Nutrient biomarkers and vascular risk factors in subtypes of mild cognitive impairment: A cross-sectional study

Abstract

Objective

To investigate the interrelationships among blood nutrient biomarkers, the Framingham Stroke Risk Profile (FSRP), and cognitive impairment features in mild cognitive impairment (MCI) subjects and to verify whether nutrient biomarkers and FSRP are risk factors for MCI.

Methods

According to the criteria for MCI developed by Petersen, 81 subjects aged 50–80 years were divided into a normal control group (NC group, n = 36) and an MCI group (n = 45). Then, the MCI group was divided into an amnestic MCI (a-MCI) and a multidomain MCI (md-MCI) group. All subjects were administered a comprehensive health history to calculate their FSRP score and a thorough neuropsychological assessment of four cognitive domains. Blood samples from all subjects were collected to measure the nutrient biomarkers.

Results

FSRP score was not only associated with memory function, but also with executive function, which itself had a negative relationship with eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), and total n-3 polyunsaturated fatty acids (n-3PUFAs) levels, and a positive relationship with the ratio of n-6 PUFAs to n-3 PUFAs(n6/n3). Compared with the NC group, the concentrations of EPA, DHA, 25-hydroxy vitamin D (25OHD), and folate and the ratio of n3/n6 in the md-MCI group were significantly lower. In the a-MCI group, only DHA concentrations and the ratio of n3/n6 were significantly lower. After adjustment for potential confounding variables, low education level [Adjusted OR=8.71 (95%CI: 1.83–41.50), p trend = 0.007], decreased plasma 25OHD [Adjusted OR = 4.41 (95% CI: 1.08–17.94), p trend=0.04] and decreased plasma DHA [Adjusted OR = 6.69 (95% CI: 1.37–32.72), p = 0.02] were associated with a higher risk of MCI.

Conclusions

Several nutrient biomarkers in MCI patients, especially in md-MCI patients, were lower compared with healthy controls, suggesting that lower 25OHD and DHA levels are risk factors for MCI. However, we found no evidence that FSRP is an early biomarker of MCI.

Introduction

Mild cognitive impairment (MCI) is a syndrome defined as cognitive decline greater than expected for an individual’s age and education level, but that does not interfere notably with the activities of daily life (1). It is recognized as a high risk factor for dementia and especially Alzheimer’s disease (AD); many studies have shown that MCI tends to develop into clinically probable AD at a rate of about 10%–15% per year, whereas the rate of conversion to AD is only about 1% per year in normal older people (2, 3). This dementia increases the morbidity, disability, and healthcare costs of the growing elderly population, and decreases their quality of life and survival. However, there are currently no specific treatments.

Hypertension, diabetes, and other vascular risk factors are common among older adults, and it has been shown that these risk factors can affect brain structure (4) and have an association with the prevalence of cognitive impairment (5, 6, 7, 8, 9). The Framingham Stroke Risk Profile (FSRP) provides an estimate of the 10-year risk for future stroke based on age and the presence and severity of several cardiovascular risk factors (10). Previous studies have shown that a high FSRP score is positively related to lower cognitive function among stroke-free individuals (8, 11). Although many studies have shown that a high FSRP score is related to lower cognitive functions across cross-sectional (6, 8, 9) and longitudinal (7, 12) comparisons, some results are inconsistent. Moreover, no study has investigated the predictive power of FSRP score in MCI subgroups.

Given the substantial health and economic burden of MCI, its prevention and treatment are urgent and important. The association between diet and cognitive functions has a long history grounded not only in studies of brain development and physiology but also in studies of vascular risk factors (13). Diet is an important modifiable factor that has been shown to have direct and indirect influences on the development of cognition, which may lead to MCI and ultimately dementia (14, 15, 16). In recent years some studies have shown that nutrient biomarkers, such as: vitamin B12 (VitB12), folate, vitamin D (VitD), polyunsaturated fatty acids (PUFAs), have a close relationship with MCI (17, 18, 19, 20). However, the results of different studies are inconsistent, and the relationship between cognitive impairment and nutrient biomarkers in MCI subtypes is unknown.

In this study, we examined the nutrient biomarkers and used a modified version of the FSRP to explore the relationships among nutrient biomarkers, vascular risk factors, and MCI status and to verify whether nutrient biomarkers and FSRP are risk factors for MCI.

Methods

Participants

Overall, a total of 114 participants aged 50-80 years were recruited from the Department of Neurology of the Affiliated Zhongnan Hospital of Wuhan University and the Fifth Hospital of Wuhan between September 2012 and April 2013. Among them, 81 subjects met the inclusion criteria and provided full clinical data, of which 45 were MCI and 36 were normal control (NC). Exclusion criteria included history of incident stroke, abnormal renal function, current delirium, severe depression, cancer, and inability to understand or answer the study questionnaires. All included study participants received a full medical examination that consisted of structured questionnaires, information about education and comorbidities, a clinical examination, and a neuropsychological assessment. The study procedures were approved by the Medical Ethical Committee of the Zhongnan Hospital of Wuhan University. (Approval Number:2012083)

Diagnostic and neuropsychological assessment

A trained clinical neurologist interviewed patients and caregivers for the diagnosis of MCI or NC. All patients and NC underwent a thorough neuropsychological assessment. To obtain a global index of cognitive impairment, we administered Mini Mental State Examination (MMSE) (21), Montreal Cognitive Assessment (MoCA) (22), Clinical Dementia Rating (CDR) (23), Activity of Daily Living (ADL) (24). Cognitive domains were examined with the following most widely used tests: 1. Memory: Rey Auditory Verbal Learning Test (RAVLT) (25); Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) word recall (26); 2. Executive functions: Digit Symbol Substitution Test (DSST) (27); Trail-Making Test-Part B (TMT-B) (28); 3. Language: Verbal fluency test (VF) (29); 4. Visuo-spatial functions: Clock Drawing Test (CDT) (30). Hamilton Depression Rating Scale (HAMD) (31) used to evaluate the depressive symptom.

MCI was diagnosed according to Winblad et al. (32) consensus criteria: 1.there is evidence of cognitive deterioration, shown by either objectively measured decline over time or subjective report of decline by self or informant in conjunction with objective cognitive deficits; 2.a CDR score = 0.5; 3.the MMSE score is between 24 and 30; 4.the MoCA score is between 19 and 25; 5. Basic ADL score is normal; 6.failure to meet DSM-IV criteria (33) for dementia. Specifically, if patients’ nonmemory domains were intact, they were classified as amnestic MCI (a-MCI); if they had deficits in one or more nonmemory domains, they were diagnosed as multidomain MCI (md-MCI). Nondemented participants without MCI and who had normal neuropsychological and functional performance were considered as NC.

Plasma measurements

Peripheral venous blood was obtained from participants after overnight fasting. The plasma or serum was then removed and stored at –80°C until assay.

Serum folate and VitB12 was measured with a chemiluminescence-based assay on an Immulite analyzer (Beckman Coulter Inc., USA). Radioimmunoassay measured ethylenediaminetetraacetic acid (EDTA) plasma 25-OH vitamin D (25OHD) (DiaSorin Corp., Stillwater, Minnesota, USA). Gas chromatography equipped with a flame ionization detector quantified EDTA plasma fatty acid concentrations. Total plasma lipids were extracted from plasma with 2ml of methanolbenzene (4: 1, vol: vol) containing 0.2ml internal standard (heptadecanoic acid) by the classical Folch method (34). The lipids were then transesterified with Acetyl Chloride in the fume cupboard, and then at 100°C for 1h. The resultant fatty acid methyl esters (FAMEs) were extracted with 6% potassium carbonate. FAMEs were separated and quantified using a Varian-450 gas chromatograph equipped with a 30-m capillary column (internal diameter: 0.25mm). The injector temperature was set at 280 °C and the detector temperature at 250 °C. Column conditions were 150 °C for 1 min, increasing by 5 °C/min to 240 °C for 2min, then increasing by 20 °C/min to 290 °C for 10min. Helium was used as the carrier gas(flow rate: 0.8ml/min). FAMEs were identified based on the retention time to authentic lipid standards.

Definition of covariates or mediators

We collected self-reported data on sex, age and education (highest grade or years of formal education). Each participant’s weight while not wearing shoes was measured on a digital freestanding scale and height was measured with a height gauge. Body mass index (BMI) was calculated as weight/height2 in kg/m2. The FSRP (10) was calculated as an estimate of the 10-year risk of stroke. This incorporates age, systolic blood pressure, diabetes mellitus, current cigarette smoking, history of heart disease, atrial fibrillation, left ventricular hypertrophy (LVH), and the use of antihypertensive medication. Systolic blood pressure, diabetes, current cigarette smoking, history of heart disease and current use of antihypertensive medication were determined by interview. Atrial fibrillation and LVH were defined via electrocardiogram (ECG).

Statistical analysis

All statistical calculations were performed using SPSS 17.0. The subjects’ characteristics were summarized using means and standard deviations or frequencies, as appropriate. First, subjects were categorized into three groups based on their cognitive status (i.e., NC or a-MCI or md-MCI). Intergroup differences were analyzed by the chi-square test or one-way analysis of variance (ANOVA) as appropriate. Secondly, correlation between variables was evaluated by Pearson’s or Spearman’s correlation coefficient (r) as appropriate. Thirdly, unadjusted and fully adjusted logistic regressions were performed to predict the status of MCI (a-MCI and md-MCI) from nutrient biomarkers concentration as a tertile variable. The sample size of 81 subjects assumed accounting for up to eight covariables in multivariate analyses. Fourthly, univariate and multiple (i.e., fully adjusted and stepwise backward models) logistic regressions were used to examine the association between the tertiles of nutrient biomarkers (independent variable) and MCI status (dependent variable) whilst taking the potential confounders into account. P-values <0.05 were considered as significant.

Standard protocol approvals, registrations, and patient consents

All participants signed written informed consent forms. The study procedures were approved by the Medical Ethical Committee of the Zhongnan Hospital of Wuhan University (Approval Number:2012083). And the study has been registered in the web of Chinese Clinical Trial Registry (Registration number is Chi CTR-OCS-12002876).

Results

Clinical characteristics

Full clinical data were available for 81 subjects in our study. They were divided into the following groups: NC, a-MCI and md-MCI, according to cognitive assays. As indicated in Table 1, the demographic differences among these groups were not significant (P > 0.05) except for their education level. Among these groups, the md-MCI group had the highest FSRP score and the NC group had the lowest score, but the difference was barely significant (P = 0.052).

Table 1.

Clinical characteristics

Clinical characteristics	NC (n = 36)	MCI (n = 45)		P-valuea
		a-MCI (n = 19)	md-MCI (n = 26)
age (years)	62.67 ± 7.94	65.16 ± 9.04	68.15 ± 9.52*	0.056
gender (male/female)	14/22	11/8	14/12	0.317
BMI (kg/m2)	24.07 ± 3.29	23.19 ± 3.04	23.64 ± 3.87	0.659
Education level (years)	13.33 ± 2.53	11.89 ± 3.81	9.15 ± 4.66*	0.001
FSRP	8.11 ± 4.10	9.68 ± 4.88	10.85 ± 4.23*	0.052

Data are presented as mean ± SD. * P < 0.05 as compared with NC group; aComparisons are based on chi-square test or ANOVA as appropriate; a-MCI, amnestic MCI; BMI, Body Mass Index; FSRP, Framingham Stroke Risk Profile; MCI, mild cognitive impairment; md-MCI, multidomain MCI; NC, Normal control

Neuropsychological assessment

As shown in Table 2, patients with cognitive impairment had lower scores in Moca, RAVLT(immediate recall, delayed recall, right recognition), ADAS-Cog, DSST, TMT-B, VF assessments compared with the NC group (P < 0.05). Patients with cognitive impairment had higher scores in RAVLT false recognition and TMT-B than control subjects (P < 0.05). The md-MCI patients had lower scores in CDT and MMSE assessments compared with the NC group (P < 0.05) but not the a-MCI group. Compared with the a-MCI group, the md-MCI group had significantly lower scores in MMSE, Moca, DSST, and CDT, but higher scores in TMT-B.

Table 2.

Neuropsychological assessment of subjects

Cognitive functiona		NC (n = 36)	a-MCI (n = 19)	md-MCI (n= 26)
Global cognitive function	MMSE	28.47 ± 1.03	27.79 ± 1.32	25.88 ± 2.41*&
	Moca	26.19 ± 1.09	22.79 ± 1.75*	20.54 ± 2.80*&
Memory functions	RAVLT, immediate	41.53 ± 7.38	28.53 ± 10.91*	28.42 ± 9.18*
	RAVLT, delayed	9.17 ± 2.75	4.63 ± 2.97*	5.38 ± 2.59*
	RAVLT, right recognition	14.47 ± 0.74	12.53 ± 1.98*	12.31 ± 2.77*
	RAVLT, false recognition	0.97 ± 1.34	3.26 ± 2.58*	2.58 ± 2.35*
	ADAS-Cog	23.42 ± 2.87	18.00 ± 3.11*	17.15 ± 3.02*
Executive functions	DSST	34.92 ± 7.74	26.58 ± 6.55*	19.77 ± 6.41*&
	TMT-B	76.89 ± 26.19	106.68 ± 31.23*	150.46 ± 61.51*&
Language functions	VF	43.17 ± 7.06	37.37 ± 7.88*	34.12 ± 5.93*
Visuo-spatial functions	CDT	3.94 ± 0.23	3.89 ± 0.32	2.92 ± 1.29*&

Data are presented as mean ± SD. * P < 0.05 as compared with NC group; &P < 0.05 as compared with a-MCI group; aComparisons based on ANOVA test; ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive Subscale; ADL, activities of daily living; a-MCI, amnestic mild cognitive impairment; CDR, clinical dementia rating; CDT, clock-drawing task; DSST, Digit Symbol Substitution Test; md-MCI, multidomain MCI; MMSE, Mini-Mental State Examination; Moca, Montreal Cognitive Assessment; NC, Normal control; RAVLT, Rey Auditory Verbal Learning Test; TMT-B, Trail-Making Test-Part B; VF, Verbal fluency test

Plasma concentrations of nutrient biomarkers

Plasma nutrient biomarkers are presented in Table 3. For the vitamins, the 25OHD and folate concentrations were lower in the md-MCI group compared with the control group (P < 0.05), but there was no difference in VitB12 concentrations. For the PUFAs, EPA concentrations in the md-MCI group were significantly lower than in the NC group (P < 0.05), and the patients with cognitive impairment had lower DHA concentrations (P = 0.050). There were no significant differences for any n-6 PUFAs (P > 0.05), but the ratio of n-6/n-3 in both the a-MCI group and the md-MCI group was higher than in the NC group (P < 0.05).

Table 3.

Plasma nutrient biomarkers concentration

Plasma nutrient biomarkers	NC (n = 36)	a-MCI (n = 19)	md-MCI (n = 26)	Pa
25OHD (ng/ml)	22.66±14.28	19.76±13.62	14.43±6.54*	0.036
Folate (mmol/L)	9.57±5.13	8.79±3.25	6.67±2.66*	0.008
VitB12 (mmol/L)	377.39±205.68	367.42±173.19	308.12±112.35	0.175
PUFAs (mg/L)
ALA	15.38±6.86	14.41±4.24	16.66±6.78	0.489
E PA	19.16±1.21	16.59±14.44	12.89±4.7*	0.022
DHA	29.42±9.40	24.17±9.65*	2448±8.21*	0.050
Total n-3 PUFAs	63.97±23.97	55.18±22.32	5403±16.34	0.147
LA	166.85±83.85	165.31±52.36	170.68±59.67	0.659
AA	67.96±2456	71.75±25.29	72.53±26.28	0.752
Total n-6 PUFAs	234.81±98.76	237.06±68.57	243.21±71.55	0.926
n-6/n-3 ratio	3.92±1.32	468±1.49*	4.61±1.01*	0.046

Data are presented as mean ± SD. * P < 0.05 as compared with NC group; aComparisons based on ANOVA test; 25OHD, 25-hydroxyvitamin D; AA, arachidonic acid; ALA, α-linolenic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; LA, linoleic acid; PUFAs, polyunsaturated fatty acids; VitB12, vitamin B12

Correlations between cognitive function and FSRP

According to the data supplied by the subjects, FSRP score was positively related to age (r = 0.788, P = 0.000) (data not shown). Figure 1 indicates that there was a negative correlation between FSRP and RAVLT immediate recall (r = –0.248, P = 0.026), RAVLT delayed recall (r = –0.334, P = 0.002), RAVLT recognition (r = –0.309, P = 0.005) and DSST (r = –0.331, P = 0.003). There was no correlation between FSRP and the other cognitive tests (P > 0.05, data not shown)

Figure 1.

FSRP and cognitive function (n = 81)

Correlations between plasma nutrient biomarkers and FSRP

Figure 2 indicates that there was a negative correlation between FSRP score and the plasma levels of DHA (r = –0.267, P = 0.016), EPA (r = –0.265, P = 0.017) and n-3PUFAs (r = –0.274, P = 0.013). FSRP score was positively correlated with the ratio of n-6/n-3PUFAs (r = 0.321, P = 0.003). There was no correlation between FSRP score and the other nutrient biomarkers (P > 0.05, data not shown).

Figure 2.

FSRP and plasma nutrient biomarkers (n = 81)

Correlations between cognitive function and plasma nutrient biomarkers

Univariate logistic regression was used to select those nutrient biomarkers that significantly predicted cognitive function. These biomarkers were 25OHD, EPA, DHA, n-6/n-3 ratio. As Table 4 shows in the univariate models, compared with the highest tertiles, the lowest tertiles of 25OHD (P = 0.02), EPA (P = 0.03) and DHA (P = 0.03) were associated with a higher risk of MCI. The lowest tertile of the ratio of n-6/n-3 was associated with a lower risk of MCI (P = 0.002). There was no association between the second tertile of these biomarkers and MCI status. In addition, age (P = 0.049) and education level (P = 0.005) were also associated with a higher risk of MCI. In the model adjusted only for demographics (data not shown), the associations between age (adjusted OR = 1.97, 95%CI: 1.12–3.47, P = 0.019), education level (adjusted OR = 8.20, 95%CI: 2.10–31.99, P = 0.002), 25OHD (adjusted OR = 4.92, 95%CI: 1.32–18.34, P = 0.018), DHA (adjusted OR = 4.56, 95%CI: 1.05–19.77, P = 0.043), and the n-6/n-3 ratio (adjusted OR = 0.19, 95%CI: 0.05–0.70, P = 0.013) and MCI status remained significant. After added adjustment for FSRP score, the associations between age and DHA and MCI status became insignificant (P > 0.05). Additionally, the education level and the tertiles of plasma 25OHD, DHA, and the n-6/n-3 ratio were selected by a stepwise backward logistic regression model to explain MCI status, but the association for the n-6/n-3 ratio became insignificant (P > 0.05).

Table 4.

Univariate and multiple logistic regressions showing the cross-sectional association between MCI status[a](dependent variable) and plasma nutrient biomarkers concentrations (independent variable) adjusted for subjects’ characteristics (n = 81)

		Unadjusted model		Fully adjusted model[b]			Stepwise forward model
	OR	95%CI	P-value	OR	95%CI	P-value	OR	95%CI	P-value
Age (+10y)	1.67	1.003-2.79	0.049	1.55	0.69-3.48	0.29		——
Gender								——
male		Ref			Ref
female	0.51	0.21-1.24	0.14	2.25	0.83-6.11	0.110
Education level
College or above		Ref			Ref			Ref
High school	2.06	0.72-5.92	0.18	2.24	0.72-7.03	8.56	1.28	0.35-4.67	0.71
Middle school or below	6.12	1.73-21.61	0.005	8.56	2.15-34.04	0.002	8.71	1.83-41.50	0.007
FSRP								——
Q3 (≥13)		Ref			Ref
Q2 (8, 9, 10, 11, 12)	1.85	0.66-5.19	0.25	0.80	0.20-3.29	0.76
Q1 (≤7)	2.86	0.89-9.19	0.08	0.47	0.07-3.12	0.43
25OHD
Q3 (≥22.47)		Ref			Ref			Ref
Q2 (12.57-22.47)	2.47	0.827-7.393	0.11	1.66	0.47-5.84	0.43	3.12	0.80-12.17	0.10
Q1 (≤12.57)	4.04	1.30-12.59	0.02	5.37	1.41-20.50	0.01	4.41	1.08-17.94	0.04
EPA								——
Q3 (≥16.54)		Ref			Ref
Q2 (10.66-16.54)	2.89	0.96-8.72	0.06	3.09	0.84-11.42	0.09
Q1 (≤10.66)	3.40	1.11-10.40	0.03	3.10	0.81-11.84	0.10
DHA
Q3 (≥30.54)		Ref			Ref			Ref
Q2 (22.36-30.54)	1.82	0.62-5.35	0.28	2.04	0.54-7.68	0.29	2.01	0.47-8.51	0.35
Q1 (≤22.36)	3.46	1.12-10.67	0.03	3.96	0.95-16.56	0.06	6.69	1.37-32.72	0.02
n-6/n-3 ratio
Q3 (≥4.57)		Ref			Ref			Ref
Q2 (3.75-4.57)	1.00	0.31-3.22	1.00	0.93	0.26-3.28	0.91	2.00	0.49-8.23	0.34
Q1 (≤3.75)	0.15	0.05-0.49	0.002	0.20	0.05-0.75	0.02	0.30	0.07-1.24	0.10

25OHD, 25-hydroxyvitamin D; CI, confidence interval; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; FSRP, Framingham Stroke Risk Profile; n-6/n-3 ratio, ratio of n-6 PUFAs to n-3 PUFAs; OR, Odds ratio; a. Based on Winblad et al. consensus criteria; b. adjusted on age, gender, education and FSRP.

Discussion

To our knowledge, this is the first study to specifically investigate nutrient biomarkers and vascular risk factors in a sample group of elderly without dementia. This study indicates that among those elderly without dementia, memory functions and executive functions are correlated with FSRP score, which itself associated with changes in the plasma levels of EPA, DHA, and total n-3PUFAs. Furthermore, lower vitamin D, EPA, DHA, folate levels and the n6:n3 ratio in MCI patients, especially in md-MCI patients, were lower compared with healthy control subjects. Likewise, we found that lower 25OHD and DHA levels may be risk factors for MCI. However, we found no evidence that FSRP score is an early biomarker of cognitive impairment.

Increasing evidence has shown that there is an association between FSRP score and cognition (6, 7, 8, 9, 11, 12). In a large national sample (n = 23,752) that was stroke-free and cognitively normal at baseline, followed for an average of four years and culled of participants who developed clinical stroke in the 4-year interval, Unverzagt et al. (11) found that FSRP score was linearly related to the rate of incidence of cognitive impairment. Other studies have shown that increased stroke risk as measured by total FSRP score is related to lowered functions of cognitive domains cross-sectionally (6, 7, 8, 9) and longitudinally (7, 12). One cross-sectional study (8) of 7377 adults aged 50 years or above found that FSRP was associated with immediate verbal memory, delayed verbal memory, semantic verbal fluency, and processing speed. In contrast, a cross-sectional study (6) of 2175 participants found that FSRP was associated with visual-spatial memory, attention, organization, scanning, and abstract reasoning. Among 1755 Framingham Offspring participants, Nishtala et al. (9) found that FSRP was only associated with executive function. The longitudinal study (12) consisted of 235 men who were stroke- and dementia-free at baseline and who were reassessed on a cognitive battery 3 years later. FSRP score was inversely related to verbal fluency but not word list learning, word list recall, pattern comparison, or digit span. While Kaffashian et al. (7) found that FSRP score was associated with faster decline in verbal fluency, vocabulary, and global cognition, but no association was observed for memory and reasoning among 4153 men and 1657 women (mean age, 55.6 years at baseline) in the Whitehall II study. We found that FSRP score was correlated with memory functions and executive functions. Although the results of these studies are different, we can nonetheless conclude that the FSRP indeed plays a vital role in cognitive decline. To the best of our knowledge, our study is the first study to investigate FSRP score in MCI subgroups, and we found that the md-MCI group had higher FSRP scores than the a-MCI group, whose FSRP score was higher than NC group. Because of a limited number of participants, this difference was barely significant (P = 0.052). Despite this, we suggest that the higher the FSRP score, the greater chance an individual will be in the md-MCI subgroup. It has been proposed that it may be useful to distinguish specific subtypes of MCI (35). For example, subjects in the a-MCI subgroup may be at increased risk of developing AD, whereas subjects in the md-MCI subgroup may be at increased risk of developing VaD (35). This proposal has not yet been tested in a clinical setting, but a population-based study failed to find support for the proposal (36). However, Rasquin et al. (37) found that being in the md-MCI subtype has a high sensitivity for identifying people at risk for developing AD or VaD. Moreover, previous researches indicate that patients with md-MCI are more likely to progress to dementia (38, 39). Therefore, we can propose that the higher the FSRP score, the higher the risk of dementia, especially vascular dementia, but more evidence is needed to be conclusive. In our study, we failed to find that FSRP can be an early biomarker of cognitive impairment. In both the univariate and multivariate models, there was no association between the lower tertile of FSRP score and a higher risk of MCI (P > 0.05). In contrast to our results, in a 10-year follow up to the Whitehall II cohort study, Kaffashian et al. [7] found a greater decline in the highest FSRP quartile compared with the lowest risk quartile (P = 0.03) for global cognition. They proposed that elevated stroke risk at midlife is associated with accelerated cognitive decline over 10 years. Nevertheless, the Whitehall II study could not definitely establish that a high FSRP score is a risk factor for MCI.

In comparisons between healthy controls and MCI, we found differences in plasma nutrient status, especially in the md-MCI group. Despite mounting evidence of an association between nutrient status and cognitive impairment in the elderly, few studies have examined the relationship between nutrient status and MCI, and their results are inconsistent. Moreover, no study has examined the role of nutrient status in MCI subtypes. In the two prospective longitudinal studies (40, 41), high PUFA intake appears to be protective against the development of MCI. Evidence among cross-sectional or case-control studies, however, is less clear (19, 20, 42, 43). Four clinical trials have provided convincing evidence for the use of n-3 fatty acids to treat MCI, and all of them have shown a possible benefit in patients with MCI (16, 44, 45, 46). Dacks et al. (47) provided an overview of the relationship between n-3 PUFAs and cognitive aging and dementia, and argued that the available evidence does not support the idea that n-3PUFAs supplements protect against cognitive decline or dementia. A variety of studies have suggested that long-chain n-3 fatty acids (EPA and DHA) have multiple mechanisms of action in the brain and vascular system that could protect against cognitive decline and dementia (48), and n-3 fatty acids reduce cardiovascular risk factors (49). Our study found that DHA and n-6/n-3 ratio had a significant association with a higher risk of MCI in the model adjusted only for demographics. After added adjustment for the FSRP factor, the association between DHA and MCI became insignificant (P > 0.05). This may suggest that low DHA concertrations may contribute to vascular risk factors resulting in cognitive decline.

Our data also show an independent association between lower 25OHD levels and higher risk of MCI. This result is consistent with another cross-sectional study consisting of 43 MCI patients and 52 healthy people (18). We made further efforts to explore the association between 25OHD and vascular risk factors. Compared with the highest tertile in our study, the lowest tertile of 25OHD had a significant association with a higher risk of MCI in the model adjusted only for demographics. After added adjustment for the FSRP factor, the association remained significant (P < 0.05). This result shows that VitD may help to prevent neurodegenerative diseases of aging not merely through protection against vascular risk factors, and is concordant with the idea that VitD exerts its neuroprotective effects through multiple targets, including reducing vascular factors (50), decreasing the neurotoxicity (51), resisting oxidative stress (52), increasing neurotrophic factor levels (53), stimulating the innate immunity and phagocytosis of macrophages to reduce the accumulation of Aβ42 peptide (54), acting as a neurosteroid hormone by crossing the blood-brain barrier, and binding to the VitD receptor present in neurons (55), and many other targets.

In this study we did not find any significant difference in VitB12 levels between MCI and controls, but folate concentrations were lower in the MCI groups (a-MCI and md-MCI) compared with the NC group, consistent with the report of Quadri et al. (17). A single-center, randomized, double-blind controlled trial (VITACOG) of high-dose folate, VitB6, and B12 supplementation conducted in the United Kingdom (56), found that that high doses of B vitamins can be used to reduce the rate of atrophy of the brain in elderly people with MCI, but the study was not powered sufficiently to detect an effect of treatment on cognition. Two other randomized controlled trials (57, 58) found that there was no significant effect of B vitamins on cognition. Coincidently, a meta-analysis (59) conducted before 2011 showed that supplementation of VitB12, B6, and folate alone or in combination does not appear to improve cognitive function in individuals with or without existing cognitive impairments.

The present study has several limitations. First, this was a two-center study. Multicenter trials with larger participants are needed. Second, although we were able to control for many characteristics likely to modify the association between nutrient biomarkers and MCI, residual confounders might be still present, such as the serum parathyroid hormone concentration (60). Finally, the use of a cross-sectional design in our study prevents the determination of a causal relationship, and a scenario of reverse causation remains possible, with MCI leading to low nutrition concertrations because of memory loss. Further prospective analyses in a variety of memory clinics are needed to clarify whether NC subjects with low nutrient biomarker concentrations are more likely to develop MCI than their counterparts with a normal nutrition status, and whether nutrient biomarkers could be prognostic factor for the development of dementia among patients with MCI.

In conclusion, the findings we report here suggest that the abnormal concentrations of many indicators of blood nutrient biomarkers in MCI subjects indicate that they may be potential biomarkers for the early diagnosis of cognitive impairment and they are a potential target for early intervention for MCI and earlier stage. However, we found no evidence that FSRP score is an early biomarker of cognitive impairment, and more evidence is needed to be conclusive.

Acknowledgements: The study was financially supported by Grant Number 303132178 from the Natural Science Foundation of Hubei Province.

Conflict of interest: The authors declare that we have no conflict of interest.

Ethical Standards: The study procedures were approved by the Medical Ethical Committee of the Zhongnan Hospital of Wuhan University (Approval Number:2012083).