incident dementia independent of cerebral small-vessel diseases and vascular risk
factors. J. Alzheimers Dis. 49:503–13
82. Morris MC, Tangney CC. 2011. A potential design flaw of randomized trials of vitamin supplements.
JAMA 305:1348–49
83. Nagy Z, Jobst KA, Esiri MM, Morris JH, King EM-F, et al. 1996. Hippocampal pathology reflects
memory deficit and brain imaging measurements in Alzheimer’s disease: clinicopathologic correlations
using three sets of pathologic diagnostic criteria. Dementia 7:76–81
84. Nagy Z, Smith MZ, Esiri MM, Barnetson L, Smith AD. 2000. Hyperhomocysteinaemia in Alzheimer’s
disease and expression of cell cycle markers in the brain. J. Neurol. Neurosurg. Psychiatry 69:565–66
85. Narayan SK, Firbank MJ, Saxby BK, Stansby G, Hansrani M, et al. 2011. Elevated plasma homocysteine
is associated with increased brain atrophy rates in older subjects with mild hypertension. Dement. Geriatr.
Cogn. Disord. 31:341–48
86. Nie T, Lu T, Xie L, Huang P, Lu Y, Jiang M. 2014. Hyperhomocysteinemia and risk of cognitive decline:
a meta-analysis of prospective cohort studies. Eur. Neurol. 72:241–48
87. Nilsson K, Gustafson L, F¨
aldt R, Andersson A, Brattstr ¨
om L, et al. 1996. Hyperhomocysteinaemia—a
common finding in a psychogeriatric population. Eur. J. Clin. Investig. 26:853–59
88. Nurk E, Refsum H, Tell GS, Engedal K, Vollset SE, et al. 2005. Plasma total homocysteine and memory
in the elderly: the Hordaland Homocysteine study. Ann. Neurol. 58:847–57
89. Obeid R, Herrmann W. 2006. Mechanisms of homocysteine neurotoxicity in neurodegenerative diseases
with special reference to dementia. FEBS Lett. 580:2994–3005
90. Obeid R, Kasoha M, Knapp JP, Kostopoulos P, Becker G, et al. 2007. Folate and methylation status in
relation to phosphorylated Tau protein(181P) and β-amyloid(1–42) in cerebrospinal fluid. Clin. Chem.
53:1129–36
91. Obeid R, Kostopoulos P, Knapp JP, Kasoha M, Becker G, et al. 2007. Biomarkers of folate and vitamin
B12 are related in blood and cerebrospinal fluid. Clin. Chem. 53:326–33
92. Oulhaj A, Jerner´
en F, Refsum H, Smith AD, de Jager CA. 2016. Omega-3 fatty acid status enhances the
prevention of cognitive decline by B vitamins in mild cognitive impairment. J. Alzheimers Dis. 50:547–57
93. Oulhaj A, Refsum H, Beaumont H, Williams J, King E, et al. 2010. Homocysteine as a predictor of
cognitive decline in Alzheimer’s disease. Int. J. Geriatr. Psychiatry 25:82–90
94. Pacheco-Quinto J, Rodriguez de Turco EB, Derosa S, Howard A, Cruz-Sanchez F, et al. 2006. Hyper-
homocysteinemic Alzheimer’s mouse model of amyloidosis shows increased brain amyloid beta peptide
levels. Neurobiol. Dis. 22:651–56
95. Park SY, An SA, Lee HB, Kim Y, Kim NK, et al. 2013. Different impact of hyperhomocysteinemia on
cerebral small vessel ischemia and cervico-cerebral atherosclerosis in non-stroke individuals. Thromb.
Res. 131:e12–16
96. Peng Q, Lao X, Huang X, Qin X, Li S, Zeng Z. 2015. The MTHFR C677T polymorphism contributes to
increased risk of Alzheimer’s disease: evidence based on 40 case-control studies. Neurosci. Lett. 586:36–42
97. Pfeiffer CM, Caudill SP, Gunter EW, Osterloh J, Sampson EJ. 2005. Biochemical indicators of B vitamin
status in the US population after folic acid fortification: results from the National Health and Nutrition
Examination Survey 1999–2000. Am. J. Clin. Nutr. 82:442–50
236 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
98. Pfeiffer CM, Osterloh JD, Kennedy-Stephenson J, Picciano MF, Yetley EA, et al. 2008. Trends in
circulating concentrations of total homocysteine among US adolescents and adults: findings from the
1991–1994 and 1999–2004 National Health and Nutrition Examination Surveys. Clin. Chem. 54:801–13
99. Philip D, Buch A, Moorthy D, Scott TM, Parnell LD, et al. 2015. Dihydrofolate reductase 19-bp deletion
polymorphism modifies the association of folate status with memory in a cross-sectional multi-ethnic
study of adults. Am. J. Clin. Nutr. 102:1279–88
100. Pikula A, Beiser AS, DeCarli C, Himali JJ, Debette S, et al. 2012. Multiple biomarkers and risk of clinical
and subclinical vascular brain injury: the Framingham Offspring Study. Circulation 125:2100–7
101. Popp J, Lewczuk P, Linnebank M, Cvetanovska G, Smulders Y, et al. 2009. Homocysteine metabolism
and cerebrospinal fluid markers for Alzheimer’s disease. J. Alzheimers Dis. 18:819–28
102. Pynn CJ, Henderson NG, Clark H, Koster G, Bernhard W, Postle AD. 2011. Specificity and rate of
human and mouse liver and plasma phosphatidylcholine synthesis analyzed in vivo. J. Lipid Res. 52:399–
407
103. Rajagopalan P, Hua X, Toga AW, Jack CR, Weiner MW, Thompson PM. 2011. Homocysteine effects
on brain volumes mapped in 732 elderly individuals. Neuroreport 22:391–95
104. Ravaglia G, Forti P, Maioli F, Martelli M, Servadei L, et al. 2005. Homocysteine and folate as risk factors
for dementia and Alzheimer disease. Am. J. Clin. Nutr. 82:636–43
105. Refsum H, Nurk E, Smith AD, Ueland PM, Gjesdal CG, et al. 2006. The Hordaland Homocys-
teine Study: a community-based study of homocysteine, its determinants, and associations with disease.
J. Nutr. 136:1731S–40
106. Refsum H, Smith AD, Ueland PM, Nexo E, Clarke R, et al. 2004. Facts and recommendations about
total homocysteine determinations: an expert opinion. Clin. Chem. 50:3–32
107. Refsum H, Ueland P, Nyga˚ rd O, Vollset SE. 1998. Homocysteine and cardiovascular disease. Annu. Rev.
Med. 49:31–62
108. Reitz C, Tang MX, Miller J, Green R, Luchsinger JA. 2009. Plasma homocysteine and risk of mild
cognitive impairment. Dement. Geriatr. Cogn. Disord. 27:11–17
109. Risacher SL, Shen L, West JD, Kim S, McDonald BC, et al. 2010. Longitudinal MRI atrophy biomarkers:
relationship to conversion in the ADNI cohort. Neurobiol. Aging 31:1401–18
110. Robbins MA, Elias MF, Budge MM, Brennan SL, Elias PK. 2005. Homocysteine, type 2 diabetes mellitus,
and cognitive performance: The Maine-Syracuse Study. Clin. Chem. Lab. Med. 43:1101–6
111. Rosenberg IH, Rosenberg LE. 1998. The implications of genetic diversity for nutrient requirements:
the case of folate. Nutr. Rev. 56:S47–53
112. Sachdev PS, Valenzuela M, Wang XL, Looi JC, Brodaty H. 2002. Relationship between plasma homo-
cysteine levels and brain atrophy in healthy elderly individuals. Neurology 58:1539–41
113. Satija A, Yu E, Willett WC, Hu FB. 2015. Understanding nutritional epidemiology and its role in policy.
Adv. Nutr. 6:5–18
114. Schafer JH, Glass TA, Bolla KI, Mintz M, Jedlicka AE, Schwartz BS. 2005. Homocysteine and cognitive
function in a population-based study of older adults. J. Am. Geriatr. Soc. 53:381–88
115. Selhub J. 1999. Homocysteine metabolism. Annu. Rev. Nutr. 19:217–46
116. Selhub J, Bagley LC, Miller J, Rosenberg IH. 2000. B vitamins, homocysteine, and neurocognitive
function in the elderly. Am. J. Clin. Nutr. 71:614S–20
117. Selley ML. 2003. Increased concentrations of homocysteine and asymmetric dimethylarginine and de-
creased concentrations of nitric oxide in the plasma of patients with Alzheimer’s disease. Neurobiol. Aging
24:903–7
118. Selley ML. 2007. A metabolic link between S-adenosylhomocysteine and polyunsaturated fatty acid
metabolism in Alzheimer’s disease. Neurobiol. Aging 28:1834–39
119. Seshadri S, Beiser A, Selhub J, Jacques PF, Rosenberg IH, et al. 2002. Plasma homocysteine as a risk
factor for dementia and Alzheimer’s disease. N. Engl. J. Med. 346:476–83
120. Seshadri S, Wolf PA, Beiser AS, Selhub J, Au R, et al. 2008. Association of plasma total homocysteine
levels with subclinical brain injury: cerebral volumes, white matter hyperintensity, and silent brain infarcts
at volumetric magnetic resonance imaging in the Framingham Offspring Study. Arch. Neurol. 65:642–49
121. Smith AD. 2008. The worldwide challenge of the dementias: a role for B vitamins and homocysteine?
Food Nutr. Bull. 29:S143–72
www.annualreviews.org •Homocysteine and Cognitive Impairment 237
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
122. Smith AD, de Jager CA, Refsum H, Rosenberg IH. 2015. Homocysteine lowering, B vitamins, and
cognitive aging. Am. J. Clin. Nutr. 101:415–16
123. Smith AD, Smith SM, de Jager CA, Whitbread P, Johnston C, et al. 2010. Homocysteine-lowering
by B vitamins slows the rate of accelerated brain atrophy in mild cognitive impairment: a randomized
controlled trial. PLOS ONE 5:e12244
124. Smith GD, Ebrahim S. 2003. ‘Mendelian randomization’: Can genetic epidemiology contribute to un-
derstanding environmental determinants of disease? Int. J. Epidemiol. 32:1–22
125. Snyder HM, Corriveau RA, Craft S, Faber JE, Greenberg SM, et al. 2015. Vascular contributions to
cognitive impairment and dementia including Alzheimer’s disease. Alzheimers Dement. 11:710–17
126. Sontag JM, Sontag E. 2014. Protein phosphatase 2A dysfunction in Alzheimer’s disease. Front. Mol.
Neurosci. 7:16
127. Stabler SP. 2013. Clinical practice. Vitamin B12 deficiency. N. Engl. J. Med. 368:149–60
128. Stern Y, Liu XH, Albert M, Brandt J, Jacobs DM, et al. 1996. Application of a growth curve approach
to modeling the progression of Alzheimer’s disease. J. Gerontol. A Biol. Sci. Med. Sci. 51:M179–84
129. Strike SC, Carlisle A, Gibson EL, Dyall SC. 2015. A high omega-3 fatty acid multinutrient supplement
benefits cognition and mobility in older women: a randomized, double-blind, placebo-controlled pilot
study. J. Gerontol. A Biol. Sci. Med. Sci. 71:236–42
130. Sudduth TL, Powell DK, Smith CD, Greenstein A, Wilcock DM. 2013. Induction of hyperhomocys-
teinemia models vascular dementia by induction of cerebral microhemorrhages and neuroinflammation.
J. Cereb. Blood Flow Metab. 33:708–15
131. Tacconi M, Wurtman RJ. 1985. Phosphatidylcholine produced in rat synaptosomes by N-methylation
is enriched in polyunsaturated fatty acids. PNAS 82:4828–31
132. Tangney CC, Aggarwal NT, Li H, Wilson RS, Decarli C, et al. 2011. Vitamin B12, cognition, and brain
MRI measures: a cross-sectional examination. Neurology 77:1276–82
133. Toda N, Okamura T. 2012. Cerebral blood flow regulation by nitric oxide in Alzheimer’s disease.
J. Alzheimers Dis. 32:569–78
134. Troen AM, Chao WH, Crivello NA, D’Anci KE, Shukitt-Hale B, et al. 2008. Cognitive impairment
in folate-deficient rats corresponds to depleted brain phosphatidylcholine and is prevented by dietary
methionine without lowering plasma homocysteine. J. Nutr. 138:2502–9
135. Troen AM, Shea-Budgell M, Shukitt-Hale B, Smith DE, Selhub J, Rosenberg IH. 2008. B-vitamin
deficiency causes hyperhomocysteinemia and vascular cognitive impairment in mice. PNAS 105:12474–
79
136. Tucker KL, Qiao N, Scott T, Rosenberg I, Spiro A, 3rd. 2005. High homocysteine and low B vitamins
predict cognitive decline in aging men: the Veterans Affairs Normative Aging Study. Am. J. Clin. Nutr.
82:627–35
137. Ueland PM, Refsum H, Beresford SA, Vollset SE. 2000. The controversy over homocysteine and car-
diovascular risk. Am. J. Clin. Nutr. 72:324–32
138. Univ. Oxford. 2014. Taking B vitamins won’t prevent Alzheimer’s disease. Univ. Oxford News Events,
July 15. http://www.ox.ac.uk/news/2014-07-15-taking-b-vitamins-won%E2%80%99t-prevent-
alzheimer%E2%80%99s-disease
139. van der Zwaluw NL, Dhonukshe-Rutten RA, van Wijngaarden JP, Brouwer-Brolsma EM, van de Rest
O, et al. 2014. Results of 2-year vitamin B treatment on cognitive performance: secondary data from an
RCT. Neurology 83:2158–66
140. Vermeer SE, van Dijk EJ, Koudstaal PJ, Oudkerk M, Hofman A, et al. 2002. Homocysteine, silent brain
infarcts, and white matter lesions: The Rotterdam Scan Study. Ann. Neurol. 51:285–89
141. Vogiatzoglou A, Refsum H, Johnston C, Smith SM, Bradley KM, et al. 2008. Vitamin B12 status and
rate of brain volume loss in community-dwelling elderly. Neurology 71:826–32
142. Vogiatzoglou A, Smith AD, Nurk E, Drevon CA, Ueland PM, et al. 2013. Cognitive function in an
elderly population: interaction between vitamin B12 status, depression, and apolipoprotein E ε4: the
Hordaland Homocysteine Study. Psychosom. Med. 75:20–29
143. Vry MS, Haerter K, Kastrup O, Gizewski E, Frings M, Maschke M. 2005. Vitamine-B12-deficiency
causing isolated and partially reversible leukoencephalopathy. J. Neurol. 252:980–82
238 Smith ·Refsum
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

NU36CH09-Smith ARI 10 June 2016 11:7
144. Wald DS, Kasturiratne A, Simmonds M. 2010. Effect of folic acid, with or without other B vitamins, on
cognitive decline: meta-analysis of randomized trials. Am. J. Med. 123:522–27.e2
145. Wald DS, Kasturiratne A, Simmonds M. 2011. Serum homocysteine and dementia: meta-analysis of
eight cohort studies including 8669 participants. Alzheimers Dement. 7:412–17
146. Wang JZ, Xia YY, Grundke-Iqbal I, Iqbal K. 2013. Abnormal hyperphosphorylation of tau: sites, regula-
tion, and molecular mechanism of neurofibrillary degeneration. J. Alzheimers Dis. 33(Suppl. 1):S123–39
147. Watkins SM, Zhu X, Zeisel SH. 2003. Phosphatidylethanolamine-N-methyltransferase activity and
dietary choline regulate liver-plasma lipid flux and essential fatty acid metabolism in mice. J. Nutr.
133:3386–91
148. Wei W, Liu Y-H, Zhang CE, Wang Q, Wei Z, et al. 2011. Folate/vitamin-B12 prevents
chronic hyperhomocysteinemia-induced tau hyperphosphorylation and memory deficits in aged rats.
J. Alzheimers Dis. 27:639–50
149. Wen Y, Yang S, Liu R, Simpkins JW. 2004. Transient cerebral ischemia induces site-specific hyper-
phosphorylation of tau protein. Brain Res. 1022:30–38
150. Wen Y, Yang SH, Liu R, Perez EJ, Brun-Zinkernagel AM, et al. 2007. Cdk5 is involved in NFT-like
tauopathy induced by transient cerebral ischemia in female rats. Biochim. Biophys. Acta 1772:473–83
151. Whalley LJ, Staff RT, Murray AD, Duthie SJ, Collins AR, et al. 2003. Plasma vitamin C, cholesterol and
homocysteine are associated with grey matter volume determined by MRI in non-demented old people.
Neurosci. Lett. 341:173–76
152. Whiley L, Sen A, Heaton J, Proitsi P, Garc´
ıa-G ´
omez D, et al. 2014. Evidence of altered phosphatidyl-
choline metabolism in Alzheimer’s disease. Neurobiol. Aging 35:271–78
153. Whitwell JL, Josephs KA, Murray ME, Kantarci K, Przybelski SA, et al. 2008. MRI correlates of neu-
rofibrillary tangle pathology at autopsy: a voxel-based morphometry study. Neurology 71:743–49
154. Williams JH, Pereira EA, Budge MM, Bradley KM. 2002. Minimal hippocampal width relates to plasma
homocysteine in community-dwelling older people. Age Ageing 31:440–44
155. Williams RJ. 1986. Biochemical Individuality: Basis for the Genetotrophic Concept. New Canaan, CT: Keats
156. Wilson RS, Beckett LA, Barnes LL, Schneider JA, Bach J, et al. 2002. Individual differences in rates of
change in cognitive abilities of older persons. Psychol. Aging 17:179–93
157. Yan J, Ginsberg SD, Powers B, Alldred MJ, Saltzman A, et al. 2014. Maternal choline supplementation
programs greater activity of the phosphatidylethanolamine N-methyltransferase (PEMT) pathway in
adult Ts65Dn trisomic mice. FASEB J. 28:4312–23
158. Yang LK, Wong KC, Wu MY, Liao SL, Kuo CS, Huang RFS. 2007. Correlations between folate, B12,
homocysteine levels, and radiological markers of neuropathology in elderly post-stroke patients. J. Am.
Coll. Nutr. 26:272–78
159. Ye W, Blain SW. 2010. S phase entry causes homocysteine-induced death while ataxia telangiectasia and
Rad3 related protein functions anti-apoptotically to protect neurons. Brain 133:2295–312
160. Yuki D, Sugiura Y, Zaima N, Akatsu H, Takei S, et al. 2014. DHA-PC and PSD-95 decrease after loss
of synaptophysin and before neuronal loss in patients with Alzheimer’s disease. Sci. Rep. 4:7130
161. Zhang CE, Tian Q, Wei W, Peng JH, Liu GP, et al. 2008. Homocysteine induces tau phosphorylation
by inactivating protein phosphatase 2A in rat hippocampus. Neurobiol. Aging 29:1654–65
162. Zhuo JM, Wang H, Pratic`
o D. 2011. Is hyperhomocysteinemia an Alzheimer’s disease (AD) risk factor,
an AD marker, or neither? Trends Pharmacol. Sci. 32:562–71
163. Zylberstein DE, Lissner L, Bj¨
orkelund C, Mehlig K, Thelle DS, et al. 2011. Midlife homocysteine and
late-life dementia in women. A prospective population study. Neurobiol. Aging 32:380–86
164. Zylberstein DE, Skoog I, Bj¨
orkelund C, Guo X, Hult´
en B, et al. 2008. Homocysteine levels and lacunar
brain infarcts in elderly women: the prospective population study of women in Gothenburg. J. Am.
Geriatr. Soc. 56:1087–91
www.annualreviews.org •Homocysteine and Cognitive Impairment 239
Annu. Rev. Nutr. 2016.36:211-239. Downloaded from www.annualreviews.org
Access provided by University of Oxford - Bodleian Library on 11/10/16. For personal use only.

1
SUPPLEMENTAL MATERIAL
for
Homocysteine, B vitamins and cognitive impairment
A. David Smith and Helga Refsum
Contents
Does high folate status impair cognition? Page 2
Supplemental Figures 1 – 6 Page 10
Supplemental Figure 1. Changes in episodic memory over time in relation to
plasma total homocysteine (tHcy) and plasma folate.
Supplemental Figure 2. Meta-analysis of prospective studies on the association of
tHcy with incident dementia.
Supplemental Figure 3. Cognitive decline in Alzheimer’s disease.
Supplemental Figure 4. Regulation of protein phosphatase-2A (PP2A) by
methylation.
Supplemental Figure 5. Brain atrophy in VITACOG trial in subjects with Mild
Cognitive Impairment.
Supplemental Figure 6. Effect of B vitamin treatment on MMSE and semantic
memory in Mild Cognitive Impairment.
.

2
Does high folate status impair cognition?
In spite of much evidence, which will not be reviewed here, of the beneficial cognitive effects
of a good folate status (2; 14; 17), there have also been several reports that high intake or
high blood concentrations of folate may be detrimental to some cognitive functions. We will
briefly review these reports here.
The first report came in 2005 from the Chicago Health and Aging Project (CHAP) (7), a
study on 3,718 community-dwelling elderly with average age 74 y. 91% of the cognitive
assessments (4 different tests) in this study were done after the introduction of mandatory
folic acid fortification. Vitamin consumption was estimated from a Food Frequency
Questionnaire and records of supplement use were made. It was found that those in the 4th
and 5th quintiles of total folate intake (349 -1719 g/day) had a faster rate of cognitive decline
over 6 y than those in the 1st quintile (63-221g/day). Notably, participants who consumed
400-1200 g/day of folic acid from supplements showed greater cognitive decline than non-
supplement users.
The second study came from the participants in the NHANES 1999-2002 cohort, who had
been exposed to mandatory folic acid fortification, starting in 1996 and fully implemented in
1998. Out of 1459 subjects (average age 70 y), 346 had low B12 status (defined as serum
B12 < 148 pmol/L or serum MMA > 210 nmol/L), 20.7% had serum folate > 59 nmol/L and
67% reported consuming folic acid-containing supplements (8). Only one cognitive test was
used in this study, the Digit Symbol-Coding test, which is more sensitive to impairment than
the MMSE and is primarily a test of cognitive speed. It was found that elderly subjects with
poor vitamin B12 status and high serum folate had a greater risk of anemia and of cognitive
impairment than those with low B12 and normal folate status (Figure 1). In contrast, those
with normal B12 and high folate had a lower risk of cognitive impairment, while those with
low B12 and high folate had a 5-fold increased risk of cognitive impairment. Although the
number of subjects in the latter category was fairly small, at 42, the odds ratio was robust at
5.0 (95% CI 2.7, 9.5).
Figure 1. Anemia and cognitive impairment according to folate and B12 status in senior participants in the
NHANES survey 1999–2002). Asterisks indicate significant differences. High folate was defined as serum
folate > 59 nmol/L and low B12 as serum B12 < 148 pmol/L or serum MMA > 210 nmol/L. Drawn from data in
Table 2 of reference (8).

3
Another report from the NHANES cohort (9) examined whether the presence of
unmetabolized folic acid in serum was related to cognitive impairment. Circulating folic acid
was detected in 33% of the participants, and in those with poor B12 status, the presence of
folic acid in serum was associated with poorer scores in the cognitive test, whereas the level
of 5-methyltetrahydrofolate was not related to cognitive scores. The authors pointed out that
their result was consistent with the finding described above from the CHAP cohort (7) that
participants consuming folic acid supplements showed a greater cognitive decline than those
who did not. The authors came to the tentative conclusion: “Our findings for cognitive test
performance in the subjects with low vitamin B-12 status may be consistent with the idea that
folic acid harms the nervous system via a mechanism that involves circulating unmetabolized
folic acid.”
A study on 549 members of the Framingham Heart Study cohort (aged 74.8 y) is consistent
with the hypothesis that it may be unmetabolized folic acid that is harmful. This study was
carried out before the introduction of mandatory folic acid fortification and recorded the
MMSE scores of the cohort over an 8 y period according to the baseline vitamin B12 status.
Participants with serum B12 of < 258 pmol/L showed faster rate of decline in MMSE than
those with B12 > 258 pmol/L but there was a striking effect of folic acid-containing
supplements: taking supplements had no effect on decline in those with B12 > 258 pmol/L
but markedly increased the rate of decline in those with B12 < 258 pmol/L (Figure 2)
Figure 2. Annual change in Mini-Mental State Examination (MMSE) scores during the 8 years of the original
Framingham cohort (n = 549) for users and nonusers of folic acid containing supplements, stratified according
to vitamin B-12 status. Results are controlled for age, sex, educational achievement, and baseline serum
creatinine status, body mass index, and smoking status. Points represent -coefficients. Error bars represent 95%
confidence intervals for the -coefficients. (Redrawn from Morris et al. (10)).
Two reports from Australia have also shown that high folate status is related to poorer
cognition. The first is a study on the Australian Imaging Biomarker Lifestyle (AIBL) cohort,
which includes 760 normal elderly, 130 with MCI and 205 with AD (3). A significantly
lower red cell folate was found for the AD patients compared with controls. However, the
relationship between cognitive test scores and red cell folate in the whole cohort was complex
and best fitted a quadratic inverted U-relationship. Long-term memory, total memory and
global cognition all showed this relationship, with cognitive test performance improving up
to a red cell folate of about 1,500 nmol/L for males and up to 1,200 nmol/L for females
whereas higher red cell folate in both sexes was associated with worse cognitive

4
performance. The authors suggested “excessively elevated red cell folate levels may be
associated with a detrimental cognitive state.” An important additional conclusion from this
study is that the high folate status is not due to the methylfolate trap phenomenon that occurs
in B12 deficiency (16). (See also below).
The second study from Australia combined the AIBL cohort with two other cohorts to make a
total of 1,354 subjects of whom 687 did not have cognitive impairment (6). This cohort had
blood samples taken after the introduction of voluntary folic acid fortification but only 17%
had samples taken after mandatory fortification was introduced. The authors divided the
cohort into those with ‘low’ B12 levels < 250 pmol/L and those with ‘high’ red cell folate
>1,594 nmol/L (90th centile). A score < 24 on MMSE was defined as cognitive impairment.
Setting the group with normal red cell folate (< 1,594 nmol/L) and B12 > 250 pmol/L as
reference, there were 39 subjects with high red cell folate and low B12 with an adjusted odds
ratio for cognitive impairment of 3.45 (1.60-7.43). Even among the 97 subjects with high red
cell folate and normal B12, the risk of cognitive impairment was increased with an odds ratio
of 1.74 (1.03-2.05). The median B12 in the latter group was 383 pmol/L. Thus, in contrast to
the NHANES study above, it was not only those with poor B12 status who had an increased
risk of impairment if they also had high folate, but also those with B12 in the normal range.
Red cell folate levels above 1,600 nmol/L are unusual unless the person consumes folic acid
by fortification or as a supplement: this is consistent with 18% of those with high folate
reporting that they took supplements, whereas less than 6% of those with normal folate did.
There are other studies on this topic as well (1; 5). In particular one study from UK is cited as
an argument against the theory (1). Voluntary fortification had been introduced some years
before the study. According to the authors “There was no evidence of modification by high
folate status of the associations of low B12 with anaemia or cognitive impairment in the
setting of voluntary fortification”. However, this study does not really examine very high
folate levels since those defined as high folate (> 30 nmol/L) have much lower folate than the
values in the US studies (>59 nmol/L). When the UK authors used a threshold of ≥60 nmol/L,
low B12 was associated with cognitive impairment (OR 2.46 (0·90, 6.71) but did not reach
significance, probably because the number with impairment was only 7 subjects. Even in
those with modestly elevated folate > 30 nmol/L and low B12 status, there was an increased
risk increase of cognitive impairment (OR 1.50 (0.91, 2.46), but again it did not reach
significance. Overall, our interpretation of this study is that it supports the studies from USA
and Australia that the combination of high folate with low B12 is indeed harmful in relation
to cognition, but that this particular study lacked power.
Another study in the USA also reported no significant cognitive impairment in a subgroup of
the SALSA cohort with low B12 (<148 pmol/L) and high serum folate (>45.3 nmol/L), but
only 22 out of a total of 1525 subjects had this combination and odds ratios were not reported
(5). It is thus possible, as suggested by the authors, that the study was underpowered to detect
an effect on the relatively insensitive cognitive tests used.
In conclusion, the above reports show that in the elderly intake of folic acid supplements (7;
10) and/or a high serum (8) or red cell folate status (3; 6) are associated with an increased risk

5
of cognitive impairment. One report suggests that the effect on cognition is due to circulating
unmetabolized folic acid (9). These conclusions have been challenged by a study on elderly
in the UK (1), but this study is clearly underpowered, and if anything, tends to support the
other studies.
A recent study on the effect of the dihydrofolate reductase (DHFR) 19-bp deletion
polymorphism on cognitive function gives another example of how high folate levels can
impair cognition (12). In this study of 1402 community living adults from the Boston area,
higher serum folate concentrations (above 40.3 nmol/L) in the whole cohort were associated
with improved performance on a memory test. However, the 23% of the cohort with the
DHFR del/del genotype had worse memory scores (ß ± SE = - 0.24 ± 0.10, P < 0.05) than
those with the del/ins and ins/ins genotypes. Most importantly, the effect of the
polymorphism depended on folate status so that carriers of the del/del genotype with high
plasma folate had worse memory scores than those of both non-carriers with high folate and
del/del carriers with normal folate (ß-interaction = 0.26 ± 0.13, P < 0.05). (Figure 3).
Figure 3: Interaction of DHFR 19-bp deletion polymorphism and plasma folate with respect to memory. The
graph shows mean (95% CI) adjusted factor memory scores (FACmem) by genotype below and above the
plasma folate cutoff of 17.8 ng folate/mL (40.3 nmol/L). Memory scores were adjusted in the general linear
model for ethnicity, age, sex, education, estimated glomerular filtration rate, diabetes, hypertension,
apolipoprotein E4 genotype, plasma tHcy and vitamins B12 and B6. *P < 0.05, ** P < 0.01. Figure 2 from
reference (12), with permission.
The authors concluded that in a population with a similar frequency of the DHFR del/del
polymorphism and similar folate levels, for every 4 individuals who would benefit from high
folate, one person would risk memory impairment.
Possible mechanisms
Before discussing possible mechanisms for the association of high folate with cognitive
impairment, we must consider if the high folate is consequence of low B12 status, which is

6
known to cause cognitive impairment. Quinlivan (13) suggested that the high serum folate in
this situation is a consequence of the methylfolate trap that follows B12 deficiency. If this is
the case, the high serum folate (as 5-methytetrahydrofolate) would be accompanied by low
red cell folate since the cells cannot take up 5-methytetrahydrofolate. However, two studies
have now found that in people with low B12 and high folate, the red cell folate is also
elevated: the SALSA study (5) and the Australian study (6). So we can dismiss the
methylfolate trap as an explanation.
A novel approach was adopted by Selhub et al. (15)who compared two NHANES cohorts,
one from before mandatory folic acid fortification and the other from after fortification. In
this way, the authors were able to show that in subjects with B12 <148 pmol/L, the
relationship between folate level and tHcy was non-linear. The concentration of tHcy was
15.5 mol/L at a folate level of 7 nmol/L and it fell as the folate level increased until about
20 nmol/L, but it then started to increase again as the folate rose towards 44 nmol/L. In other
words, there was a U-shaped relationship between tHcy and serum folate. A similar, if less
marked, relationship was observed for MMA. The authors explored several hypotheses to
explain what appears to be an inhibition of B12-dependent pathways in people with high
folate and low B12. They suggested that folic acid or a metabolite could oxidise the cobalt
atom in cobalamin, thereby inactivating the cofactor.
It is of note that the Australian authors (3) found a similar U-shaped relationship between
tHcy and red cell folate in the whole AIBL cohort, not just in those with low B12: tHcy levels
fell as folate rose from 300 to 1,500 nmol/L at which point there was an inflection with tHcy
rising as folate increased from 1,500 to 3,000 nmol/L (Figure 3). No such relationship was
found for serum folate in this cohort. It should be noted that red cell folate levels above about
1,500 nmol/L are rarely seen in populations not exposed to voluntary or mandatory folic acid
fortification, or in people who do not consume folic acid supplements(4). Voluntary
fortification was permitted in Australia and mandatory fortification was introduced in 2009.
Figure 3. The association of serum and red cell folate with tHcy in the AIBL cohort. Modified from Figure 4B
in reference (3), with permission.

7
The authors speculated that the U-shape might be caused by inhibition of methionine
synthase by pro-oxidants that are common in aging, which is similar to Selhub’s suggestion
(15). To explain the link with cognitive impairment they wrote “…we hypothesize that
because there is a direct relationship between homocysteine and cognition in each clinical
group, the association of highly elevated red cell folate with worse cognition is a reflection of
the impact of elevated homocysteine upon cognition.”
An important finding has provided an additional explanation: Miller et al. (5) confirmed in
the SALSA cohort the elevation of tHcy and of MMA in those with high serum folate (>45.3
nmol/L) and low B12 (< 148 pmol/L) but also found that the concentration of
holotranscobalamin was markedly lower in in the low B12/ high folate group than in the
group with low B12 /normal folate. Thus, the effect of high folate seems to extend to the
availability of B12 itself, not only its catalytic activity.
Unmetabolized folic acid. In the study by Morris in 2010 (9), carried out after mandatory
folic acid fortification, it was found that 33% of the NHANES cohort had detectable
unmetabolized folic acid in their plasma. Importantly, the impairment in cognition at high
folate levels in those with low B12 status was only found in those with circulating
unmetabolized folic acid; it was not significantly related to high levels of
methyltetrahydrofolate (MTHF). In contrast, those with normal B12 status showed an
improvement in cognition as the MTHF level increased but no significant improvement as the
folic acid level increased. The authors suggested the following hypothesis: “Our findings for
cognitive test performance in the subjects with low vitamin B-12 status may be consistent
with the idea that folic acid harms the nervous system via a mechanism that involves
circulating unmetabolized folic acid.”
The findings of Philip et al. (12). described above are consistent with this hypothesis,
although they are not a critical test of it. It appears that the del/del polymorphism of DHFR
has limited effect on plasma and red cell folate concentrations (4; 11). However in subjects
with the del/del genotype, a subgroup with high folic acid intake (> 500 µg/d) showed an
increased prevalence of high plasma umetabolized folic acid (4). It is possible, therefore, that
the memory impairment seen at high folate levels in those with the del/del genotype might be
related to an increase in circulating umetabolized folic acid.
Conclusion
Overall, the results suggest that high folic acid intake leading to high folate status could be
harmful to cognitive health, in particular in those with low B12 status. Increasing evidence
suggest that this may be a unique effect of folic acid, but we cannot ignore that the effect
might be explained by the efficiency of folic acid to raise circulating folate status or that it
adversely affects intracellular or tissue metabolism of folate. At the moment, the evidence
suggests that one should be careful in supplementing folic acid without either ruling out low

8
B12 status or by combining folic acid with B12. The fact that even some elderly with B12 in
the normal range may have an increased risk of cognitive impairment at high folate status
might be related to the high prevalence (10 to 48% in different populations) of the del/del
polymorphism of DHFR (12).
The public health significance of these findings for the elderly may be considerable. A large
number of countries have now introduced mandatory fortification of flour with folic acid and
the use of folic acid containing supplements is very common: from 33 – 67% of Americans
take such supplements regularly (18). If we take the approximate prevalence of high folate
and low B12 status after fortification as 4% of the elderly (8), then about 1.8 million
American elderly might be at increased risk of cognitive impairment by this cause alone (18).
References
1. Clarke R, Sherliker P, Hin H, Molloy AM, Nexo E, et al. 2008. Folate and vitamin B12 status in
relation to cognitive impairment and anaemia in the setting of voluntary fortification in the UK. Br J
Nutr 100:1054-9
2. Durga J, van Boxtel MP, Schouten EG, Kok FJ, Jolles J, et al. 2007. Effect of 3-year folic acid
supplementation on cognitive function in older adults in the FACIT trial: a randomised, double blind,
controlled trial. Lancet 369:208-16
3. Faux NG, Ellis KA, Porter L, Fowler CJ, Laws SM, et al. 2011. Homocysteine, vitamin B12, and folic
acid levels in Alzheimer's disease, mild cognitive impairment, and healthy elderly: baseline
characteristics in subjects of the Australian Imaging Biomarker Lifestyle study. J Alzheimers Dis
27:909-22
4. Kalmbach RD, Choumenkovitch SF, Troen AP, Jacques PF, D'Agostino R, Selhub J. 2008. A 19-base
pair deletion polymorphism in dihydrofolate reductase is associated with increased unmetabolized folic
acid in plasma and decreased red blood cell folate. J Nutr 138:2323-7
5. Miller JW, Garrod MG, Allen LH, Haan MN, Green R. 2009. Metabolic evidence of vitamin B12
deficiency, including high homocysteine and methylmalonic acid and low holotranscobalamin, is more
pronounced in older adults with elevated plasma folate. Am J Clin Nutr 90:1586-92
6. Moore E, Ames D, Mander A, Carne R, Brodaty H, et al. 2014. Among vitamin B12 deficient older
people, high folate levels are associated with worse cognitive function: combined data from three
cohorts. J Alzheimers Dis 39:661-8
7. Morris MC, Evans DA, Bienias JL, Tangney CC, Hebert LE, et al. 2005. Dietary folate and vitamin
B12 intake and cognitive decline among community-dwelling older persons. Arch Neurol 62:641-5
8. Morris MS, Jacques PF, Rosenberg IH, Selhub J. 2007. Folate and vitamin B-12 status in relation to
anemia, macrocytosis, and cognitive impairment in older Americans in the age of folic acid
fortification. Am J Clin Nutr 85:193-200
9. Morris MS, Jacques PF, Rosenberg IH, Selhub J. 2010. Circulating unmetabolized folic acid and 5-
methyltetrahydrofolate in relation to anemia, macrocytosis, and cognitive test performance in American
seniors. Am J Clin Nutr 91:1733-44
10. Morris MS, Selhub J, Jacques PF. 2012. Vitamin B-12 and folate status in relation to decline in scores
on the Mini-Mental State Examination in the Framingham Heart Study. J Am Geriatr Soc 60:1457-64

9
11. Ozaki M, Molloy AM, Mills JL, Fan R, Wang Y, et al. 2015. The dihydrofolate reductase 19 bp
polymorphism Is not associated with biomarkers of folate status in healthy young adults, irrespective of
folic acid intake. J Nutr 145:2207-11
12. Philip D, Buch A, Moorthy D, Scott TM, Parnell LD, et al. 2015. Dihydrofolate reductase 19-bp
deletion polymorphism modifies the association of folate status with memory in a cross-sectional
multi-ethnic study of adults. Am J Clin Nutr 102:1279-88
13. Quinlivan EP. 2008. In vitamin B12 deficiency, higher serum folate is associated with increased
homocysteine and methylmalonic acid concentrations. Proc Natl Acad Sci U S A 105:E7; author reply
E8
14. Selhub J, Bagley LC, Miller J, Rosenberg IH. 2000. B vitamins, homocysteine, and neurocognitive
function in the elderly. Am J Clin Nutr 71:614S-20S
15. Selhub J, Morris MS, Jacques PF. 2007. In vitamin B12 deficiency, higher serum folate is associated
with increased total homocysteine and methylmalonic acid concentrations. PNAS 104:19995-20000
16. Shane B, Stokstad EL. 1985. Vitamin B12-folate interrelationships. Annu Rev Nutr 5:115-41
17. Smith AD. 2008. The worldwide challenge of the dementias: A role for B vitamins and homocysteine?
Food Nutr Bull 29:S143-72
18. Smith AD, Kim YI, Refsum H. 2008. Is folic acid good for everyone? Am J Clin Nutr 87:517-33

10
Supplemental Figures
Supplemental Figure 1. Changes in episodic memory over time in relation to plasma total
homocysteine (tHcy) and plasma folate. Associations between scores on an episodic
memory test (Kendrick Objective Learning Test, KOLT) at follow-up and changes in plasma
concentrations of tHcy (n = 1,670) and folate (n = 1,660) between baseline and follow-up
over a 6 y period. The reference value for KOLT is the approximate value associated with no
change over the 6 y period in tHcy or folate levels, respectively. Solid lines are the estimated
concentration-response curves; dashed lines show the 95% confidence intervals. (Modified
from reference 88).

11
Supplemental Figure 2. Meta-analysis of prospective studies on the association of
tHcy with incident dementia. Modified from reference 7.

12
Supplemental Figure 3. Cognitive decline in Alzheimer’s disease. Expected course of
CAMCOG scores (global cognition) in patients with AD aged 70 y according to baseline
tHcy concentrations at the time of diagnosis. The plots were generated from a non-linear
mixed model adjusted for several covariates from patients in OPTIMA. Redrawn from
reference 93.

13
Supplemental Figure 4. Regulation of protein phosphatase-2A (PP2A) by methylation (from
Sontag & Sontag (126) with permission. The formation of the PP2A/Bα holoenzyme, the
primary Ser/Thr tau phosphatase in vivo, is believed to be controlled by Leu-309 methylation
of PP2A catalytic subunit by LCMT1. This reaction requires the supply of SAM, the
universal methyl donor, and is inhibited by SAH. The PP2A methylesterase, PME-1, can
demethylate and inactivate PP2A through distinct mechanisms, and form a complex with
inactive PP2A enzymes. Those inactive complexes could be reactivated via the action of the
PP2A activator PTPA, allowing for subsequent methylation of PP2AC subunit. Many brain
Ser/Thr protein kinases, including GSK3β, oppose the action of PP2A/Bα and promote tau
phosphorylation. Inhibition and/or down-regulation of PP2A can enhance tau
phosphorylation directly by preventing its dephosphorylation, or indirectly by up-regulating
tau kinases.

14
Supplemental Figure 5. Brain atrophy in VITACOG trial in subjects with Mild Cognitive
Impairment. Global brain atrophy rate in relation to baseline tHcy concentration. Subjects
were treated for 2 y with B vitamins (folic acid, vitamins B6 and B12). Plotted from data in
Table S2, reference 123.

15
Supplemental Figure 6. Effect of B vitamin treatment on MMSE and semantic memory.
Subjects (n = 266) with Mild Cognitive Impairment in the VITACOG trial were treated with
placebo or B vitamins (folic acid, vitamins B6 and B12) for 2 years. The graphs illustrate the
longitudinal effect of treatment on the MMSE and semantic memory scores and its
interaction with baseline tHcy level using generalized linear mixed effect models. ‘Low’ and
‘High’ baseline tHcy refer to values below and above the median (11.3 μmol/L), respectively.
From Figure S2 of de Jager et al. (18).