Serum vitamin D in patients with mild cognitive impairment and Alzheimer's disease

Abstract Objectives To determine the relevance of Mini‐Mental State Examination (MMSE), serum 25‐hydroxyvitamin D (25(OH)D3), and 1,25(OH)2D3 concentrations to mild cognitive impairment (MCI) and various stages of Alzheimer's disease (AD). Materials and Methods The study included 230 participants (>74 years) allocated to three main groups: 1‐healthy subjects (HS, n = 61), 2‐patients with MCI (n = 61), and 3‐ patients with Alzheimer's disease (AD) subdivided into three stages: mild (n = 41), moderate (n = 35), and severe AD (n = 32). The cognitive status was evaluated using MMSE. Serum 25 (OH)D3 (ng/ml) and 1,25(OH)2D3 concentrations (pg/ml) were determined by competitive radioimmunoassay. Results MMSE scores and 25(OH)D3 were decreased in MCI and all stages of the AD in both genders. MMSE variability was due to gender in HS (11%) and to 25(OH)D3 in MCI (15%) and AD (26%). ROC analysis revealed an outstanding property of MMSE in diagnosis of MCI (AUC, 0.906; CI 95%, 0.847–0.965; sensitivity 82%; specificity, 98%) and AD (AUC, 0.997; CI 95%, 0.992–1; sensitivity, 100%; specificity, 98%). 25(OH)D3 exhibited good property in MCI (AUC, 0.765; CI 95%, 0.681–0.849; sensitivity, 90%; specificity, 54%) and an excellent property in diagnosis of AD (AUC, 0.843; CI 95%, 0.782–0.904; sensitivity, 97%; specificity, 79%). Logistic analyses revealed that, in MCI, MMSE could predict (or classify correctly) with 97.6% accuracy (Wald, 15.22, β, −0.162; SE, 0.554; OR = 0.115:0.039–0.341; p = .0001), whereas 25(OH)D3 with 80% accuracy (Wald, 41,013; β, −0.213; SE, 0.033; OR = 0.808: 0.757–863; p = .0001). 25(OH)D3 was the only significant predictor for the severe AD and contributed to MMSE variability. Age and gender were significant predictors only in the moderate AD. In patients with MCI, 25(OH)D3 and 1,25(OH)2D3 were correlated men, but in case of the AD, they were correlated in women. Conclusions MMSE and serum 25(OH)D3 concentrations could be useful biomarkers for prediction and diagnosis of MCI and various stages of the AD. The results support the utility of vitamin D supplementation in AD therapy regimen.


| INTRODUCTION
Alzheimer's disease (AD) is the most common cause of cognition impairment in elderly populations. AD is characterized by dementia with progressive loss of memory, intellectuality, disturbance of language ability, impairment in social performance, and reduced independence (i.e., the need for caregiver support in daily life). Although a definitive diagnosis of AD can only be made based on histopathological examination of brain specimens, the clinical diagnosis of AD could have a high degree of accuracy if dementia is diagnosed using a cognitive score (Creavin et al., 2016;Votruba, Persad, & Giordani, 2016). In addition to age and gender, the Mini-Mental State Examination (MMSE) has been regarded as a useful instrument for evaluating the cognitive state of patients (Folstein, Folstein, & McHugh, 1975) and used as a predictor of AD (Musicco et al., 2009). Mild cognitive impairment (MCI) is subclinical complaint of memory function in elderly people. It has been reported that 10%-20% of individuals over the age of 65 years suffer from MCI (Petersen, 2011), with high potential of converting to AD (Devanand et al.,2008;Ganguli et al., 2011;Petersen et al., 2001;Ritchie & Touchon, 2000).
There are several measures that could distinguish AD from control subjects. These include decreased metabolism of fluorodeoxyglucose (Silverman et al., 2001), increased uptake of amyloid (Small et al., 2006), elevated levels of tau or its phosphorylated form, and decreased amyloid β42 in CSF (Hansson et al., 2006;Querfurth & LaFerla, 2010;Sunderland et al., 2003). However, these approaches are either invasive or very expensive. Therefore, there is still a need for developing diagnosis as well as treatment approaches to diseases characterized by dementia. Also, as to our knowledge, no study of possible use of both 25(OH)D 3 and 1,25(OH) 2 D 3 , separately or in combination with MMSE, as predictors in diagnosis and prediction of MCI and various stages (mild, moderate, and severe) of AD. Accordingly, this study was conducted to evaluate utility of MMSE, serum 25(OH) D 3, and 1,25(OH) 2 D 3 concentrations in prediction and diagnosis patients with MCI and the various stages of AD.

| Participants
A total of 230 individuals from Fukuoka University Hospital were included in this study. The participants were allocated to three groups: I-Healthy subjects (HS), II-patients with MCI and III-patients with AD main group classified, according to disease severity, into three stages defined as 1-mild AD, 2-moderate AD, and 3-severe AD. Diagnosis of MCI was performed according to Petersen's criteria (Petersen, 2011;Petersen et al., 2001;Ritchie & Touchon, 2000). The severity of cognitive impairment in patients with AD was evaluated using MMSE scores: mild AD (27 ≥ MMSE > 20), moderate AD (20 ≥ MMSE > 10), and severe AD (10 > MMSE) (Feldman, Van Baelen, Kavanagh, & Torfs, 2005;O'Bryant, Humphreys, & Smith, 2008). Each participant was clinically evaluated by set of tests that included questionnaire and a proxy interview, assessment of past and present illness, neurological and physical examinations, blood chemistry, and neuroimaging with computed tomography and/or magnetic resonance imaging. Some participants in the HS group had hypertension (eight of 33; 24%) and/ or hypercholesterolemia (four of 33; 12%), regarded above the baseline blood pressure (systolic 139/diastolic 89 mmHg) and cholesterol (219 mg/dl). All groups were gender-balanced except there were two times as many women as men in the moderate and severe AD groups.
This difference is consequent to grouping according to the clinical classification to MCI or AD. All participants were free of hepatic and renal disorders. The ethical permission for this study was obtained from the ethical committee of Fukuoka University Hospital. The study was performed in accordance with the ethical standards of the 1964 Helsinki Declaration. Written informed consent was obtained from all participants or their relatives prior to their participation in the study.
We excluded participants with any present or earlier history of vitamin D supplementation.

| Samples preparations and analyses
Peripheral blood was collected from each participant and centrifuged at 400 x g for 20 min. The sera obtained were stored at −80°C until use. Total serum concentrations of 25(OH)D 3 and 1,25(OH) 2 D 3 were determined by competitive radioimmunoassay using two respective antibodies. A 25-OH vitamin D 125 I RIA Kit (DiaSorin Inc. MN, USA) was used to assay 25(OH)D 3 . Briefly, after pretreatment of the samples with acetonitrile 300 to remove proteins, the sample extracts containing 25(OH)D 3 were incubated with 125 I-25(OH)D 3 and sheep anti-25(OH)D 3 antibody for 90 min at room temperature. Cellulose-conjugated anti-sheep IgG antibody was added to the precipitated reactive complex and free 125 I-25(OH)D 3 was removed by centrifugation. The radioactivity in each precipitate was assayed using a γ-counter (ARC-950, Hitachi-Aloka Medical Ltd, Japan), and concentrations were determined according to a standard curve. A 1,25(OH) 2 D 3 RIA Kit (Immunodiagnostic Systems Ltd, Boldon, England) was used to assay 1,25(OH) 2 D 3 . The principle of this assay system was the same as that above except a column technique was also employed to remove lipids during sample pretreatment.

| Statistical analyses
ANOVA one-way was conducted on the variables (age, MMSE, 25(OH)D3 and 1,25(OH)2D3) between groups (HS, MCI, mild AD, moderate AD and severe D), with gender as covariate, to detect the following: 1-the main effect, differences between the variables of the groups, 2-The groups within each gender, and each gender's variable between groups were compared to evaluate the effects of gender. Homogeneity was verified. Tukey's multiple comparisons post hoc test was applied whenever ANOVA detected significant differences. Bivariate correlations among the variables were evaluated by Pearson's correlation coefficient. As MMSE could be seen as both risk factor and outcome of the disease, a linear regression analysis was also conducted to determine the regression coefficients, statistical significance of regression model (t value), and proportion of MMSE (dependent) contributed by independent variables (age, gender, 25(OH)D3, and 1,25(OH) 2 D3) derived from the multiple correlation coefficient (Adjusted R 2 ).
The predictors were also tested with univariate logistic regression analyses to assess the contribution of each predictor alone to each group. Then, multivariate-forward selection analyses were conducted to assess the contribution of the predictors in combination to increase the statistical power and account for the individual differences in prediction. Variables which had a p value of >.05 were excluded.
The followings were calculated: β: logistic regression coefficient describes the size and direction of the relationship between a predictor and the disease (predictive value). Positive predictive value is the probability that a subject classified as a patient by the test belongs in the patient group becomes more likely as the predictor increases.
Negative predictive value is the probability that a subject classified as a nonpatient by the test belongs in the nonpatient group. It also indicates the inverse relationship between the predictor and the disease (decreased predictor means increased disease odd). Odd ratio (OR): the ratio of the odds, calculated as the exponent of β. OR is the measurement of likelihood and indicates that when the predictor is raised by one unit the odds ratio of the outcome increase by a factor equal to the OR value, that is, the odds of participants in the dependent variable (patients) increase by a factor equivalent to OR value with 95% confidence interval (CI). Correct classification, CC (accuracy rate (%) of the predictor to diagnose or distinguish two compared variables), and Wald value (significance of predictor contribution) were also measured.
Receiver operating characteristic (ROC) analysis provides useful information regarding the ability of a predictor to classify subjects into the relevant groups, and to compare the performance of more than one predictor. ROC was conducted to calculate area under the ROC curve (AUC), sensitivity, and specificity. Cutoff values at which optimal balance of sensitivity and specificity can be obtained were derived according to Youden Index. Sensitivity (with optimal 95% confidence interval) is the probability that a test result will be positive when the disease is present (true positive rate-the probability that a patient will be accurately classified by the test). Specificity (with optimal 95% confidence interval) is probability that a test result will be negative when the disease is not present (true negative rate-the probability that a nonpatient will be accurately classified by the test). The AUC is a measure of the efficacy of the test. The

| MMSE scores in patients with MCI and AD
The MMSE scores of HS women and men were 28.0 ± 1.9, 29.1 ± 0.9, respectively. Figure 1 shows that MMSE scores were decreased in MCI and AD. A significant difference for the main effect, between groups, was detected for MMSE (F(4,225) = 722.076; p = .000). There was no difference in the MMSE values between women (26.0 ± 2.4) and men (26.0 ± 1.8) with MCI. The MMSE scores were decreased in mild AD (women 23.9 ± 2.0, men 23.2 ± 1.6), moderate AD (women 16.1 ± 2.5, men 17.5 ± 2.0), and severe AD (women 4.6 ± 4.1, men 4.5 ± 3.4). The decrease in MMSE scores in AD was more than that observed in MCI in both genders (p = .000 for moderate and severe AD vs. MCI) except in mild AD (women, p = .030; men, p = .002 vs. MCI). In addition, significant differences (p = .000) were detected among the various stages of AD in women and men analyzed separately. However, no significant gender-dependent difference was detected for the same stage of AD between women and men when compared to each other.

| Serum 25(OH)D 3 concentrations in patients with MCI and AD
In HS, the mean serum concentrations of 25(OH)D 3 were 26.18 ± 7.18 ng/ml and 27.42 ± 8.05 ng/ml in women and men participants, respectively. A significant (F(4,225) = 25.869, p = .000) main effect of 25(OH)D 3 was obtained in MCI and AD.  (16.79 ± 5.32 ng/ml), and severe AD (13.95 ± 5.08 ng/ml) were significantly lower than HS (p = .000). On the other hand, it can be seen from Figure 2b that in the men patients, the concentrations of 25(OH)D 3 were significantly (p = .000) lower than HS in mild AD (17.59 ± 6.95 ng/ml) and severe AD (15.36 ± 4.08 ng/ml). However, no significant (p = .105) difference was detected between HS and the moderate AD (21.09 ± 6.32 ng/ml). No significant difference was detected among the AD stages for the same gender, or between the genders in each group.

| Univariate and multivariate logistic regression analyses
The prediction values were evaluated for each of MCI and AD groups against HS. Univariate analysis of each predictor alone (Table 4) shows that significant (p = .0001) prediction by age (

| DISCUSSION
The present results show that MMSE and 25(OH)D 3 (but not 1,25(OH) 2 D 3 ) were decreased in MCI and various stages of AD.
Although MMSE is one of the most widely used tools in the evaluation of cognitive status, there is still a debate about its diagnostic accuracy.
Some studies have reported that MMSE lacks diagnostic specificity and has limited diagnostic accuracy, particularly for distinguishing between normal cognition and MCI, and MCI from demential patients with AD (Chapman et al., 2016) and in measuring the progression of Alzheimer's disease (Clark et al., 1999). On the other hand, MMSE has been regarded as a good first step in the evaluation of cognitive status and effectively separating those with mild AD from normal aging and MCI (Benson, Slavin, Tran, Petrella, & Doraiswamy, 2005). USC B: unstandardized regression coefficient; SC Beta: standardized coefficient; R 2 : squared multiple correlation coefficient; SE: standard errors of the regression coefficients; Sig: two-sided observed significance levels (p) for the t statistics. CI: confidence interval; *Significant p values. The predictors 25(OH)D3, 1,25(OH)2D3 and gender (independent variable) were used to predict MMSE (dependent variable). et al., 1975). MMSE had also been reported to predict converters to AD (Devanand et al., 2008;Palmqvist et al., 2012). The present study highlights the value of MMSE and 25(OH)D 3 in the differential diagnosis and prediction of MCI, mild AD, moderate AD, and severe AD at a sensitivity rate >80. The differences among the results reported for MMSE could be attributed to the analyses approach such as selection of the cutoff values, and the patients' cultural, education, and demographic specificities. It is also noteworthy to mention that MMSE could be influenced by changes that could accompany dementia. A low level of the MMSE score is associated with low plasma phosphate (Haglin & Backman, 2016).
Vitamin D3 is produced in the skin from 7-dehydrocholesterol under the influence of UV light. Vitamin D is metabolized first to 25(OH)D 3 in the liver, then undergoes 1α-hydroxylation to the hormonal form 1,25(OH) 2 D 3 in the kidney (Bikle, 2014). The relation of 25(OH)D 3 and 1,25(OH) 2 D 3 is farther than that between a substrate and its product. 25(OH)D 3 and 1,25(OH)2D 3 are synthesized, regulated, and changed differently in variable diseases.
However, serum 25(OH)D3 is negatively correlated with TNFα, IL-1β or IL-6 levels in healthy subjects and patients with MCI, but positively with late-onset AD (Dursun et al., 2016). As TNF is increased in AD (Gezen-Ak et al., 2013), it is possible that the increased TNFα is responsible for the decreased 25(OH)D 3 . These mechanisms could play a role in maintaining 1,25(OH) 2 D 3 level against reduced 25(OH)D 3 .
The serum vitamin D level is associated with its activity in the brain. Serum 25(OH)D 3 concentration is correlated with regional cerebral blood flow (Farid et al., 2012), brain volume and gray matter thickness (Brouwer-Brolsma et al., 2015;Buell et al., 2010;Hooshmand, Lökk, & Solomon, 2014) and clearance of aggregated Aβ in the AD brain (Durk et al., 2014;Masoumi, Goldenson, & Ghirmai, 2009;Yu et al., 2011). On the other hand, low serum 25(OH)D 3 is associated with neuronal damages (Gezen-AK, Yilmazer, & Dursun, 2014), MCI (Annweiler et al., 2012), multidomain MCI (Yin et al., 2015) and AD (Annweiler et al., 2015;Balion, Griffith, & Strifler, 2012;Chei et al., 2014;Littlejohns, Henley, & Lang, 2014). The present results showed that 25(OH)D 3 is involved in the decrease of MMSE, and predict MCI, mild AD, and moderate AD. It was the only significant predictor of severe AD. It differentiated the disease from HS at a sensitivity >90%, but exhibited only a poor-weak diagnostic power when the evaluation was carried out among the stages of AD. It should be noted that although 25(OH)D 3 was decreased in MCI and AD, no difference was observed between women and men but it predicted a 2.5 times higher incidence of AD in women than in men. This result is in line with that reported 1.5-3 times higher incidence of AD in women than the incidence in men (Baum, 2005). Our results also showed that the decrease of 25 (