点击显示 收起
1 From the University Department of Psychiatry, Royal Liverpool University Hospital, Liverpool, United Kingdom (CM); the Department of Pathology, Wrexham Maelor Hospital, Wrexham, United Kingdom (PH); the University Hospital of Wales, Cardiff, United Kingdom (RE); and the Wales College of Medicine, Division of General Practice, Wrexham, United Kingdom (AM)
2 Supported by a grant from North Wales Research Committee. The epidemiologic investigations were funded by the Medical Research Council as part of the multicenter Cognitive Function and Ageing study (MRC CFA study).
3 Reprints not available. Address correspondence to A McCaddon, Gardden Road Surgery, Rhosllanerchrugog, Wrexham, North Wales, UK, LL14 2EN. E-mail: andrew{at}mccaddon.demon.co.uk.
See corresponding editorial on page 1259.
ABSTRACT
Background: An elevated blood concentration of homocysteine is an established risk factor for cognitive impairment and dementia, but associations between cognition and methylmalonic acid (MMA), a related metabolic marker of vitamin B-12 deficiency, are less clear.
Objective: The aim was to determine the utility of serum MMA and holotranscobalamin as markers of vitamin B-12 status in relation to cognitive function and to investigate their association with discrete cognitive domains.
Design: This was a cross-sectional survey of 84 nondemented elderly participants (aged >69 y) from the Welsh cohort of the Medical Research Council's Cognitive Function and Ageing Study. Cognitive status was determined by Mini-Mental State Examination (MMSE) and the Cognitive Section of the Cambridge Mental Disorders of the Elderly Examination (CAMCOG).
Results: Nearly one-half (43%) of the persons selected had likely metabolically significant vitamin B-12 deficiency. Higher MMA concentrations were associated with lower MMSE scores independent of age and education (P = 0.007). MMA concentration correlated inversely with CAMCOG scores of ideational praxis (P < 0.05) and language comprehension (P < 0.05) and expression (P < 0.01). Serum folate correlated weakly but significantly with language (P < 0.05), remote memory (P < 0.05), and constructional and ideational praxis scores (P < 0.05 and P < 0.01, respectively).
Conclusion: The high prevalence of likely metabolically significant vitamin B-12 deficiency in the elderly is associated with lower cognitive function scores and particularly with lower scores of language comprehension and expression.
Key Words: Vitamin B-12 • methylmalonic acid • holotranscobalamin • cognitive function • Mini-Mental State Examination • folate • elderly • aging • basal ganglia • language
INTRODUCTION
Vitamin B-12 is essential for 2 mammalian metabolic reactions–the conversion of methylmalonyl-coenzyme A to succinyl-conenzyme A and of homocysteine to methionine; the latter reaction is also folate dependent. Serum concentrations of methylmalonic acid (MMA) increase in vitamin B-12 deficiency, whereas homocysteine concentrations increase in both folate and vitamin B-12 deficiency. The advent of assays for these metabolites has facilitated the detection of early and "subtle" deficiencies of these vitamins (1).
Of these 2 metabolites, homocysteine has attracted interest with regard to cognitive function and aging. Elevated plasma homocysteine is associated with an increased risk of vascular disease and vascular dementia (2, 3), and there are widely confirmed reports of increased blood concentrations in patients with Alzheimer disease (AD) and mild cognitive impairment (4-7). Homocysteine concentrations predict the rate of cognitive decline in the healthy elderly (8-10), and hyperhomocysteinemia is an independent risk factor for the development of dementia, including AD (11, 12). However, not all studies confirm a direct relation between homocysteine and cognition (13-16).
Associations between cognition and MMA are less well documented (17-20). We present findings from a cross-sectional analysis of blood concentrations of MMA, vitamin B-12, and folate in 84 healthy elderly persons and report their relation with specific cognitive domains. To explore associations with vitamin B-12, we also determined the proportion of the vitamin bound to the vitamin B-12 carrier protein transcobalamin. This fraction [designated as holotranscobalamin (holoTC)] delivers the vitamin to tissues and is an early marker of changes in vitamin B-12 status (21).
SUBJECTS AND METHODS
Subjects
This sample of persons was part of the Medical Research Council's (MRC) Cognitive Function and Ageing Study (CFAS) cohort (www.cfas.ac.uk), a longitudinal multicenter study of persons aged 65 y who were randomly selected from population-based registers in each of 6 sites (22, 23). To establish that holoTC could be reliably measured from stored blood collected during CFAS, subjects were selected to be similar to an earlier study of holoTC intraindividual variability (24); that is, from nondemented persons who were living in their own homes in North Wales. We decided on 7 replicates of our original sample of 12 and so sampled 84 persons whose sex and age (to within 5 y) matched to the original sample. Any persons with a study diagnosis of dementia (AGECAT organicity score of O3 or above) were not included in the sample (25). AGECAT is a computerized diagnostic algorithm that uses scores from the Geriatric Mental State schedule to produce levels on each of 8 diagnostic syndromes, including organic disorder. These are then weighted to produce a differential diagnosis, which is highly concordant with a psychiatric opinion (25). The study received ethical approval from the North East Wales Local Research Ethics Committee.
Cognitive tests
Cognitive testing was conducted by trained interviewers using both the Mini-Mental State Examination (MMSE) (26) and an extended version of the Cognitive Section of the Cambridge Mental Disorders of the Elderly Examination (CAMCOG)—the cognitive battery of the Cambridge examination for mental disorders of the elderly (27). The MMSE consists of 19 items and tests the domains of orientation, language, attention, memory, and praxis with a maximum correct score of 30. The CAMCOG tests these same domains plus those of calculation, abstract thinking, and perception by using a total of 60 items. Within CAMCOG, language is examined by using 17 items: 9 for comprehension and 8 for expression. Education was recorded as the total number of years of formal education.
Blood sampling and laboratory methods
Collection of blood samples was carried out in collaboration with the European Prospective Investigation of Cancer group who have established an efficient laboratory procedure for the separation and safe storage of samples in liquid nitrogen (28). A 30-mL sample of blood was taken from MRC CFAS respondents in EDTA-coated, citrate-coated, or plain bottles, couriered overnight to a central laboratory, then separated and stored in –80 °C freezers as several separate aliquots of serum, plasma, red cells, and buffy coats.
For the present study, 3 aliquots of serum from each subject, identified by barcode, were packed in dry ice and sent to the University Hospital of Wales, Cardiff. On receipt, they were stored at –80 °C. Assays were performed within 6 to 7 y of the date of sample collection. Before assay, the aliquots were thawed at room temperature, pooled, and mixed thoroughly. HoloTC concentrations were estimated in singlicate by using a radioimmunoassay kit (Axis-Shield Diagnostics Ltd, Dundee, United Kingdom). The total volume available did not allow measurement in duplicate. Vitamin B-12 and folate concentrations were measured by using an immunoassay analyser (Beckman Coulter Inc, Chaska, MN). Because renal function is a determinant of MMA concentration (29), creatinine was measured by means of an Abbott Aeroset (Abbott Laboratories Ltd, Abbott Park, IL). MMA concentrations were quantified by using an in-house gas chromatography–mass spectroscopy method based on the one described by Rasmussen (30). "Likely metabolically significant" vitamin B-12 deficiency was defined as an MMA concentration greater than the laboratory upper limit of normality (0.42 µmol/L) in the absence of significant renal impairment (creatinine <120 µmol/L).
Statistical analysis
Descriptive statistics and ordinary least squares regression were carried out by using the Basic Statistics and Regression Analysis modules of STATISTICA version 6.0 (Statsoft Inc, Tulsa, OK). Regression analysis with missing cases in the dependent and independent variables was achieved by multivariate imputation with the program MICE version 1.14 (31). Regression parameters are presented with their associated SEs in parentheses. An R2 value for goodness-of-fit for the general linear models was calculated as (1 – residual dispersion)/null dispersion and expressed as a percentage.
RESULTS
The sample consisted of 42 men and 42 women with an age range from 69 to 93 y. Descriptive data are presented in Table 1. Cognitive scores from MMSE ranged from 18 to 30 with a median of 26. Due to our sampling criteria, no persons had a diagnosis of dementia. However, 31% (n = 26) of the sample could be considered cognitively impaired but not demented, with scores ranging from 18 to 23.
View this table:
TABLE 1. Description of the study population1
Seventeen persons (20%) were "vitamin B-12 deficient" by standard criteria, and 4 persons (5%) were "folate deficient" as defined by the laboratory reference range (Table 1). An additional 4 persons who had folate concentrations of 5.2 µmol/L had borderline folate deficiency. The holoTC concentration accounted for 9% to 80% of the total vitamin B-12 concentration.
MMA concentrations were obtained for only 76 patients; 8 samples had insufficient serum for analysis. Of the 42 persons with an elevated MMA concentration, 9 also had an elevated creatinine concentration. Thirty-three persons (43%) therefore had likely metabolically significant vitamin B-12 deficiency.
Relations between laboratory variables
No significant correlation was observed between vitamin B-12 and folate concentrations (Pearson correlation = 0.09, P = 0.43). Vitamin B-12 was highly correlated with holoTC (Pearson correlation = 0.7, P < 0.001). As expected, we found that high MMA related to low holoTC concentrations (Pearson correlation = –0.32, P = 0.005) and to low vitamin B-12 concentrations (Pearson correlation = –0.34, P = 0.003). No significant correlation was observed between MMA and creatinine concentration (Pearson correlation = 0.03, P = 0.8). Total vitamin B-12 concentration correlated inversely with the percentage of vitamin B-12 present as holoTC (Pearson correlation = –0.30, P = 0.006).
Relations between vitamin status and cognition
A significant correlation was observed between MMSE cognitive scores and MMA concentration. In a regression model including sex, age, education, creatinine, folate, and vitamin B-12, MMA, or holoTC, in turn, as independent variables, MMSE score decreased with increasing age and MMA [MMSE = 34.0 –0.12(0.05) x age – 2.42(1.02) x MMA; P = 0.007]. Age and MMA contributed 0.078 and 0.071, respectively, to the total R2 value of 0.22 for the regression analysis, indicating that each accounted for 7% of the variance in MMSE scores.
In a similar model with total CAMCOG score as the dependent variable, only age was significantly correlated with cognitive score (Table 2). CAMCOG data sets were incomplete for some persons (Table 1). Multivariate imputation allowed the inclusion of 7 case subjects with incomplete data. This showed effects for age and folate (Table 2).
View this table:
TABLE 2. Parameter estimates for models including vitamin B-12 or holotranscobalamin with the Cognitive Section of the Cambridge Mental Disorders of the Elderly Examination score as the dependent variable, before and after replacement of missing data by multivariate imputation1
To determine which cognitive domains in the CAMCOG were associated with folate and which with MMA, general linear models were constructed with each domain as a dependent variable. The independent variables sex, age, education, creatinine, and folate were included in each model. These were combined, in turn, with vitamin B-12, MMA, or holoTC, resulting in 3 separate models for each cognitive domain. The data were modeled in this way because of the significant correlations between these variables.
No significant relations were observed between cognitive domain scores and concentrations of either total vitamin B-12 or holoTC, nor were there any relations between the independent variables and CAMCOG scores for domains of orientation, learning memory, attention, or ideomotor praxis.
Scores of remote memory, recent memory, abstraction, and perception decreased with increasing age. Education and folate status were positively associated with remote memory retrieval (Table 3).
View this table:
TABLE 3. Parameter estimates of significant models of cognitive domain scores derived from the Cognitive Section of the Cambridge Mental Disorders of the Elderly Examination1
Increasing MMA concentrations were associated with lower scores for language comprehension and language expression (Table 3). The strongest correlation was between MMA and language expression score. Increasing MMA and decreasing folate concentrations were associated with lower scores of ideational and total praxis. Decreasing folate concentrations were also associated with lower scores of constructional praxis.
DISCUSSION
The high prevalence (43%) of likely metabolically significant vitamin B-12 deficiency in this sample confirms earlier findings in elderly subjects (32-35) and agrees with the observed prevalence in the British National Diet and Nutrition Survey of persons aged 65 y (36). This survey reported that 20% of persons had a subnormal vitamin B-12 concentration and 46% had elevated MMA, although the latter value was found only in institution-living participants. It is likely that elevated MMA in elderly subjects reflects genuine deficiency, because it can be lowered with vitamin B-12 supplementation (35-38). Similarly, the observed 5% prevalence rate for folate deficiency, with an additional 5% of persons being "borderline deficient," is consistent with previous reports of folate status in this non-fortified population (39).
HoloTC concentrations shared a similar distribution to our earlier study of biological variation of holoTC (24), which suggests that the CFAS protocol for blood collection and storage leads to suitable samples for holoTC analysis. We found that in some patients as much as 80% of total vitamin B-12 was bound to transcobalamin. This is probably a consequence of the low vitamin B-12 status in this population. Refsum et al (40) showed that the proportion of vitamin B-12 existing as holoTC changes according to vitamin status; at low concentrations, a higher proportion exists as holoTC.
Neuropsychiatric disorders due to vitamin B-12 deficiency are common, and measurements of serum MMA can be useful in the diagnosis of these patients (41). However, not all studies have confirmed a correlation between MMA and impaired cognitive function (42, 43), and MMA assays are rarely used in the diagnostic screening of such patients. This may be partly because it is unclear whether treating patients with established dementia and elevated MMA is beneficial (44, 45) or because MMA assays are not routinely available.
There are conflicting data as to whether MMA concentrations are elevated in patients with dementia. Kristensen et al (46) and Refsum and Smith (20) found significantly elevated concentrations in dementia patients, especially in those with a diagnosis of AD. Conversely, Campbell et al (47) found no significant difference in MMA in patients with dementia or cognitive impairment compared with cognitively intact elderly Latinos. In discussing their findings, Campbell et al note the interesting observation of a quadratic relation between serum vitamin B-12 and cognitive function scores, initially reported by Renvall et al (48) and also observed in an Italian centenarian population (15). These observations suggest a nonlinear relation between total vitamin B-12 and cognitive function, where deficiency is more closely associated with mild or moderate cognitive impairment than established dementia. The reason for this is unclear, but it perhaps explains the discrepancy between our findings and those of Campbell et al; our population was relatively cognitively intact.
We found that the association between MMA concentration and global cognitive score was largely mediated via its relation with language scores. The association between MMA and language was particularly strong for scores of language expression. Garcia et al (17) reported a significant correlation between MMA concentration and some scores of the California Verbal Learning Test, but this relation did not remain significant after a multivariate analysis. In contrast to our findings, in a recent Swedish study of 209 community dwelling healthy elderly subjects, the simultaneity index and block design correlated with serum MMA (18). However, this study focused on tests of psychomotor ability and mental speed rather than language-related skills.
A cross-sectional study of 116 elderly persons in Northeast Georgia found that those with high MMA had lower performance scores on tests relating to information processing speed, memory, verbal fluency, and nonverbal reasoning (19). However, in contrast to our own findings, the associations in this study were modest, especially in relation to other determinants such as age and education.
Our observations raise the question of what "biological substrate" may underlie the relation between MMA and language. Studies of MMA toxicity within the context of methylmalonic aciduria suggest that either MMA itself or closely related metabolites inhibit complex II of the electron transport chain of oxidative phosphorylation (49, 50). This appears to differentially affect brain tissue, with the basal ganglia and hippocampus being particularly vulnerable (50). Deficient basal ganglia function may explain the reported inverse correlation of MMA with speed and motor coordination (18). However, there is also evidence for a role of the basal ganglia in relation to language, probably in suppressing competing alternatives in the late "integrational" processes of language comprehension and in the lexicalization stage of language production (51). In addition to its effect on these subcortical structures, MMA is also an inhibitor of mitochondrial creatine kinase in the cerebral cortex (52).
Despite such potential mechanisms, the possibility remains that MMA is simply a marker for some other pathogenic factor, such as intracellular vitamin B-12 concentration, which plasma total vitamin B-12 or holoTC may not adequately represent, or homocysteine, which usually correlates with MMA. Homocysteine concentrations were not available for the present study. Sample collection and processing in CFAS precluded the determination of reliable homocysteine measurements; the samples were neither transported on ice nor processed within one hour of collection (53). We found no significant correlations between holoTC and cognitive scores despite previous reports of low holoTC concentrations in patients with AD and correlations between holoTC concentration and cognitive scores in the healthy elderly (20); our study perhaps lacked sufficient power to detect such relations.
In conclusion, the high prevalence of likely metabolically significant vitamin B-12 deficiency in the elderly appeared to be associated with impaired cognitive function. This was largely mediated by an association with impaired language comprehension and expression. Determination of MMA is a useful addition to epidemiologic studies of vitamin B status and cognition; assays of this metabolite can assist in differentiating between the effects of vitamin B-12 and folate on discrete cognitive domains.
ACKNOWLEDGMENTS
We thank Heather Wheatley for performing the MMA assays.
CM and PH participated in the concept and design of the study and in the statistical analysis and interpretation of the data. RE was responsible for organizing laboratory analyses. AM participated in the concept and design of the study and was responsible for initially drafting the article. All authors provided input into the final draft of the article. AM and PH are shareholders in COBALZ Limited, a private company developing combined vitamin B-12 and antioxidant supplements. None of the other authors has any conflict of interest to disclose.
REFERENCES
Related articles in AJCN: