OUP user menu

Challenges of long-term nutrition intervention studies on cognition: discordance between observational and intervention studies of vitamin B12 and cognition

Cherie McCracken
DOI: http://dx.doi.org/10.1111/j.1753-4887.2010.00325.x S11-S15 First published online: 1 November 2010


Conducting long-term nutrition intervention studies on cognition can be challenging. The gaps in current methodology are addressed via a case study of the relationship between vitamin B12 and cognition in people aged 60 and older. There is robust evidence from many observational studies, both cross-sectional and longitudinal, showing that a deficit of the vitamin is associated with poor or declining cognition in this age group, but supplementation of the vitamin in trials does not bring about improved cognition. The evidence from observational studies as well as clinical trials is reviewed here, and the potential difficulties in conducting long-term nutritional intervention studies in this area are highlighted.

  • cognition
  • community dwelling
  • elderly
  • methylmalonic acid (MMA)


It has been known since the 1950s that deficiency in vitamin B12 is linked to neuropsychiatric disorders, particularly difficulty with cognitive functioning and dementia.1 In people aged 60 years and older, this link between low vitamin B12 levels and poor cognition has been recorded in many observational studies. Why then is this association not upheld in intervention studies?


Vitamin B12 is a water-soluble vitamin found in food mainly from animal sources. Average daily consumption of 5–30 µg contributes to a store of approximately 2–5 mg, and the average adult daily requirement is 2.4 µg. Given this, it takes time for frank deficiency to manifest clinically, especially because folic acid can correct hematological (but not neurological) abnormalities caused by low levels of B12. Tissue vitamin B12 deficiency is associated with differing levels of serum B12. Vitamin B12 is carried on one of two carrier proteins (haptocorrin and transcobalamin), which, when saturated, becomes holotranscobalamin (holoTC). Haptocorrin binds the majority of serum B12 but, unlike holoTC, it does not deliver it to the metabolically active cells. Therefore, measurement of holoTC gives a more accurate representation of the tissue availability of vitamin B12 than serum B12, with holoTC carrying between 9% and 80% of serum B122 and higher percentages being carried with low vitamin status.3


Suboptimal levels of availability from low intake, poor intestinal absorption, or impaired use are reflected in higher levels of the B12 metabolites homocysteine (Hcy) and methylmalonic acid (MMA). However, Hcy level is not specific to B12; deficiencies of folic acid and vitamin B6 also cause high levels of Hcy. MMA is thus a more specific indicator of B12 deficiency. Renal damage is the only other known biochemical cause of high levels of MMA, but it is also indicated by levels of serum creatinine. It is therefore usual to also measure the level of serum creatinine when investigating levels of MMA. A combined approach to assessing B12 status based on a mathematical model of the relationship between MMA and holoTC has recently been proposed, although its utility in clinical settings has yet to be determined.4


Dementia, particularly Alzheimer's disease (AD), and the risk of developing dementia have been associated with the action of Hcy.58 Incident AD has also been associated with low levels of B12 and folate,9 although Selhub et al.10 reported important longitudinal evidence of an adverse interaction between high folate levels and low B12 levels to the detriment of cognitive functioning. High levels of Hcy have also been implicated in cognitive impairment11 and risk of cognitive decline.12,13 Similarly, high levels of MMA have been associated with AD14,15 and with cognitive performance.217


Plausible biological mechanisms exist for the action of both of these metabolites on cognitive function. In the case of Hcy, it affects cognitive function either by causing neuronal death, through it or its excitotoxic derivatives activating N-methyl-D-aspartate (NMDA) receptors,18 or through vascular damage.19 MMA is also neurotoxic, causing degeneration of the basal ganglia, possibly through the inhibition of respiratory chain complex II, and it may be similar to Hcy in that MMA induces changes that NMDA receptor blockade can prevent.20 Excessive MMA will prevent normal fatty acid synthesis and, if it is incorporated into fatty acids, will bring about demyelination, which can cause cognitive disruption, including speech impairment and memory loss.21


There is a wealth of studies examining the link between vitamin B12 levels and associated biochemical markers indicative of deficiency (holoTC, folate, Hcy, MMA, and creatinine) and cognition in older adults. Given that B12 administration may improve cognitive function in individuals with memory problems or with minimal cognitive impairment but not established dementia,22 the focus of this review is on studies that have examined B12 and related markers within the context of cognitive decline or impairment. This emphasis is particularly important given the potential of nutritional intervention to modify risk factors. Attention is also focused on MMA rather than Hcy because MMA is a more specific marker of B12 deficiency than Hcy, although some studies included in this review measured both.


Five cross-sectional observational studies were found that matched the criteria of elderly community-dwelling respondents without dementia whose vitamin B12 and MMA levels were examined in relation to their cognition.225 Most studies used a combination of levels of MMA and B12 to define deficiency, and this prevalence ranged from 25% to 43%. All studies showed an association between higher levels of MMA and cognitive impairment; however, in one study, this was part of low B12 definition rather than explicit,25 and in another, the relationship was lost in multivariate analysis.16 Hin et al.24 define cognitive impairment as scoring less than 22 out of 30 on the Mini-Mental State Examination (MMSE).26 Using multiple regression analysis to control for the effects of age, sex, and smoking, they found almost four times the odds of being in the impaired category with an MMA level in the highest versus the lowest quartile (odds ratio [OR], 3.7; 95% confidence interval [CI], 1.7–8.0). The significance of this relationship may be somewhat attenuated because serum creatinine was not taken into account, but significant relationships were also found between impairment and lower levels of holoTC and B12 (OR, 3.0 and 2.2, respectively). Morris et al.25 modeled cognitive impairment as scoring less than 34 on the Digit Symbol Coding test controlling for age, sex, race, education, and creatinine (among others); they found that low B12 and high MMA were associated with impairment (OR, 1.7; 95% CI, 1.01–2.9), but the primary importance of their study lay in the uncovering of the relationships among B12, folate, and cognition, with high levels of folate being protective with normal B12 (OR, 0.4; 95% CI, 0.2–0.9) but conferring five times the level of risk with low B12 (OR, 5.0; 95% CI, 2.7–9.5). Evidence also exists for the influence of high MMA levels on specific cognitive domains. Using regression modeling to allow for confounders, MMA has been associated with poor results on tests of matrix reasoning23 and with problems in tests of language expression, language comprehension, and praxis2; for example, there is a decrease of 3 points (95% CI, 1–5 points) on the language expression component of the cognitive part of the Cambridge Examination for Mental Disorders of the Elderly27 for each additional unit of MMA.2 Although clearly indicative of risk, the evidence above needs to be examined in longitudinal studies to avoid the possibility of reverse causation.


Only two longitudinal observational studies examining the role of B12 and MMA in cognitive decline could be found,28,29 possibly because examining MMA is labor intensive and consequently expensive.29 These were both robust, large-scale studies of community-dwelling older adults – one in England and the other in the United States – examining change over 10 years and 8 years, respectively. Each modeled the data to take confounding into account. Clarke et al.28 measured cognition at three time points using the MMSE and found that a doubling of MMA concentration was associated with a 50% faster rate of cognitive decline; likewise, a doubling of holoTC was associated with a 30% slower rate of cognitive decline. These results are endorsed in the study by Tangney et al.29 in which the rate of cognitive decline was significantly associated with higher MMA concentrations and lower levels of B12, having been similarly adjusted for confounding with other factors, such as vitamin markers, age, sex, and education.


The studies mentioned above provide robust evidence of the relationship between MMA, B12, and cognitive impairment and decline. The relationship between vitamin B12, MMA, and cognition is consistently in the expected direction, is consistent in different geographical settings (e.g., Canada, England, the United States, and Wales), shows strength of association, is independent of known confounders, and shows the risk or protective factor before the decline or preservation of cognition. Furthermore, the relationship is known to be biologically plausible and to be manifest in animal studies20 and in disorders of impaired synthesis of B12 in early age, such as cobalamin C disease.30,31 Given that many of the conditions for causality according to the Bradford Hill criteria are fulfilled, it could be expected that results from intervention studies would confirm the observational evidence.


Seven intervention studies examining the effect of supplementation of vitamin B12 on MMA and cognition in people aged 60 years and older were found.2037 All were placebo controlled; all administered a battery of cognitive tests examining many cognitive domains; and all measured cognition, B12, and MMA (among other biochemical measures) at baseline and completion. The studies were also similar in other regards. Most were conducted in Europe, with one in Hong Kong.32 Most were random samples of community-dwelling older adults, although two sampled volunteers through newspaper advertisements.33,36 While two studies specifically sampled people with low values of B12,32,33 four others reported that the respondents were deficient in B12.2037 Most used multivariate techniques to analyze their data, with two studies using univariate analysis.32,33 Six of the trials reported that the intervention had significantly reduced the level of MMA (and Hcy, and that all measures of vitamin B12 status were improved), but this was not detailed in the seventh.35 Despite this proof of biochemical principle, six studies found that the intervention did not affect cognition, while the seventh found improvement in the delayed recall of a verbal word-learning test only.33


Given that the intervention brought about a change in the biochemical measures, why has a relatively homogeneous group of trials failed to show an effect on cognition, especially since the weight of the observational evidence would indicate the contrary? It is possible that the trials were underpowered. Only one study reported a power calculation.36 Eussen et al.37 powered their study to detect a difference of 3 points in delayed recall of a 15-word learning test based on the work of van Asselt et al.,33 even though this was the only significant change in cognition that van Asselt et al. found, and the authors themselves state that “it cannot be excluded that the improvement represents a practice effect.”33 Given this, it may be optimistic to expect such a change, and Eussen's study may, thus, also be underpowered.

Most of the trials initially gave weekly doses of 1 mg of cyanocobalamin, followed by monthly administration. Although this is far in excess of the 0.006 mg that Bor et al.38 found to saturate all the vitamin B12-related variables, it is conceivable that the dosage of vitamin B12, although enough to correct high levels of B12 metabolites, may not be sufficient to make neurological changes. Doses of 2 mg per week have been reported to correct reversible myelopathy caused by B12 deficiency, albeit in a younger individual.39

All of the trials reviewed herein were of short duration, i.e., 6 months or less, with one trial lasting only 3 months. All of the authors, in critiquing their own work, stated that the duration of treatment was short. Pittock et al.39 reviewed resolution of neurological abnormalities in B12-deficient patients for periods of up to 3 years. Likewise, the longitudinal observational evidence cited earlier documented change over periods of 8 and 10 years.

An alternative explanation could be that the relationship is not actually true and that B12 deficiency and raised metabolites may be confounding the relationship between an associated unknown causal agent and cognitive impairment. Observational studies can take known confounders into account, but randomized trials, by virtue of their randomization, take into account unknown confounders. Thus, the potential exists that some unknown variable is the causative agent, but until trials of longer duration with higher dosages are undertaken, it is impossible to tell. Only when such trials are performed can gaps in the current methodology be redressed.


Declaration of interest.  The author has no conflict of interest. She received a small honorarium for writing this article. This work was commissioned by the Nutrition and Mental Performance Task Force of the European branch of the International Life Sciences Institute (ILSI Europe). Industry members of this task force are Abbott Nutrition, Barilla G. & R. Fratelli, Coca-Cola Europe, Danone, Dr Willmar Schwabe, DSM, FrieslandCampina, Kellogg Europe, Kraft Foods, Martek Biosciences Corporation, Naturex, Nestlé, PepsiCo International, Pfizer, Roquette, Soremartec – Ferrero Group, Südzucker/BENEO Group, Unilever. For further information about ILSI Europe, please call +32-2-771-00-14 or email: info{at}ilsieurope.be. The opinions expressed herein are those of the authors and do not necessarily represent the views of ILSI Europe. The coordinator for this supplement was Ms Agnes Meheust, ILSI Europe.


View Abstract