Abstract
Low plasma amyloid-β (Aβ) is linked to Alzheimer’s disease. Since vitamin D cleared brain Aβ in vitro, this 8-week trial examined whether vitamin D increased plasma Aβ40. Vitamin D insufficient adults (6/18 M/F; 64.3 ± 10.9 y) were randomized to placebo or vitamin (50,000 IU/week) treatments. The vitamin group experienced greater plasma Aβ40 change than controls, +14.9 ± 12.0 and +12.8 ± 12.8 pg/mL (p = 0.045; effect size, 0.228). Change in Aβ40 for older participants (≥60 y) was +18.3 ± 33.6 and –3.2 ± 44.5 pg/mL for vitamin (n = 4) and placebo (n = 4) groups (effect size, 0.295). Thus, vitamin D may increase plasma Aβ, particularly in older adults, suggesting decreased brain Aβ.
INTRODUCTION
Alzheimer’s disease (AD) is a gradual neurodegenerative disorder characterized by neocortical and subcortical neurofibrillary tangles and amyloid-β (Aβ) plaques [1]. Aβ decreasing treatments are desired, but prior clinical trials have not been successful [2].
Plaques are composed of the isoforms Aβ42 and Aβ40 [3]. One mechanism by which Aβ is cleared involves active transportation across the blood-brain barrier via the abluminal lipoprotein receptor-related protein 1 and luminal P-glycoprotein (P-gp) [4]. In vitro and in vivo findings suggest that low P-gp reduces Aβ transportation to the periphery [5–10].
Vitamin D insufficiency has been previously associated with a higher AD risk [11]. Notably, Aβ clearance increased in vivo and in vitro after 1,25OHD3 administration, the active form of vitamin D [12, 13]. Interestingly, low plasma Aβ levels have been connected to an increased AD risk [14]. Thus, plasma Aβ concentrations might rise after vitamin D supplementation. We hypothesized that older, vitamin D insufficient and cognitively healthy subjects receiving vitamin D supplements would have a greater rise in plasma Aβ40 than controls after eight weeks.
MATERIALS AND METHODS
Study design
We undertook a parallel-arm, double-blinded randomized control trial design. Volunteers were recruited in Phoenix by electronic advertisement and word-of-mouth, and were screened for Mini-Mental State Examination (MMSE) scores greater than 26, 50+ years of age, serum total 25OHD less than 30 ng/mL, and the absence of diabetes, kidney disease, liver disease, statin usage, or blood disorders. Participants provided written consent, knowing that a placebo and not vitamin D might be received. Once randomized into groups, participants consumed either 50,000 IUs of cholecalciferol (Bio Tech Pharmacal; Fayetteville, AR) or a similar in appearance flour capsule once a week for eight weeks. Follow-up visits occurred within a week after the eighth supplement was consumed. The Arizona State University Institutional Review Board approved the protocol.
Measurements
Serum total 25OHD, plasma Aβ40, vitamin D intake (IU-d), and physical activity (METS/week) data were collected before and after the eight-week trial. Serum total 25OHD was quantified using liquid chromatography, tandem mass spectrometry (Sonora Quest Laboratories, Phoenix, AZ). Plasma Aβ40 was detected using a commercially available ELISA (Aβ40 Human ELISA Kit; Invitrogen; intrassay CV: 6.5%). All participants reported as fasting and had blood drawn into serum separator or sodium heparin vacutainers. After centrifugation, serum samples were transferred into a transport tube for Sonora Quest pick-up, and a protease inhibitor (Pefabloc) was immediately added to plasma samples, which were stored at –80°C and thawed once. Vitamin D intake was quantified using the Brief Vitamin D Questionnaire, and physical activity was measured using the Community Health Activities Model Program for Seniors questionnaire, both of which have been previously validated [15, 16].
Statistics
A univariate analysis of variance was conducted to compare by group the change from baseline in plasma Aβ40 and serum total 25OHD concentrations at the end of the eight-week trial. Three adjustment models were used to evaluate mean differences: Model 1 controlled for the respective baseline measurement; model 2 controlled for the respective baseline measurement and METs/week; and model 3 controlled for the respective baseline measurement, METs/week, and age. Model covariates were determined using Pearson, Spearman, and partial correlations. Change in Aβ40 as a function of change in 25OHD by age bracket was evaluated using Pearson correlations. Independent t-tests were used to examine differences in baseline values between the two groups. Skewed data were log-transformed for analyses. Data are reported as the mean ± SE. Effect sizes are included to aid interpretation (partial eta squared, ). Analysis was conducted using IBM SPSS Statistics v.22 (Chicago, IL).
RESULTS
General characteristics
Qualifying men (n = 6) and women (n = 18) were stratified by gender, age, vitamin D status, and BMI, and randomly assigned to the vitamin D group (n = 12; 9/3 F/M) or control group (n = 12; 9/3 F/M). One female participant who was randomized to the vitamin D group declined to initiate the study, citing concern regarding vitamin D status. Technical error analyzing plasma for Aβ40 reduced sample sizes to 10 in the vitamin D group (62.5 ± 3.4 y; 24.9 ± 1.3 kg/m2; 7 F) and 9 in the control group (64.3 ± 4.3 y; 27.1 ± 2.3 kg/m2; 7 F). Age, vitamin D intake, and BMI did not differ between groups at baseline, but physical activity tended to be higher in the vitamin D group compared to controls (1128.9 ± 219.5 and 575.3 ± 139.8 METs/week, respectively; p = 0.059).
Serum total 25OHD and plasma Aβ40
Serum total 25OHD and plasma Aβ40 concentrations did not differ between groups at baseline (Table 1). Following the eight-week trial, the vitamin D group mean change in serum total 25OHD was greater than controls by 26.8 ng/mL (Table: Model 1: = 0.902, p < 0.001; Model 2: = 0.886, p < 0.001; Model 3: = 0.903, p < 0.001). Concomitantly, the vitamin D group experienced greater mean plasma Aβ40 change than controls by 2.1 pg/mL. The effect of vitamin D supplementation on plasma Aβ40 change was significant when controlling for baseline plasma Aβ40, but adding physical activity and age to the model weakened the effect size and significance was lost (Table: Model 1: = 0.228, p = 0.045; Model 2: = 0.197, p = 0.085; Model 3: = 0.179, p = 0.116).
Correlation between 25OHD change and Aβ40 change
Figure 1 illustrates plasma Aβ40 change as a function of serum total 25OH change for (a) subjects younger than 60 years (n = 11) and (b) subjects 60 years or older (n = 8). There was no relationship between change values in the younger group (R2 = 0.001), whereas there was a strong association in the older group (R2 = 0.176). Moreover, change in plasma Aβ40 for the older subjects was +18.3 and –3.2 pg/mL for the vitamin D and placebo groups, respectively (= 0.295).
Correlations among baseline values and Aβ40 change
Five statistically significant relationships were observed: (1) baseline METs/week with baseline plasma Aβ40 (Spearman R = –0.83; p = 0.002); (2) age with plasma Aβ40 change (Spearman R = –0.48; p = 0.039); (3) baseline METs/week with plasma Aβ40 change (Spearman R = 0.60; p = 0.008); (4) baseline METs/week with age (Spearman R = –0.61; p = 0.008); and (5) baseline plasma Aβ40 with plasma Aβ40 change (Pearson R = –0.75; p < 0.001). Furthermore, the relationship between baseline plasma Aβ40 and plasma Aβ40 change remained significant independent to group difference (R = –0.82; p < 0.001).
DISCUSSION
Our findings suggest that vitamin D supplementation may enhance brain Aβ transportation to the periphery. Participants ingesting 50,000 IU weekly for eight weeks experienced a greater rise in plasma Aβ40 than participants ingesting a placebo. The effect was significant when controlling for baseline plasma Aβ40, but was attenuated after controlling for physical activity and age. Additionally, since we observed a strong correlation between Aβ40 change and total 25OHD change exclusively in participants of at least 60 years, older individuals might respond more from vitamin D treatment than younger individuals, a phenomenon future research should consider.
There are several strengths to this study, including its novel concept, exclusively vitamin D insufficient study population, younger study population, and blood-based measurements. There are also several limitations, such as the potential for plasma Aβ40 confounders (e.g., plasma protein binding, Aβ production from peripheral organs and platelets, hydration, and unknown covariance), questionnaire bias, and small sample size. Additionally, ApoE4, the allele associated with AD, was not tested, though those taking statins were excluded because of its association with elevated serum cholesterol [17]. All together, our findings should be interpreted as suggested.
More vitamin D research is justifiable because elderly, vitamin D insufficient individuals have a significantly higher AD risk and are susceptible to insufficiency [18]. Though perhaps more significantly, little is known about what modifies plasma Aβ, limiting its use as predictive marker. The power of studies that use plasma Aβ as an outcome measurement could be boosted if confounding covariates were consistently controlled. Interestingly, different protective mechanisms might lower or raise plasma Aβ. For example, while we found that vitamin D supplementation might raise plasma Aβ, one study found that higher exercise was associated with lower plasma Aβ [19]. In our study, the effect size of vitamin supplementation was weakened and significance was lost when physical activity was controlled.
In conclusion, we show the potential for vitamin D as a primary treatment strategy and the identification of plasma Aβ modifiers is necessary to maximize its utility.
