Homocysteine, Vitamin B12, and Folic Acid Levels in ... - IOS Press

Journal of Alzheimer’s Disease 27 (2011) 909–922 DOI 10.3233/JAD-2011-110752 IOS Press

909

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 Noel G. Fauxa , Kathryn A. Ellisa,b , Lorine Porterc , Chris J. Fowlera , Simon M. Lawsd,e , Ralph N. Martinsd,e , Kelly K. Pertilea , Alan Rembacha , Chris C. Rowef , Rebecca L. Rumblea , Cassandra Szoekeg , Kevin Taddeid,f , Tania Taddeid,e , Brett O. Trounsona , Victor L. Villemagnea,f , Vanessa Wardd,e , David Amesb,h , Colin L. Mastersa , the AIBL research groupi and Ashley I. Busha,∗ a The

Mental Health Research Institute, The University of Melbourne, Parkville, VIC, Australia of Psychiatry, The University of Melbourne, Academic Unit for Psychiatry of Old Age, Kew, VIC, Australia c Southern Health, Clayton, VIC, Australia d Centre of Excellence for Alzheimer’s Disease Research & Care, School of Exercise Biomedical and Health Sciences, Edith Cowan University, Joondalup, WA, Australia e Sir James McCusker Alzheimer’s Disease Research Unit (Hollywood Private Hospital), Perth, WA, Australia f Department of Nuclear Medicine and Centre for PET, The University of Melbourne, Austin Health, Heidelberg, VIC, Australia g CSIRO Molecular and Health Technologies, Parkville, VIC, Australia h National Ageing Research Institute, Royal Melbourne Hospital, Parkville, VIC, Australia i http://www.aibl.csiro.au/partners.html b Department

Accepted 13 July 2011

Abstract. There is some debate regarding the differing levels of plasma homocysteine, vitamin B12 and serum folate between healthy controls (HC), mild cognitive impairment (MCI), and Alzheimer’s disease (AD). As part of the Australian Imaging Biomarker Lifestyle (AIBL) study of aging cohort, consisting of 1,112 participants (768 HC, 133 MCI patients, and 211 AD patients), plasma homocysteine, vitamin B12, and serum and red cell folate were measured at baseline to investigate their levels, their inter-associations, and their relationships with cognition. The results of this cross-sectional study showed that homocysteine

∗ Correspondence to: Professor Ashley Bush, The Mental Health Research Institute, 155 Oak Street, Parkville, VIC 3052, Australia. Tel.: +6 13 9389 2914; Fax: +6 13 9387 5061; E-mail: [email protected].

ISSN 1387-2877/11/$27.50 © 2011 – IOS Press and the authors. All rights reserved

910

N.G. Faux et al. / Homocysteine in AD

levels were increased in female AD patients compared to female HC subjects (+16%, p-value < 0.001), but not in males. Red cell folate, but not serum folate, was decreased in AD patients compared to HC (−10%, p-value = 0.004). Composite z-scores of short- and long-term episodic memory, total episodic memory, and global cognition all showed significant negative correlations with homocysteine, in all clinical categories. Increasing red cell folate had a U-shaped association with homocysteine, so that high red cell folate levels were associated with worse long-term episodic memory, total episodic memory, and global cognition. These findings underscore the association of plasma homocysteine with cognitive deterioration, although not unique to AD, and identified an unexpected abnormality of red cell folate. Keywords: Alzheimer’s disease, biomarker, folate, homocysteine, vitamin B12

INTRODUCTION Alzheimer’s disease (AD) is the leading cause of dementia. There are estimated to be 36.5 million cases of dementia world wide, and its prevalence is predicted to double every 20 years to 115.4 million by 2050 [1]. AD is an incurable progressive condition, characterized by memory loss and cognitive decline, leading to death within 7–10 years after diagnosis [2]. A relationship between plasma homocysteine levels, cognitive performance, and the risk of developing AD has been repeatedly reported. Numerous studies have shown that homocysteine levels are positively correlated with cognitive impairment and decline [3–8], though some others report no such correlation [9], possibly due to a shorter follow-up time (2.7 years [9] compared with 5 years [7] and 9.5 years [8]). Plasma homocysteine levels have been reported to be elevated in AD compared to healthy subjects [4, 10, 11], although one study found no such difference [12] and another reported no difference between AD and HC. Additionally, homocysteine was elevated in vascular dementia [6] or in both vascular dementia and mixed AD/vascular dementia subjects [13]. These inconsistencies may be attributed to different fasting states, as well as small sample sizes. Some studies have also shown that elevated plasma homocysteine increases the risk for AD [10, 14] and dementia [14]. However, one study has shown that elevated homocysteine is a risk factor for vascular dementia but not for AD [15]. High levels of homocysteine could impair cognition by several possible mechanisms. Homocysteine can act as an agonist at the glutamate binding site of the NMDA receptor [16], inducing toxic calcium influx [17–19], and impairing long-term potentiation in the hippocampus, crucial for memory formation (reviewed in [20]). Also, increased homocysteine can inhibit S-adenosyl-1-methionine dependent methylation, leading to decreases in DNA and protein

methylation [21–26], which impair the metabolism of neurotransmitters and alter gene expression. Homocysteine also promotes toxic radical formation caused by amyloid-␤:copper complexes implicated in AD pathogenesis [27]. Homocysteine is an amino acid central to the biosynthesis of methionine, which requires both folate and cobalamin (vitamin B12). Levels of plasma homocysteine increase with age and inversely correlate with levels of vitamin B12 and folate in the blood [28], which are influenced by diet [29, 30]. There is some disagreement in the literature on differences in blood vitamin B12 and/or folate levels between healthy controls (HC) and AD patients. A number of studies have reported that vitamin B12 is lower in AD compared to HC [10, 31, 32], while others have not [11, 13, 33]. One study found a possible gender difference, reported a decrease in vitamin B12 in males AD patients, but not female patients [34]. Several studies have shown lower folate levels in AD compared to HC [10, 31, 35], though other studies have not detected a difference [11, 13, 34]. This disunity in the literature may be due to small cohort sizes, with most studies having less than 300 subjects. Recently a meta-analysis of a large number of case-control and cross-studies showed that in AD subjects, vitamin B12 and serum folate were lower compared to the HC subjects [36]. Thus, it is unclear if the apparent elevation of homocysteine in AD patients is a result of low vitamin B12 and/or folate levels, due to the AD disease pathology, or a result of another underlying condition, or dietary behavior. The Australian Imaging Biomarker Lifestyle (AIBL) Study of Aging [37] aims to further understand the causes of AD, to help identify biomarkers for diagnosis, and to inform the development of preventative strategies. Here we examine the AIBL cohort for the status of plasma homocysteine, vitamin B12 and folate (serum and red blood cell) levels in HC, mild cognitive


impairment (MCI), and AD participants, and examine the association between these values and cognition. MATERIALS AND METHODS Participant recruitment The recruitment and characteristics of this cohort have been previously described [37]. Briefly, the AIBL study is a two-site (Melbourne and Perth), multi-disciplinary, prospective longitudinal study of aging, integrating data from neuroimaging, biomarkers, clinical and neuropsychological measurements, and lifestyle information. Eligible volunteers were aged over 60 years, and fluent in English. The volunteers were recruited either in response to a media appeal or after being informed of the study by their treating physician. The volunteers were drawn from one of three groups: 1) individuals with AD as defined by NINCDS-ADRDA criteria [38]; 2) individuals with MCI (diagnosed according to a protocol based on the criteria of Winblad et al. [39] and Petersen et al. [40]), which is associated with an increased risk for the development of AD [39, 40]; and 3) cognitively healthy individuals (HC) (i.e., those participants who were not clinically classified as AD or MCI and did not meet the exclusion criteria). All participants’ clinical files were reviewed by a clinical review panel, comprised of psychiatrists, a neurologist, a geriatrician, and neuropsychologists. The AIBL study is compliant with, and is running parallel to, the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The AIBL study was approved by the institutional ethics committees of Austin Health, St. Vincent’s Health, Hollywood Private Hospital, and Edith Cowan University. Written informed consent was obtained from all study participants. Cohort size This cross-sectional study was conducted using data collected at baseline assessment on 1,112 participants (768 HC, 133 MCI, and 211 AD). Of these, 8 HC, 3 MCI, and 6 AD participants did not have results for homocysteine, vitamin B12, serum folate or red cell folate, due to either test failure or failed venepuncture. These missing samples were missing from each group equally (χ2 = 2.235, p-value = 0.327). The final cohort numbers were 760 HC, 130 MCI, and 205 AD (total of 1,095 participants).

911

Biochemistry The baseline blood samples were taken from overnight fasted participants and fractionated as previously described [37]. Aliquots of fresh blood were forwarded to a clinical pathology laboratory (Royal Melbourne Pathology in Melbourne, and PathWest Laboratory Medicine WA in Perth) for analysis. The list of tests included, but was not limited to, plasma homocysteine (IMMULITE 2000 - competitive immunoassay), serum folate, red cell folate (ADVIA Centaur Assay - chemiluminescent), and plasma vitamin B12 (ADVIA Centaur Assay - competitive immunoassay) levels. The stated reference ranges are the ranges established in the clinical pathology laboratory in accordance with the national guidelines (http://www.nata.asn.au/, http://www.health.gov.au/npaac). An aliquot of blood (0.5 ml) was also assayed for APOE genotype. Neuropsychological composite z-scores Four neuropsychological composite z-scores were calculated: 1. Short-term episodic memory composite z-score, composed of the following tests: logical Memory 1 – Story A [41], Californian Verbal Learning Task (CVLTII) [42] immediate recall, Rey Complex Figure Test (RCFT) [43] three minute delay. 2. Long-term episodic memory composite z-score, composed of logical memory II [41], CVLTII delayed recall, and RCFT 30 minute delay. 3. Total episodic memory composite z-score comprised of all the tests from both the short- and long-term episodic memory composite z-scores. 4. Global cognitive composite z-score, composed of the following tests, Stroop (Victoria version) incongruent trial [44], verbal fluency category fluency (fruit/furniture) switching [45], digit symbol coding [46], and the tests used for the Total episodic memory composite z-scores. The neuropsychological composite z-scores were generated by calculating individual z-scores for each test, then taking the average of the set of test z-scores. The individual z-scores were calculated as defined by equation 1. zi =

Txi − Mhc(T ) SDhc(T )

(1)

Where z is the z-score, x is the score for the neuropsychological test T, i is the ith participant, Mhc is

912


the mean of the HC for test T, and SDhc is the standard deviation of the HC group for test T. Folate dietary intake All participants were given the food frequency questionnaire (FFQs) developed by the Cancer Council of Victoria (CCV). The FFQ was developed as a means of assessing intakes of foods and nutrients in large epidemiological studies [47]. The FFQ [48] is a modification of a questionnaire that was developed in the late 1980 s to measure diet in the Melbourne Collaborative Cohort Study and has been designed specifically for Australian adults [49]. It is a 74-item self-administered questionnaire that is optically scanned, provides analysis of dietary intake, and has been validated relative to seven-day weighed food records in pre-menopausal women [50]. The CCV FFQ was completed on one occasion by participants attending their baseline visit. AD participants were assisted by their carer in completing the questionnaire. Respondents were asked to indicate how often they ate each of the 74 food items using the 10 frequency response options ranging from ‘Never’ to ‘ ≥ 3’ times per day. The FFQ also contains 3 photographs of scaled portions for four foods (used to calibrate a portion size), and questions on the overall frequency of consumption of foods such as bread. The 74 food items are grouped into four categories: 1) cereal foods, sweets and snacks; 2) dairy products, meats and fish; 3) fruit; and 4) vegetables. The responses were analyzed for usual daily nutrient intake and grams consumed (including but not limited to folate) at Cancer Council (Carlton, Victoria) using software based on the Australian nutrient database (NUTTAB95) [47]. As part of the medication use questionnaire, all participants were asked to also report if they were taking any supplements, i.e., herbal supplements, vitamins, etc., and the frequency, per day. To obtain the amount of folate in each of the reported supplements/vitamins, we collected the ingredients of the supplements (in mg) either from the manufacture’s web site or by contacting the manufacturer. These values were then added to the participants’ nutritional intake data to give the total nutrient intake. Brain imaging A total of 287 of the participants (25.8%) of the AIBL study underwent Pittsburgh compound B positron emission topography (PiB-PET) imaging using the methodology described previously [51, 52].

Briefly, 3D T1 MPRAGE and a T2 turbospin echo and FLAIR sequence MRI was acquired for screening and co-registration with the PET images. PET standardized uptake value (SUV) data acquired 40–70 minutes post C11 PiB injection were summed and normalized to the cerebellar cortex SUV, resulting in a region to cerebellar ratio termed the SUV ratio (SUVR). In this study, 14 of the participants in the PiB-PET subcohort failed venepuncture, thus the PiB-PET subcohort size for this study was 273 (24.5%).

Statistical analysis All statistical analysis was performed with R version 2.10.1 [53], using the following packages: effects [54], ggplot2 [55], MASS [56], and car [57]. Pearson’s Chi square (χ2 ) test was used to assess the differences in the distributions of the missing data, genders, and the APOE ␧4 carriers. Analysis of variance (ANOVA) test was used to assess the difference in age and SUVR ratio across the clinical classifications, followed by Tukey honest significant differences to test the pair-wise comparisons. The homocysteine, red cell folate, serum folate, vitamin B12 levels and the neuropsychological composite z-scores where checked for normality, through the inspection of the histograms and the quartile-quartile plots. For those that deviated from normality, BoxCox [58] analysis was performed. The transformed data were then checked for normality. For homocysteine and vitamin B12, natural log transformation was required. For both red cell folate and serum folate, square root transformation was required. The analysis of the difference between the clinical classifications and the continuous data, i.e., homocysteine, vitamin B12, serum folate and red cell folate, was by analysis of covariance (ANCOVA) with age, gender, and site as confounding variables. For the analysis of serum and red cell folate, dietary folate was also included as a confounding variable. The results presented are the adjusted means and standard errors (se). The pairwise analysis was corrected for multiple testing by controlling for false discovery rate [59]. The p-values presented are the adjusted p-values. To assess the relationships between homocysteine and red cell folate, as well as the relationships between homocysteine, serum and red cell folate with the four neuropsychological composite z-scores, multiple regression analysis was performed. For all of these analyses, age, gender, site, and clinical classification were included as confounding variables. The results


presented are the adjusted response values with the 95% confidence interval (CI). All initial models (ANCOVA and multiple regression) consisted of all terms and all possible interactions. Non-significant terms were initially removed through the use of the step procedure, which uses Akaike’s information criterion [60]. Further model simplification was performed manually by progressively removing terms which showed p-values ≥ 0.05. Model simplification stopped when all terms were significant or when the last term to be removed was the central hypothesis. Finally, the assumptions of regression analysis (homoscedasticity, normal distribution of the residuals and absence of influential observations) were tested by graphical inspection of diagnostic plots. For the analysis of serum and red cell folate against clinical classification, the distribution of the residuals for the original linear model deviated from normality. This was corrected through the use of generalized linear model with an inverse Gaussian error structure using a square root link function.

RESULTS Demographics of the AIBL cohort Table 1 shows the demographic makeup of the analyzed cohort. Across the entire cohort, there was a significant female bias (p-value < 0.001), and both the HC and AD groups individually exhibited a significant female bias (p-value < 0.001 and 0.002, respectively). There was a significant age difference between the groups (p-value < 0.001); the AD partici-

913

pants are significantly older than both the HC and the MCI participants (p-value < 0.001 and 0.003) and the MCI participants significantly older than the HC participants (p-value < 0.001). As expected there was a significant deviation from the expected ratios of APOE ␧4 carriers between the groups (p-value < 0.001); the AD group having the highest proportion of APOE ␧4 carriers. In concordance with the previous report of the AIBL PiB-PET subcohort [61, 62], there was a significant difference between the groups (p-value < 0.001), with the HC having significantly lower SUVR than MCI and MCI having significantly lower SUVR than AD (TukeyHSD, HC vs. MCI p-value < 0.001, HC vs. AD p-value < 0.001 and MCI vs. AD p-value < 0.001).

Differences in homocysteine levels in AD There is a significant difference in plasma homocysteine levels between the clinical classifications (Table 2; p-value = 0.001). There was a significant interaction between gender and clinical classification (F = 3.7753, p-value = 0.023). The age and site adjusted means differences between the clinical classification within the genders showed a significant difference between female AD and female HC (Table 3). Female AD patients had a 16% increase in plasma homocysteine levels compared to female HC subjects (p-value < 0.001). Furthermore, the adjusted means differences between females and males within the clinical classifications, revealed a significant difference between only male and female HCs (p-value < 0.001): male HC homocysteine levels are 14% higher than female HCs. Homocysteine levels of female AD participants are similar to male AD participants (Fig. 1A,

Table 1 Cohort demographics. Demographics, neuropsychological and biological descriptive statistics of the AIBL cohort

Age (years) Gender % female ApoE-␧4 pos. (%) SMem z-score LMem z-score TMem z-score Cog z-score PiB-PET (SUVR)

HC N(760) n(172)

MCI N(130) n(54)

AD N(205) n(47)

F/χ2

p-value

70 (7) 57.5 26.6 −0.01 (0.68) −0.01 (0.73) −0.01 (0.68) −0.01 (0.53) 1.45 (0.44)

75.7 (7.6) 56.9 50.4 −1.66 (0.68) −1.98 (0.77) −1.82 (0.71) −1.35 (0.52) 1.93 (0.63)

78.4 (8.7) 61 62.7 −2.48 (0.44) −2.82 (0.4) −2.64 (0.38) −2.26 (0.53) 2.32 (0.40)

119.91 28.61 104.02 1022.02 1195.45 1210.88 1254.06 69.82