Vitamin B₁₂ and folate in relation to the development of Alzheimer’s disease

Methods.

Results.

Table 1 reports the number of subjects examined at baseline and prevalence (%) of low levels (defined with two cut-offs) of serum vitamin B₁₂ (≤150 and ≤250 pmol/L) and folate (≤10 and ≤12 nmol/L) by demographic characteristics and other relevant variables. After a 3-year follow-up, 78 (21.1%) subjects were diagnosed as dementia cases and 60 (16.2%) as dementia of Alzheimer’s type.

Table 2 shows the number of incident cases and relative risks of AD and dementia in relation to baseline vitamin levels. When the low levels were defined as vitamin B₁₂ ≤150 pmol/L, high crude and age-, sex-, and education-adjusted RR of AD were observed in relation to low levels of vitamin B₁₂. As adjustment of hemoglobin levels and alcohol consumption did not alter the above results, only age, sex, and educational attainment were included as covariates in the multiple Cox regression model. In contrast to AD, no relation was observed between dementia and low levels of vitamin B₁₂. When using vitamin B₁₂ ≤250 pmol/L to define low levels, 42 dementia and 37 AD cases had low B₁₂ levels, and 36 dementia and 23 AD cases had normal B₁₂ levels. The crude RR for AD in relation to low B₁₂ levels was 2.0 (95% CI = 1.2 to 3.3), and the adjusted RR was 1.8 (95% CI = 1.0 to 3.0). The crude RR for dementia was 1.5 (95% CI = 0.9 to 2.3), and the adjusted RR was 1.3 (95% CI = 0.8 to 2.1).

When the low levels were defined as folate ≤10 nmol/L, high crude and age-, sex-, and education-adjusted RR of both AD and dementia were detected in subjects with low serum folate levels (table 2). No significant changes in the results were found when hemoglobin levels and cardiovascular diseases were entered as covariates into the multiple Cox regression models, whether separate or together. Therefore, in the final model, only age, sex, and educational attainment were kept as covariates. When using folate ≤12 nmol/L to define low levels, 28 dementia and 22 AD cases had low folate levels, and 49 dementia and 37 AD cases had normal folate levels. Similar relative risks were found using the two definitions for the low levels. The crude RR for AD in relation to low folate levels was 1.7 (95% CI = 1.0 to 2.8), and the adjusted RR was 1.6 (95% CI = 0.9 to 2.7). The crude RR for dementia was 1.6 (95% CI = 1.0 to 2.5), and the adjusted RR was 1.5 (95% CI = 0.9 to 2.4).

Eleven subjects had low levels of both vitamins, and only one of them developed AD during the 3-year follow-up. Due to the small numbers, we grouped the subjects with low levels of either vitamin B₁₂ or folate together. High crude and adjusted RR for both AD and dementia were found in subjects with low levels of at least one of the vitamins (table 2). When low levels were defined as B₁₂ ≤250 pmol/L and folate ≤12 nmol/L, 15 dementia and 13 AD cases had low levels of both vitamins; 55 dementia and 46 AD cases had low levels of vitamin B₁₂ or folate; and 23 dementia and 14 AD cases had normal levels of both vitamins. Compared with subjects with normal levels of both vitamins, adjusted RR for AD in relation to low levels of at least one of the vitamins was 2.3, 95% CI = 1.2 to 4.2; adjusted RR for dementia was 1.7, 95% CI = 1.0 to 2.8.

The combined effects of the two vitamins on incidence of AD are reported in table 3. Using vitamin B₁₂ ≤150 pmol/L and folate ≤10 nmol/L to define low levels after adjusting for age, sex, and education, a clear increased risk of AD was found in subjects with low levels of both vitamins (model 1, third column). Compared with subjects who had normal levels of both vitamins, significantly high RR for AD were found for subjects who had low levels of at least one of the vitamins (model 2, third column). However, all the associations were no longer significant when cognitive functioning was taken into account (models 1 and 2, fourth column). A similar pattern was found when low levels were defined as B₁₂ ≤250 pmol/L and folate ≤12 nmol/L. The increased risk, associated with low levels of at least one of the vitamins, was modified by the MMSE (adjusted RR = 1.8, 95% CI = 1.0 to 3.4).

Due to the modifying effects of baseline cognitive functioning on the associations between low levels of the vitamins and development of AD, stratified analyses were performed by grouping subjects according to MMSE score. An MMSE score >26 was chosen to define individuals with good cognition. In comparison with subjects with normal levels of both vitamins, a clearly increased AD risk, associated with the presence of low levels in at least one of the vitamins, was found in individuals with good baseline cognitive functioning, independent of age, sex, and educational attainment. No such effect was found in subjects with weaker cognitive performance (MMSE score ≤26) ( table 4). When low levels were defined as B₁₂ ≤250 pmol/L and folate ≤12 nmol/L, the adjusted RR for AD was 7.0 (95% CI = 1.6 to 31.6) in subjects with MMSE score >26 and was 1.4 (95% CI = 0.7 to 2.7) in subjects with MMSE score ≤26. Comparable mean and median values were found across all groups in tables 3 and 4⇓.

To understand why the risk effect of low levels of vitamin B₁₂ or folate was present only in subjects with good cognition, we compared the living arrangements of subjects with good cognition and those with lower cognitive performance. Subjects with good cognition were more likely to live at home rather than in an institution (97% vs 3%), and subjects living at home, especially when living alone, may have a decreased intake of vitamins because of improper eating habits. Therefore, it was thought that living place might correlate with vitamin levels. We repeated the analysis by excluding those subjects who were living in an institution and found a similar risk effect of low levels of vitamin B₁₂ or folate in individuals with MMSE score >26 (RR = 2.7 [1.0 to 7.7] and 6.5 [1.4 to 29.5] for the lower and higher cut-off levels). In addition, subjects living at home had similar living arrangements (living alone or not) independent of their cognitive status. Finally, we compared individuals with MMSE scores ≤26 and >26 with respect to number of meals per day and number of cooked meals per day that they consumed. The results did not reveal any significant difference between the two groups (p > 0.85).

Discussion.

We examined low serum levels of vitamin B₁₂ and folate in relation to the incidence of AD and dementia in a community-based population of people aged 75 years and older who were nondemented and not treated with vitamin B₁₂ and folate. In both analyses where two different cut-off points for vitamin B₁₂ and folate were used to define low vitamin levels, neither low levels of vitamin B₁₂ nor folate alone significantly affected the risk of AD. However, as compared with subjects with normal levels of both vitamins, subjects with low levels of vitamin B₁₂ or folate had double the risk of developing AD, whichever cut-off level was chosen. This risk was even stronger in subjects with good cognition at baseline (MMSE score >26).

In the current study, we prospectively examined both vitamin B₁₂ and folate levels in relation to the incidence of AD in a community. The relationship between B₁₂ deficiency and dementia in elderly people aged 75 and older has been investigated previously in a longitudinal study.7 A cohort of 410 nondemented volunteers were followed for 5 years, and similar dementia incidences in subjects with low serum vitamin B₁₂ levels <150 pg/mL (corresponding to 110 pmol/L) (13.6%) and subjects with normal B₁₂ levels (14.7%) were reported. However, because of the small number of AD cases, this study could only examine dementia cases in general and not specifically AD.7

Formerly, low levels of serum folate were reported in subjects with AD,3 dementia,3 impaired cognition,13,14,32⇓⇓ and vascular diseases.12 We could not study the effects of low folate levels on vascular dementia because of the few cases of this dementia type. A risk effect of low folate levels on AD and dementia was detected as compared with normal levels. This finding is in line with Snowdon et al.’s finding that low serum folate was strongly associated with atrophy of the cerebral cortex.33

Two previous studies from the Kungsholmen Project found that people who had low levels of both vitamin B₁₂ and folate showed memory deficits,14 and no joint effect of vitamin B₁₂ and folate was found in any of the memory tasks.13 These results are in line with the findings from the current study.

Three main limitations of our study must be addressed. First, serum B₁₂ level is generally not regarded as a sensitive indicator of metabolic B₁₂ deficiency in the elderly.27,34⇓ When using vitamin B₁₂ radioimmunoassays, falsely high serum vitamin B₁₂ values may occur and mask a true deficiency,35,36⇓ and low levels may not correspond to B₁₂ deficiency at tissue level. Serum methylmalonic acid (MMA) and homocysteine levels have been suggested as more specific markers for vitamin B₁₂ and folate deficiencies.37-39⇓⇓ It has been reported that 76 or 80% of elderly individuals with elevated MMA and/or homocysteine levels indicating metabolic B₁₂ deficiency have serum B₁₂ level ≤258 pmol/L.29 The degree of elevation of metabolite concentrations was similar in subjects with serum B₁₂ ≤147 and 148 to 258 pmol/L. A more recent article suggested that 78% of subjects with serum cobalamin concentrations <140 pmol/L had deficiency at tissue level, whereas the majority of subjects with B₁₂ level equal to 140 to 258 pmol/L had no deficiency.40 Because of the disconcordant reports concerning the most appropriate cut-off for serum B₁₂ and folate deficiencies, we used two different cut-off levels. The lower cut-off level of 151 pmol/L is likely to include only true deficiency cases, and the higher cut-off level of 251 pmol/L might include all subjects with probable deficiencies. The associations were similar across the different cut-off points chosen. Similarly, low serum levels of folate are not considered to be perfect indicators of folate deficiency, and no specific limit for low serum folate levels in old age has been proposed so far. Therefore, we chose two cut-off points to define low levels of folate.13,29⇓ Because the imprecision in measuring vitamin deficiencies is likely to dilute the strength of the association, we do not regard this limitation as a major threat to our results.

Second, only one measurement of vitamin B₁₂ and folate levels was available. Therefore it was impossible to estimate the duration of vitamin deficiency, which may be related to development of dementia. Third, the detection of dementia among those subjects who died during the follow-up period was based on the examination of death certificates and hospital clinical records. This can lead to an underestimation of cases. It is likely that these limitations decrease the power of our study rather than bias the results. Despite this, we have found a clear association between low levels of at least one of the vitamins and AD incidence.

Biologic mechanisms related to our findings have already been proposed.17,34,41,42⇓⇓⇓ Vitamin B₁₂ or folate deficiency may increase the risk of AD because vitamin B₁₂ is necessary for the conversion of homocysteine to methionine, and vitamin B₁₂ or folate deficiency can increase homocysteine level due to slowed methylation reaction.17,42⇓ Homocysteine has a neurotoxic effect that could lead to cell death or neurologic and psychiatric disturbances such as AD.1,17,34⇓⇓ However, there is no evidence currently that vitamin B₁₂ or folate deficiency is associated with the neuropathologic hallmarks of AD. The lack of interaction between the two vitamins in relation to dementia occurrence may be explained by the common metabolic mechanisms of these two vitamins.42 The fact that we could detect a positive association only when the combined effect of the two vitamins were examined may be due to the power of the current study not being sufficient to detect a small effect. The multiple regression analyses further reduced the power of our study. For example, using normal levels of both vitamins as the unexposed group, to detect a relative risk of 2.1 for the effect of low levels of B₁₂, with α = 0.05 and 80% power, 525 subjects are required.

The fact that low levels of vitamin B₁₂ or folate increase the risk of AD, especially in subjects with good cognition, was an unexpected result. Why would subjects with high MMSE score be more prone to developing dementia if they have vitamin B₁₂ or folate deficiency? Eating habits such as number of meals per day and number of hot meals per day could not explain the observed associations. We can speculate that the effect of vitamin deficiency may be present only in the earliest prodromal phase of AD. This hypothesis could explain the inconsistent results from intervention studies in which these vitamins were supplemented.8-11,15⇓⇓⇓⇓ These studies involved a mixture of subjects with different levels of cognitive functioning. Further research concerning mechanisms underlying the observed relations are needed.