METHODS

The study population has been described previously.³ From this cohort, we selected 51 cases of histopathologically confirmed Alzheimer’s disease and 65 cognitively screened controls where serum (non-fasting) was still available. Serum holoTC was directly determined, blind to diagnoses, by Dr L Örning (Axis-Shield Ltd, Oslo), using a novel solid phase radioimmunoassay.⁶ Serum total homocysteine was determined by high performance liquid chromatography⁷ and methylmalonic acid was measured by gas chromatography/mass spectrometry.⁸

Data analysis

All results are expressed as mean with 95% confidence intervals (CI), after adjustment for age and sex. Skewed variables were log transformed before analyses. Logistic regression analysis, adjusted for age and sex, was used to estimate the odds ratios of Alzheimer’s disease. Low vitamin levels or high total homocysteine and methylmalonic acid levels were defined as those below about the 20th centile or above about the 80th centile, respectively, in the controls. Linear regression analyses, adjusted for age and sex, were used to examine the relation between the vitamin variables (log transformed) and Alzheimer’s disease in the total group, and cognitive test scores in the control population.

RESULTS

Patient characteristics are shown in table 1. The patient group was about five years older than the control group, and the proportion of women was slightly lower. To control for these differences, we always adjusted for age and sex. The cognitive test scores and prevalence of ApoE ε4 were, as expected, much lower and much higher, respectively, in the patients than in the controls. Geometric mean levels showed no significant case–control differences for serum total cobalamin, but significant differences were observed for holoTC, total homocysteine, and folate, and less so for methylmalonic acid (table 1).

Table 2 shows the risk associations of the vitamin related variables with confirmed Alzheimer’s disease. While low total cobalamin was not significantly associated with Alzheimer’s disease, low holoTC and high methylmalonic acid were associated. Low folate and raised total homocysteine concentrations were also significantly associated with Alzheimer’s disease. Including ApoE ε4, smoking status, and creatinine in the model did not change the overall results (table 2). In a conditional logistic regression model including age, sex, and all five vitamin related variables, only raised total homocysteine and low folate remained significantly associated with Alzheimer’s disease (data not shown).

Splitting the dataset in two according to the median total homocysteine level showed that low holoTC was associated with Alzheimer’s disease at high total homocysteine concentrations (odds ratio = 9.45 (95% CI, 2.31 to 38.7)), but not at all at low total homocysteine concentrations (odds ratio = 0.91 (0.15 to 5.31)). In contrast, holoTC was not significantly associated with Alzheimer’s disease at either high or low folate concentrations. Spearman correlations and linear regression analyses (adjusted for age, sex, and apoE) confirmed that holoTC was associated with Alzheimer’s disease only at high total homocysteine concentrations (data not shown). Low folate levels also showed a much stronger association at high total homocysteine concentrations (data not shown).

In the control population, we examined the association between the vitamin related variables and the cognitive scores. After adjusting for age and sex, the CAMCOG (cognitive test from the Cambridge examination for mental disorders in the elderly) score was significantly associated with serum levels of both cobalamin and holoTC (r = 0.25, p = 0.04 for both variables), whereas the mini-mental state examination (MMSE) was significantly related to holoTC (r = 0.29, p = 0.021) but not to cobalamin (r = 0.24, p = 0.059). Plasma concentrations of total homocysteine, folate, and methylmalonic acid were not significantly associated with CAMCOG or MMSE scores.

DISCUSSION

Previous reports on cobalamin in relation to Alzheimer’s disease have shown inconsistent results.^9–13 Some studies have indeed been negative.¹⁴ However, these studies have all been based on total cobalamin measurement, usually in small study populations. Our data may explain the inconsistencies. We found that the holoTC level was more strongly associated than total cobalamin with Alzheimer’s disease and with cognitive function scores in controls. The finding that the association of holoTC with Alzheimer’s disease was only present in subjects with high total homocysteine implies that its association may be caused by the effect of low tissue cobalamin on the level of total homocysteine, or by an interaction between homocysteine and cobalamin.¹⁵ Hence our data are consistent with the hypothesis that a raised total homocysteine level is associated with Alzheimer’s disease.² In relation to the association we have found between low holoTC levels and Alzheimer’s disease, it will be of interest to see if polymorphisms in the holoTC gene that influence holoTC or total homocysteine levels are associated with an increased risk of Alzheimer’s disease.^16,17

Whether low cobalamin status is a cause or a consequence of Alzheimer’s disease cannot be determined from our results. However, it is noteworthy that there was an association between holoTC levels and cognitive test scores in the control group. A similar finding in elderly people has already been reported.¹⁸ We suggest that future studies on the potential role of total homocysteine and cobalamin status in relation to Alzheimer’s disease and cognitive decline should include holoTC measurement in addition to conventional total cobalamin determinations. In patients with cognitive impairment, plasma total homocysteine and holoTC are probably better markers of cobalamin function than total cobalamin measurement.

Our findings have direct clinical relevance as they provide evidence that subjects with Alzheimer’s disease often have impaired cobalamin status, something that can readily be corrected by treatment with vitamin B-12.