Association Between High Serum Tetrahydrofolate and Low Cognitive Functions in the United States: A Cross-Sectional Study

Abstract

Background:

The relationship between serum folate status and cognitive functions is still controversial.

Objective:

To evaluate the association between serum tetrahydrofolate and cognitive functions.

Methods:

A total of 3,132 participants (60–80 years old) from the 2011–2014 NHANES were included in this cross-sectional study. The primary outcome measure was cognitive function assessment, determined by the Consortium to Establish a Registry for Alzheimer’s Disease Word Learning Test (CERAD-WL), CERAD-Delayed Recall Test (CERAD-DR), Animal Fluency Test (AF), Digit Symbol Substitution Test (DSST), and global cognitive score. Generalized linear model (GLM), multivariate logistic regression models, weighted generalized additive models (GAM), and subgroup analyses were performed to evaluate the association between serum tetrahydrofolate and low cognitive functions.

Results:

In GLM, and the crude model, model 1, model 2 of multivariate logistic regression models, increased serum tetrahydrofolate was associated with reduced cognitive functions via AF, DSST, CERAD-WL, CERAD-DR, and global cognitive score (p < 0.05). In GAM, the inflection points were 1.1, 2.8, and 2.8 nmol/L tetrahydrofolate, determined by a two-piece wise linear regression model of AF, DSST, and global cognitive score, respectively. Also, in GAM, there were no non-linear relationship between serum tetrahydrofolate and low cognitive functions, as determined by CERAD-WL or CERAD-DR. The results of subgroup analyses found that serum tetrahydrofolate levels and reduced cognitive functions as determined by AF had significant interactions for age and body mass index. The association between high serum tetrahydrofolate level and reduced cognitive functions as determined using DSST, CERAD-WL, CERAD-DR, or global cognitive score had no interaction with the associations between cognition and gender, or age, or so on.

Conclusion:

High serum tetrahydrofolate level is associated with significantly reduced cognitive function.

Keywords

Analysis of non-linear relationship cognitive functions dementia multiple logistic regression NHANES serum tetrahydrofolate

INTRODUCTION

Dementia is a heterogeneous group of progressive neurodegenerative conditions that include Alzheimer’s disease, vascular dementia, frontotemporal dementia, and dementia with Lewy bodies, and is the 7th leading cause of death worldwide [1]. According to the World Health Organization (WHO), there are approximately 55.2 million people currently living with dementia, and approximately 1.3 trillion dollars was spent on dementia worldwide in 2019 [2]. Dementia is characterized by loss of cognitive functions and results in significant disruption of quality of life [3]. However, mild cognitive impairment and subjective decline, the early stages of dementia, may represent a window in which disease progression could be slowed or reversed through early diagnosis and intervention [4]. Therefore, early assessment, diagnosis, and prevention of cognitive impairment are key to the development of effective treatments. Current methods for evaluation of cognitive impairment include clinical evaluation, cognitive screening, neuroimaging, and measurement of biomarkers in biofluids [5 –9]. Assessment tools include but are not limited to the Mini-Mental State Examination, Consortium to Establish a Registry for Alzheimer’s Disease Word Learning Test (CERAD-WL), CERAD-Delayed Recall Test (CERAD-DR), Animal Fluency Test (AF), and Digit Symbol Substitution Test (DSST), which are used clinically to diagnose cognitive impairment and dementia [5, 10]. In addition, measurement of ¹⁸F-florbetapir uptake by brain tissue using positron emission tomography is accepted by the European Medicine Agencies for determination of amyloid load in individuals with dementia [7]. However, the time-consuming nature of questionnaires and the high financial cost of neuroimaging have hindered the use of these techniques for large scale screening for cognitive impairment. Thus, analysis of biomarkers in biofluids is a key screening approach for cognitive impairment because it is time- and cost-efficient.

Micronutrients, particularly vitamins, are important for cognitive functions and strongly correlate with age-related cognitive decline and dementia [11 –13]. Therefore, it is important to determine whether serum concentrations of micronutrients correlate with severity of cognitive dysfunction. Folate is a water-soluble vitamin that plays a key role in neurological and psychological functions. Five forms of folate are present in serum: tetrahydrofolate (THF), 5-formyltetrahydrofolate, 5-methyltetrahydrofolate (5MeTHF), unmetabolized serum folic acid (UMFA), and 5,10-methenyltetrahydrofolate. Among these, THF is the main active form of folate in the human body [14]. THF and its one-carbon adducts are required for de novo synthesis of purines, thymidylate, glycine, methionine, and serine [15]. However, the relationship between serum folate status and cognitive function is unclear. Several studies have found that low folate levels were associated with cognitive decline [16 –18]. Folate deficiency may affect normal brain function through reduced synthesis of S-adenosylmethionine (SAM), resulting in reduced SAM-dependent methylation reactions, including synthesis and catabolism of dopamine, norepinephrine, adrenaline, and serotonin [16]. However, another study indicated that folate deficiency was not associated with cognitive impairment in elderly adults [19]. Furthermore, another study suggested that high folate levels were significantly associated with poorer CERAD-WL, CERAD-DR, AF, and DSST performance [20]. Also, high folate intake alone was associated with increased risk of cognitive impairment [21]. In addition, folate, a synthetic oxidized form can be found in fresh natural foods yet must be converted to THF by dihydrofolate reductase to be effective [14]. Hence, we hypothesized serum tetrahydrofolate may be associated with cognitive functions for the first time. This study aimed to clarify the relationship between serum THF status and cognitive functions for the first time via generalized linear model (GLM), multivariate logistic regression models, weighted generalized additive models (GAM) and subgroupanalyses.

MATERIALS AND METHODS

Data source

The National Health and Nutrition Examination Survey (NHANES) is a nationwide, ongoing, cross-sectional survey conducted by the National Center for Health Statistics (NCHS). It was designed to evaluate the health and nutritional status in the United States, including household interviews and physical examination. The interview consists of demographic, socioeconomic, dietary, and health-related questions. The examination includes medical, dental, physiological measurements, and laboratory analysis assessed by highly trained medical personnel. Participants in the survey were interviewed for baseline information by experienced participants at the homes of the patients, and each patient underwent medical and physical examinations in a unique mobile examination center [22]. Further details on the data collection process and analytical guidelines are publicly available on the NHANES website.

Study population

The 2011-2012 and 2013-2014 data from the NHANES assessed the cognitive functions of the survey participants. These two 2-year-cycle datasets were combined for subsequent analysis. In this study, we excluded participants with missing folate data (N = 3,922). We initially included participants over 60 years of age that had provided complete cognitive functions assessment data (N = 3,132). In total, 3,132 participants were eligible for inclusion in this study (Fig. 1).

Fig. 1

Flowcharts illustrating sample selection from NHANES 2011-2014. NHANES, the National Health and Nutrition Examination Survey; BMI, body mass index.

Specimen collection and serum THF measurement

Blood specimens were collected at home examination centers or mobile examination facilities by phlebotomists and medical technologists. Serum THF assays were performed on fresh or frozen serum. Ascorbic acid (0.5%) was sometimes added to serum prior to storage to improve stability. After collection, the specimens were frozen and shipped on dry ice by overnight mail. Once received, the samples were stored at ≤–20° until analysis. For long-term storage, specimens were frozen at ≤–70°.

Serum THF was determined using stable isotope dilution high performance liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) [23]. The method was based on a previously published method and was modified to reduce sample weights and increase throughput [24 –26]. Samples (150μl of serum) were combined with ammonium formate buffer and an internal standard mixture. Sample extraction and clean-up was performed by automated 96-probe solid phase extraction (SPE) using 96-well phenyl SPE plates. Sample preparation required approximately 1 h per 96-well plate. Folate species were separated within 6 min using isocratic mobile phase conditions and quantified using MS/MS. Quantitation was performed using a five-point linear calibration curve of peak area ratios using 1/x² weighting. The limit of detection for THF is 0.141 to 39.3 nmol/L.

Cognitive function assessment

Our primary outcome was cognitive function as determined using several cognitive function tests including CERAD-WL, CERAD-DR, AF, and DSST. The cognitive function tests were administered in person in the mobile examination center in the requested language (limited to English, Spanish, Chinese, Korean, or Vietnamese) of each study participant. The CERAD score represented immediate and delayed learning and recall ability for new verbal information [27]. After exposing the participants to ten words audibly or visually, the experienced researcher instructed the participants to recall and read aloud the words immediately and after completing the AF and DSST tests. In addition, AF was administered to assess categorical verbal fluency [28]. Participants were asked to name as many animals as possible within 1 min. One point was awarded for each named animal. Although, there is no consensus on what aspects of cognitive function the DSST measures, literature agrees that it measures working memory and sustained attention [28]. The researcher provided participants with a piece of paper with a key at the top. One hundred thirty-three boxes adjoined numbers at the bottom of the paper. The key presented pairings of 9 numbers and symbols. The participants were asked to match the corresponding symbols for the 133 boxes in 2 min.

Although there was no uniform definition of low cognitive function, previous studies based on the NHANES database defined low cognitive function as the lowest 25th percentile on the cognitive tests [29 –31]. Therefore, our study defined <25% as the threshold of low cognitive function. Participants with average CERAD-WL scores (top score: 10) <5, average CERAD-DR score (top score: 10) <5, AF score (top score: 40) <13, DSST score (top score: 105) <34, or global cognitive score (sum of CERAD-WL, CERAD-DR, AF, DSST scores; top score: 165) <10 were placed in the low cognitive function (LC) group. All other participants were placed in the normal cognitive function (NC) group.

Potential covariates

Baseline sociodemographic data were collected, including demographic and questionnaire data in NHANES. These data included gender (male, female), age (continuous, NHANES coded those with 80 years and older as simply 80 years old), body mass index (BMI; continuous), race (Hispanic, Non-Hispanic), education level (less than high school, high school, more than high school), marital status (married, widowed, divorced, separated, never married, living with partner), annual income (<75000$, ≥75000$), smoking history, and sleep disorder.

We also collected laboratory data, including information regarding hypertension, hyperlipoidemia, diabetes, and Parkinson’s disease (PD). These covariates were identified as potential covariates based on previous studies [31 –35]. Vascular risk factors such as hypertension, hyperlipidemia, and diabetes are potentially modifiable risk factors associated with 40% of dementia cases worldwide [36]. In addition, 15% –40% of patients with PD had cognitive impairment at the time of diagnosis [37]. To explore whether PD was a potential covariate that affected the relationship between THF and cognitive function, we identified patients with PD in our participant pool [34, 35]. In our study, PD cases were defined by use of the following PD-specific medications: benztropine, carbidopa, levodopa, ropinirole, methyldopa, entacapone, and amantadine [34, 35].

Statistical analysis

The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) cross-sectional checklist was administered in this study [38]. Continuous variables were reported as the mean±standard deviation and categorical variables were reported as frequencies or as percentages. Moreover, a complex and multi-level probability sampling design was performed in NHANES [39]. Therefore, the proposed weighting methodology was performed in our study. Weighted Student’s t tests (continuous variables including age, BMI, and serum THF) and weighted chi-square tests (categorical variables, including gender, race, educational level, marital status, annual income, smoking, hypertension, hyperlipoidemia, diabetes, and PD) were performed to determine any statistical difference between the means and proportions of the two groups.

We tested our main hypothesis that high serum THF is associated with low cognitive function within the framework of a GLM. In the GLM, AF, DSST, CERAD-WL, CERAD-DR, and global cognitive scores are used as continuous variables to assess the simple effects between serum THF and cognitive function. Due to non-normality of the data, a gamma-distribution and logarithmic link function were applied on the change from AF, DSST, CERAD-WL, CERAD-DR, and global cognitive scores.

Univariate and multivariate logistic regression models were weighted and generated to evaluate the association between serum THF and cognitive function (AF, DSST, CERAD-WL, CERAD-DR, and global cognitive scores are used as dichotomous variable). According to the STROBE statement [38], we simultaneously showed the results of crude (no adjustment for covariates), minimally (only adjusted for age and gender), and fully (adjusted for all covariates) adjusted analyses.

To evaluate the potential non-linear relationship between exposure and outcome, a GAM was generated in our study. If a non-linear relationship was detected, a two-piece wise linear regression model was generated to assess the threshold effect of serum THF and cognitive function based on the smoothing plot. The threshold level of serum THF at which the association between low cognitive functions began to change and became statistically different was evaluated using a recurrence method. The inflection point was moved along a predefined interval and the inflection point that showed the maximum model likelihood was investigated [40].

Subgroup analysis was conducted using stratified multivariate logistic regression model. The subgroup effect modification test used the interaction terms of serum THF between the subgroup indicators, followed by the likelihood ratio test. All analysis were conducted using the statistical software packages R ( http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empowerstats.com, X&Y Solutions, Boston, MA, USA). p values less than 0.05 (two-sides) were considered significantly different.

Consent to participate

All participants in this study signed written informed consent and all research procedures were approved by the National Center for Health Statistics Research Ethics Review Board [41]. These data are public, so the approval of the institutional review board was not required for this study.

RESULTS

Baseline characteristics of participants

Baseline characteristics are summarized in Table 1, the baseline characteristics were summarized and compared. All participants in the LC groups had significantly higher serum THF levels than those in the NC group, regardless of the cognitive assessment used to place the participants in the LC group (p < 0.05) (Table 1). For AF, 27.1% of Hispanics had LC, whereas 25.5% of Non-Hispanics had LC; for DSST, 43% of Hispanics had LC, whereas 22.6% of Non-Hispanics had LC; For CERAD-WL, 29.2% of Hispanics had LC, whereas 23.6% of Non-Hispanics had LC; for CERAD-DR, 31.8% of Hispanics had LC, whereas 26.5% of Non-Hispanics had LC; for global cognitive score, 42.2% of Hispanics had LC, whereas 22.0% of Non-Hispanics had LC (Table 1). In addition, LC was associated with older age, lower BMI, lower annual income, and diabetes (p < 0.05) (Table 1).

Table 1

Baseline characteristics of selected participants from NHANES 2011–2014

	AF score		DSST score			CERAD-WL score			CERAD-DR score				Global cognitive score
Variables	NC (N = 2,308)	LC (N = 803)	p	NC (N = 2,216)	LC (N = 799)	p	NC (N = 2,345)	LC (N = 787)	p	NC (N = 2,267)	LC (N = 860)	p	NC (N = 2,190)	LC (N = 760)	p
Age, y	69.1±6.7	71.5±6.9	<0.001	68.8±6.6	71.5±6.8	<0.001	68.8±6.6	72.4±7.0	<0.001	68.6±6.5	72.6±6.8	<0.001	68.7±6.6	71.6±6.9	<0.001
Gender			0.236			<0.001			<0.001			<0.001			0.025
Male	1,131 (49.0%)	374 (46.6%)		1,034 (46.7%)	435 (54.4%)		1,060 (45.2%)	459 (58.3%)		1,016 (44.8%)	500 (58.1%)		1,038 (47.4%)	396 (52.1%)
Female	1,177 (51.0%)	429 (53.4%)		1,182 (53.3%)	364 (45.6%)		1,285 (54.8%)	328 (41.7%)		1,251 (55.2%)	360 (41.9%)		1,152 (52.6%)	364 (47.9%)
BMI, kg/m²	29.1±6.4	28.8±6.3	0.261	29.1±6.4	28.8±6.2	0.208	29.3±6.4	28.4±6.1	<0.001	29.3±6.5	28.5±6.1	0.001	29.1±6.4	28.9±6.3	0.403
Race, n (%)			<0.001			<0.001			<0.001			<0.001			<0.001
Hispanic	439 (19.0%)	163 (20.3%)		325 (14.7%)	245 (30.7%)		417 (17.8%)	172 (24.4%)		413 (18.2 %)	193 (22.4%)		321 (14.7%)	234 (30.8%)
Non-Hispanic	1,869 (81.0%)	640 (79.7%)		1,891 (85.3%)	554 (69.3%)		1,928 (82.8%)	595 (75.6%)		1,854 (81.8%)	667 (77.6%)		1,869 (85.3%)	526 (69.2%)
Education level, n (%)			<0.001			<0.001			<0.001			<0.001			<0.001
Less than high school	491 (21.3%)	350 (43.8%)		322 (14.5%)	452 (56.7%)		494 (21.1%)	359 (45.7%)		501 (22.1%)	348 (40.6%)		323 (14.8%)	425 (56.1%)
High school	513 (22.2%)	207 (25.9%)		527 (23.8%)	176 (22.1%)		549 (23.4%)	174 (22.2%)		518 (22.9%)	203 (23.7%)		523 (23.9%)	167 (22.0%)
More than high school	1303 (56.5%)	243 (30.4%)		1366 (61.7%)	169 (21.2%)		1300 (55.5%)	252 (32.1%)		1246 (55.0%)	307 (35.8%)		1343 (61.4%)	1343 (61.4%)
Marital status, n (%)			<0.001			<0.001			<0.001			<0.001			<0.001
Married	1,305 (56.6%)	381 (47.6%)		1,287 (58.2%)	367 (45.9%)		1,307 (55.8%)	390 (49.6%)		1,263 (55.8%)	431 (50.2%)		1,277 (58.4%)	347 (45.7%)
Widowed	402 (17.4%)	237 (29.6%)		370 (16.7%)	226 (28.3%)		430 (18.4%)	216 (27.5%)		411 (18.2%)	234 (27.2%)		360 (16.5%)	219 (28.8%)
Divorced	339 (14.7%)	93 (11.6%)		330 (14.9%)	97 (12.1%)		352 (15.0%)	80 (10.2%)		342 (15.1%)	91 (10.6%)		328 (15.0%)	90 (11.8%)
Separated	55 (2.4%)	30 (3.7%)		38 (1.7%)	44 (5.5%)		57 (2.4%)	31 (3.9%)		56 (2.5%)	32 (3.7%)		38 (1.7%)	40 (5.3%)
Never married	137 (5.9%)	46 (5.7%)		127 (5.7%)	46 (5.8%)		137 (5.9%)	46 (5.9%)		134 (5.9%)	48 (5.6%)		120 (5.5%)	48 (6.3%)
Living with partner	67 (2.9%)	14 (1.7%)		60 (2.7%)	19 (2.4%)		58 (2.5%)	23 (2.9%)		57 (2.5%)	23 (2.7%)		63 (2.9%)	16 (2.1%)
Annual income			<0.001			<0.001			<0.001			<0.001			<0.001
<75000$	1,649 (74.3%)	638 (85.4%)		1,527 (71.9%)	680 (90.2%)		1,684 (75.0%)	620 (83.3%)		1,614 (74.6%)	684 (83.5%)		1,514 (72.0%)	640 (89.9%)
≥75000$	571 (25.7%)	109 (14.6%)		597 (28.1%)	74 (9.8%)		560 (25.0%)	124 (16.7%)		550 (25.4%)	135 (16.5%)		589 (28.0%)	72 (10.1%)
Smoking, n (%)			0.316			<0.001			0.558			0.955			0.017
Yes	287 (24.7%)	109 (27.2%)		250 (22.6%)	130 (31.2%)		305 (25.6%)	94 (24.1%)		289 (25.2%)	108 (25.1%)		256 (23.3%)	116 (29.4%)
No	877 (75.3%)	292 (72.8%)		858 (77.4%)	286 (68.8%)		887 (74.4%)	296 (75.9%)		858 (74.8%)	323 (74.9%)		842 (76.7%)	279 (70.6%)
Sleep disorder, n (%)			0.058			<0.001			0.023			<0.001			<0.001
Yes	103(7.3%)	57(12.2%)		77(7.4%)	48(8.02%)		94(9.7%)	49(8.1%)		79(7.5%)	60(11.5%)		82(8.1%)	44(9.1%)
No	1,302(92.7%)	410(87.8%)		963(92.6%)	550(92.0%)		878(90.3%)	556(91.9)		979(92.5%)	460(88.5%)		928(91.9%)	439(90.9%)
Hypertension, n (%)			<0.001			<0.001			0.031			0.055			<0.001
Yes	1,396 (60.5%)	558 (69.5%)		1,326 (59.8%)	560 (70.1%)		1,449 (61.8%)	520 (66.1%)		1,402 (61.8%)	565 (65.7%)		1,314 (60.1%)	526 (69.5%)
No	908 (39.3%)	243 (30.3%)		886 (40.0%)	237 (29.7%)		893 (38.1%)	264 (33.5%)		862 (38.0%)	292 (34.0%)		873 (39.9%)	231 (30.5%)
Hyperlipoidemia, n (%)			0.123			<0.001			0.037			0.003			0.066
Yes	1,306 (56.6%)	421 (52.4%)		1,247 (56.3%)	424 (53.1%)		1,330 (56.7%)	413 (52.5%)		1,286 (56.7%)	452 (52.6%)		1,244 (57.0%)	396 (53.1%)
No	985 (42.7%)	376 (46.8%)		963 (43.5%)	360 (45.1%)		1002 (42.7%)	365 (46.4%)		971 (42.8%)	396 (46.0%)		940 (43.0%)	350 (46.9%)
Diabetes			<0.001			<0.001			<0.001			0.037			<0.001
Yes	497 (21.5%)	247 (30.8%)		461 (20.8%)	247 (30.9%)		522 (22.3%)	231 (29.4%)		520 (22.9%)	232 (27.0%)		449 (20.5%)	241 (31.8%)
No	1,699 (73.6%)	521 (64.9%)		1649 (74.4%)	518 (64.8%)		1715 (73.1%)	519 (65.9%)		1649 (72.7%)	583 (67.8%)		1636 (74.7%)	485 (63.9%)
Borderline	111 (4.8%)	34 (4.2%)		105 (4.7%)	33 (4.1%)		106 (4.5%)	37 (4.7%)		96 (4.2%)	45 (5.2%)		104 (4.8%)	33 (4.3%)
PD			0.598			0.691			0.006			0.123			0.458
Yes	21 (0.9%)	9 (1.1%)		20 (0.9%)	6 (0.8%)		16 (0.7%)	14 (1.8%)		18 (0.8%)	12 (1.4%)		19 (0.9%)	10 (1.3%)
No	2,287 (99.1%)	794 (98.9%)		2,196 (99.1%)	793 (99.2%)		2,329 (99.3%)	773 (98.2%)		2,249 (99.2%)	848 (98.6%)		2,171 (99.1%)	750 (98.7%)
Serum THF, nmol/L	1.1±0.9	1.3±1.1	<0.001	1.1±0.9	1.2±1.1	0.024	1.1±0.9	1.3±1.0	0.003	1.1±0.9	1.3±1.0	<0.001	1.1±0.9	1.3±1.1	0.002

NC, normal cognitive function group; LC, low cognitive function group; BMI, body mass index; THF, tetrahydrofolate; AF, Animal Fluency test; DSST, the Digit Symbol Substitution Test; CERAD-WL, the consortium to Establish a Registry for Alzheimer’s Disease World Learning Test; CERAD-DR, the CERAD-Delayed Recall Test.

Generalized linear model analyses and multiple logistic regression analyses of serum THF and cognitive functions

To assess the simple effects between serum THF and cognitive functions, AF score, DSST score, CERAD-WL score, CERAD-DR score, and global cognitive score are used as continuous variables in the GLM. The GLM exhibited serum THF was negatively correlated with AF score (β: –0.44, 95% CI: –0.65––0.22, p = 0.000061), DSST score (β: –1.42, 95% CI: –2.1––0.74, p = 0.000044), CERAD-WL score (β: –0.13, 95% CI: –0.19––0.06, p = 0.000072), CERAD-DR score (β: –0.19, 95% CI: –0.28––0.09, p = 0.00008), and global cognitive score (β: –0.21, 95% CI: –3.01––0.35, p = 0.002647) (Table 2). However, it is more clinically meaningful to divide each cognitive function test score into a dichotomous variable. Then, we used multiple logistic regression, a GLM for analyzing dichotomous variable, to assess the relationship between serum THF and cognitive functions. Three logistic regression models were generated to characterize the association between serum THF and cognitive functions (Table 3). In the crude model (not adjusted for covariates), high serum THF correlated with low cognitive functions, as determined by AF (OR: 1.16, 95% CI: 1.07–1.26, and p < 0.001), DSST (OR: 1.11, 95% CI: 1.01–1.1, and p = 0.024), CERAD-WL (OR: 1.14, 95% CI: 1.05–1.24, and p = 0.003), CERAD-DR (OR: 1.18, 95% CI: 1.07–1.27, and p < 0.001), or global cognitive score (OR: 1.15, 95% CI: 1.05–1.25, and p = 0.002). In the minimally adjusted model (adjusted for age and gender only), the results did not differ from those obtained in the crude analysis, and it indicated that high serum THF was associated with low cognitive functions. The results for placement in the LC group according to each test were as follows: AF (OR: 1.16, 95% CI: 1.07–1.27, and p < 0.001), DSST (OR: 1.11, 95% CI: 1.02–1.21, and p = 0.02), CERAD-WL (OR: 1.14, 95% CI: 1.04–1.24, and p = 0.004), CERAD-DR (OR: 1.17, 95% CI: 1.08–1.28, and p < 0.001), and global cognitive score (OR: 1.15, 95% CI: 1.05–1.26, and p = 0.002). In the fully adjusted model (adjusted all covariates listed in Table 1), the OR trend did not differ for AF (OR: 1.14, 95% CI: 1.04–1.24, and p = 0.004), DSST (OR: 1.08, 95% CI: 1.01–1.18, and p = 0.042), CERAD-WL (OR: 1.10, 95% CI: 1.00–1.20, and p = 0.044), CERAD-DR (OR: 1.13, 95% CI: 1.03–1.23, and p = 0.008), or global cognitive score (OR: 1.12, 95% CI: 1.02–1.23, and p = 0.015) compared with the crude and minimally adjusted models. These results indicated that serum THF was an independent risk factor for the low cognitive functions, and the results were stable androbust.

Table 2

Generalized linear model analyses the association between Serum tetrahydrofolate and cognitive functions assessed by the AF, DSST, CERAD-WL or CERAD-DR

Outcome	β (95% CI)	Standard Error	p
AF	–0.44 (–0.65, –0.22)	0.11	0.000061
DSST	–1.42 (–2.10, –0.74)	0.35	0.000044
CERAD-WL	–0.13 (–0.19, –0.06)	0.03	0.000072
CERAD-DR	–0.19 (–0.28, –0.09)	0.05	0.00008
Global cognitive test	–0.21 (–3.01, –0.35)	0.07	0.002647

CI, confidence interval, the sample size of AF, DSST, CERAD-WL, CERAD-DR, global cognitive score is 3111, 3015, 3132, 3127, and 2950, respectively.

Table 3

Multiple logistic regression analyses the association between Serum tetrahydrofolate and cognitive functions assessed by the AF, DSST, CERAD-WL, or CERAD-DR.

Outcome	Crude model OR (95% CI) p	Model 1 OR (95% CI) p	Model 2 OR (95% CI) p
AF	1.16 (1.07, 1.26) <0.001	1.16 (1.07, 1.27) <0.001	1.14 (1.04, 1.24) 0.004
DSST	1.11(1.01, 1.1) 0.024	1.11 (1.02, 1.21) 0.020	1.08 (1.01, 1.18) 0.042
CERAD-WL	1.14 (1.05, 1.24) 0.003	1.14 (1.04, 1.24) 0.004	1.10 (1.0, 1.2) 0.044
CERAD-DR	1.18 (1.07, 1.27) <0.001	1.17 (1.08, 1.28) <0.001	1.13 (1.03, 1.23) 0.008
Global cognitive score	1.15 (1.05, 1.25) 0.002	1.15 (1.05, 1.26) 0.002	1.12 (1.02, 1.23) 0.015

Crude model adjusted for: none, the sample size of AF, DSST, CERAD-WL, CERAD-DR, and global cognitive score is 3111, 3015, 3132, 3127, and 2950, respectively; Model 1 adjusted for: age and gender, the sample size of AF, DSST, CERAD-WL, CERAD-DR, and global cognitive score is 2806, 2719, 2822, 2822, and 2817, respectively; Model 2 adjusted for: all covariates listed in Table 1 were adjusted, the sample size of AF, DSST, CERAD-WL, CERAD-DR, and global cognitive score is 2762, 2683, 2778, 2778, and 2774, respectively. OR, odds ratio; CI, confidence interval.

Analysis of non-linear relationships between serum THF and cognitive functions

As serum THF level is a continuous variable, we considered that there could be a non-linear relationship between serum THF and incidence of low cognitive functions. The associations between serum THF and low cognitive function were non-linear (adjusted for all covariates in Table 1) for AF (Fig. 2A), DSST (Fig. 2B), and global cognitive score (Fig. 2E).

Fig. 2

Non-linear relationship between serum THF and low cognitive functions. A) Relationship between serum THF and low AF score; B) Relationship between serum THF and low DSST score; C) Relationship between serum THF and low CERAD-WL score; D) Relationship between serum THF and low CERAD-DR score; E) Relationship between serum THF and low global cognitive score. All results were detected after adjusting for all covariates listed in Table 1. X-axis, the concentration of serum THF (nmol/L); Y-axis, the probability of low AF score; OR, red empty circles; 95% CI, blue circle.

For determination of low cognitive function using AF, the inflection point was 1.1 nmol/L, as determined using a two-piece wise linear regression model (Table 4). On the right side of the inflection point, the effect size, 95% CI, and p value were 1.25, 1.09 to 1.40, and p < 0.001, respectively. However, we detected no obvious relationship between serum THF and low cognitive function on the left side of the inflection point (OR = 0.80, 95% CI: 0.59 to 1.08, and p = 0.143).

Table 4

Non-linearity addressing by weighted two-piecewise linear model of AF score, DSST score, CERAD-WL score, CERAD-DR score, and global cognitive score

Statistic model	AF score			DSST score			CERAD-WL score			CERAD-DR score			Global cognitive score
	Serum THF (nmol/L)	OR (95% CI)	p	Serum THF (nmol/L)	OR (95% CI)	p	Serum THF (nmol/L)	OR (95% CI)	p	Serum THF (nmol/L)	OR (95% CI)	p	Serum THF (nmol/L)	OR (95% CI)	p
Fitting by standard linear model		1.14(1.04, 1.24)	0.004		1.08(0.98, 1.10)	0.04		1.10(1.00, 1.20)	0.048^*		1.13(1.03, 1.23)	0.008		1.12(1.02, 1.23)	0.015
Fitting by two-piecewise linear model	≥1.1	0.80(0.59, 1.08)	0.143	≥2.8	0.88(0.76, 1.02)	0.092	≥0.3	14.19(0.61, 328.67)	0.0981	≥0.7	0.91(0.53, 1.56)	0.7337	≥2.8	0.90(0.77, 1.04)	0.146
	≥1.1	1.25(1.09, 1.40)	<0.001	≥2.8	1.86(1.35, 2.56)	<0.001	≥0.3	1.07(0.97, 1.18)	0.2019	≥0.7	1.15(1.04, 1.28)	0.007	≥2.8	2.08(1.50, 2.89)	<0.001
	Log likelihood ratio	0.016		Log likelihood ratio	<0.001		Log likelihood ratio	0.104		Log likelihood ratio	0.429		Log likelihood ratio	<0.001

OR has been adjusted for all covariates listed in Table 1, the sample size of AF, DSST, CERAD-WL, CERAD-DR, and global cognitive score is 2762, 2683, 2778, 2774, and 2629, respectively.

For determination of low cognitive function based on DSST score, the inflection point was 2.8 nmol/L, as determined using a two-piece wise linear regression model (Table 4). On the right side of the inflection point, serum THF was significantly associated with low cognitive function (OR = 1.86, 95% CI 1.35 to 2.56, p < 0.001). On the left side of the inflection point, there was no association between serum THF and low cognitive function (OR = 0.88, 95% CI 0.76 to 1.02, p = 0.092). No non-linear relationships were observed between serum THF and low cognitive functions as determined by CERAD-WL or CERAD-DR (Fig. 2C, D; Table 4).

We also found a non-linear relationship between serum THF and global cognitive score. The inflection point was 2.8 nmol/L, as determined by a two-piece wise linear regression model. On the right side of the inflection point, the effect size, 95% CI, and p value were 2.08, 1.50 to 2.89, and <0.001, respectively. No obvious association between serum THF and global cognitive score was detected on the left side of the inflection point (OR = 0.90, 95% CI: 0.77 to 1.04, and p = 0.146) (Table 4).

Effect size of serum THF on low cognitive functions in exploratory subgroups

To further characterize the effect of serum THF on low cognitive functions, we conducted a series of subgroup analysis. In participants administered AF, no interactions were observed for gender, race, education level, annual income, medical history of hypertension, hyperlipidemia, diabetes, or PD (p for interaction = 0.892, 0.12, 0.647, 0.223, 0.238, 0.152, 0.538, and 0.395, respectively) (Table 5). Significant interactions were observed for age (p for interaction = 0.011) and BMI (p for interaction = 0.001). In participants younger than 70 years old, higher serum THF was associated with low cognitive function (OR = 1.28 95% CI 1.13 to 1.44, p < 0.01). In contrast, serum THF did not correlate with cognitive function in participants over 70 years old. Higher serum THF was also associated with low cognitive function (OR = 1.43 95% CI 1.24 to 1.66, p < 0.01) in participants with BMI greater than 30, but not in participants with BMI less than 30.

Table 5

Effect size of Serum THF on low AF score, low DSST score, low CERAD-WL score, low CERAD-DR score and low Global cognitive score in exploratory subgroups

Characteristic	Low AF score			Low DSST score			low CERAD-WL score			low CERAD-DR score			low Global cognitive score
	No of participants	OR (95% CI)	p for interaction	No of participants	OR (95% CI)	p for interaction	No of participants	OR (95% CI)	p for interaction	No of participants	OR (95% CI)	p for interaction	No of participants	OR (95% CI)	p for interaction
Gender
Male	1,375	1.17(1.04,1.32)	0.892	1,341	1.16(1.03,1.30)	0.268	1,388	1.22(1.09,1.36)	0.107	1,385	1.18(1.06,1.32)	0.83	1,310	1.16(1.03,1.31)	0.843
Female	1,431	1.15(1.02,1.30)		1,378	1.05(0.92,1.20)		1,434	1.05(0.92,1.21)		1,432	1.16(1.02,1.32)		1,355	1.14(1.00,1.30)
Age
<70	1,492	1.28(1.13,1.44)	0.011	1,479	1.16(1.03,1.32)	0.118	1,498	1.19(1.05,1.36)	0.148	1,497	1.13(0.99,1.28)	0.872	1,450	1.22(1.08,1.39)	0.07
≥70	1,314	1.02(0.90,1.15)		1,240	1.01(0.89,1.15)		1,324	1.05(0.93,1.18)		1,320	1.14(1.02,1.29)		1,215	1.04(0.91,1.18)
BMI
<25	733	0.97(0.82,1.14)	0.001	712	1.00(0.85,1.19)	0.265	740	1.11(0.95,1.29)	0.894	739	1.11(0.95,1.30)	0.721	701	1.00(0.85,1.19)	0.123
25–30	993	1.11(0.96,1.28)		971	1.20(1.04,1.39)		994	1.13(0.98,1.30)		991	1.21(1.06,1.39)		946	1.18(1.01,1.37)
>30	1,036	1.43(1.24,1.66)		1,000	1.1(0.9, 1.3)		1,044	1.17(1.0,1.36)		1,044	1.16(1.00,1.35)		982	1.26(1.08,1.47)
Race
Hispanic	535	1.31(1.10,1.56)	0.12	509	1.09(0.92,1.29)	0.692	540	1.28(1.08,1.52)	0.124	537	1.23(1.04,1.46)	0.481	495	1.17(0.99,1.39)	0.918
Non-Hispanic	2,271	1.12(1.01,1.23)		2,210	1.13(1.02,1.25)		2,282	1.10(0.99,1.21)		2,280	1.15(1.04,1.26)		2,170	1.16(1.04,1.29)
Educational level
Less than high school	720	1.10(0.95,1.27)	0.647	660	0.99(0.86,1.15)	0.114	728	1.13(0.98,1.30)	0.345	724	1.20(1.04,1.38)	0.639	639	1.02(0.88,1.19)	0.059
High school	661	1.22(1.02,1.46)		646	1.27(1.05,1.53)		663	1.26(1.04,1.52)		661	1.19(0.99,1.42)		636	1.37(1.13,1.65)
More than high school	1,422	1.17(1.02,1.35)		1,411	1.01(0.85,1.21)		1,428	1.05(0.91,1.22)		1,429	1.10(0.96,1.26)		1,388	1.09(0.91,1.30)
Annual income
<75000$	2,039	1.12(1.02,1.24)	0.223	1,966	1.07(0.97,1.18)	0.125	2,053	1.12(1.02,1.24)	0.471	2,047	1.16(1.05,1.28)	0.937	1,922	1.11(1.00,1.22))	0.077
≥75000$	641	1.29(1.06,1.58)		632	1.31(1.04,1.65)		643	1.22(1.00,1.48)		644	1.15(0.95,1.40)		624	1.39(1.11,1.74)
Sleep disorder
Yes	230	1.22(1.08,1.38)	0.11	216	1.13(1.00,1.28)	0.246	231	1.09(1.00,1.30)	0.067	211	1.15(1.01,1.30)	0.659	228	1.20(1.06,1.36)	0.159
No	2,529	1.06(0.93,1.21)		2,503	1.02(0.89,1.17)		2,591	1.13(1.00,1.29)		2,538	1.13(0.99,1.29)		2,534	1.05(0.91,1.21)
Hypertension
Yes	1,765	1.20(1.08,1.33)	0.238	1,703	1.09(0.98,1.21)	0.849	1,775	1.09(0.98,1.21)	0.166	1,773	1.16(1.04,1.28)	0.993	1,666	1.14(1.02,1.27)	0.876
No	1,036	1.07(0.92,1.26)		1,011	1.11(0.95,1.30)		1,042	1.24(1.07,1.43)		1,039	1.16(1.00,1.34)		994	1.16(0.99,1.35)
Hyperlipoidemia
Yes	1,565	1.23(1.10,1.38)	0.152	1,514	1.15(1.02,1.29)	0.329	1,578	1.12(0.99,1.26)	0.55	1,573	1.16(1.04,1.30)	0.868	1,487	1.21(1.07,1.37)	0.235
No	1,221	1.09(0.96,1.24)		1,187	1.05(0.92,1.20)		1,224	1.18(1.04,1.34)		1,224	1.18(1.04,1.33)		1,161	1.09(0.95,1.24)
Diabetes
Yes	654	1.26(1.06,1.49)	0.538	652	1.27(1.07,1.51)	0.127	661	1.14(0.96,1.35)	0.427	660	1.21(1.02,1.43)	0.574	609	1.30(1.09,1.55)	0.26
No	2,017	1.13(1.01,1.25)		1,966	1.03(0.93,1.15)		2,028	1.11(1.00,1.24)		2,026	1.13(1.03,1.25)		1,928	1.09(0.98,1.22)
Borderline	133	1.10(0.73,1.65)		126	1.20(0.78,1.85)		131	1.45(0.99,2.14)		129	1.36(0.93,1.98)		126	1.13(0.73,1.74)
PD
Yes	28	0.82(0.34,1.98)	0.395	24	0.55(0.15,2.06)	0.194	28	1.16(0.59,2.29)	0.957	28	1.05(0.53,2.11)	0.774	24	0.55(0.15,2.06)	0.166
No	2,778	1.17(1.07,1.24)		2,695	1.11(1.02,1.21)		2794	1.14(1.04,1.24)		2,789	1.17(1.07,1.27)		2,641	1.16(1.06,1.26)

Above model adjusted for all covariates listed in Table 1 were adjusted. In each case, the model is not adjusted for the stratification variable.

In DSST, CERAD-WL, CERAD-DR, or global cognitive score, there is no interaction between cognition and gender in the association between serum THF and low cognitive functions. Also, in participants administered DSST, CERAD-WL, CERAD-DR, or global cognitive score, no interactions were observed for age, BMI, race, education level, annual income, hypertension, hyperlipidemia, diabetes, or PD (Table 5).

DISCUSSION

Since the relationship between serum folate status and cognitive functions is still controversial and THF is the main active form of folate in human body. We further analyzed the relationship between THF and cognitive functions in an attempt to provide a new basis for the prevention and treatment of cognitive impairment. Interestingly, our results, which are similar to those of Bailey et al. and Morris et al., indicated that higher levels of serum THF increased the chance of low cognitive functions, including learning ability and memory [20, 21]. Moreover, this relationship was not affected by gender, age, BMI, race, education level, annual income, hypertension, or diabetes. These results support our hypothesis, which suggest that serum THF levels are negative predictors of learning ability and memory.

In addition, serum THF status and other cognitive functions, including categorical verbal fluency, processing speed, sustained attention, and working memory, had non-linear relationships. Serum THF negatively correlated with executive function/processing speed at serum THF levels greater than or equal to 2.8 nmol/L. These results explain why the previous relationship between serum folate status and cognitive functions has been controversial. Because serum folate status and serum THF status are continuous variables, there may be a non-linear relationship. GAM has obvious advantages in dealing with non-linear relations, can handle non-parametric smoothing and will fit a regression spline to the data [42]. Therefore, a variety of statistical methods are used to explore the relationship between risk factors and cognitive functions from different angles, and the results are more objective.

Interestingly, the relationship between serum THF status and categorical verbal fluency is more complicated. The results indicated that high serum level of THF was associated with worse categorical verbal fluency when serum THF was greater than or equal to 1.1 nmol/L, in participants younger than 70 years old, or in participants with BMI greater than 30. Aging can affect nervous cells to change the brain structurally and functionally via loss of synaptic function and plasticity, decreased expression of neurotrophic factors, dysregulation of histone acetylation, global downregulation of DNA methylation, and breakdown of myelin [43]. Therefore, more factors affect cognitive functions as age increases, which may explain the interaction between age and the relationship between THF and categorical verbal fluency. In addition, BMI is one of the main factors that affects cognitive functions and folate metabolism [44, 45]. This finding indicates that BMI less than or equal 30 is likely to be associated with the relationship between THF and categorical verbal fluency. However, our results only showed a relationship between THF and cognitive functions. The underlying mechanism by which age and BMI might interact with the association between THF and cognitive function requires further study. In addition, our study shows that the relationship between serum THF and cognitive function differs among different types of cognitive activities. Therefore, we need to refine different types of cognitive functions to obtain more reliable results regarding the mechanism of association between serum THF and cognitive functions.

Our study was subject to several limitations. First, the cross-sectional design of this study did not allow for determination of a causal relationship between serum THF and cognitive functions. In addition, hearing loss despite being a high-risk factor for dementia, was not adjusted for in the current study due to lack of availability of pure-tone-average data [46]. Also, the small sample size will reduce the precision of the results in all three multivariate logistic regression models. Large, randomized, controlled trials should be conducted to further characterize the association between serum THF and cognitive functions. Furthermore, the retrospective nature of the study did not allow for control of all behavioral, medical, and environmental factors that can influence cognitive function [31]. However, we have used a variety of statistical methods to analyze the data to exclude the influence of other factors, including age, gender, and education level.

These results shed new light on the phenomenon that folate remains inversely associated with dementia. Insufficient utilization of THF converted from folate may be one of the main causes of cognitive impairment. The underlying mechanism may be related to glycine. THF is a key source of glycine, which inhibits glutamate-induced damage to brain tissue [47]. THF can accumulate in the blood when glycine synthesis is reduced. However, previous studies suggested that appropriate folate supplementation may be an effective measure to prevent cognitive impairment [16–18 , 48–50]. Achieving optimal nutritional status to prevent folate-related cognitive impairment is challenging. Our results show that high serum THF status is a risk factor of cognitive impairment, which indicates that high-dosing folate supplements are not necessarily good for humans due to tissue saturation [51]. A previous study found that over-supplementation with folate had negative effects on concentrations of vitamins A, B1, B2, and B6 [52]. In addition, high levels of serum THF indicate decreased availability of THF in tissues. Therefore, serum THF is a risk factor for decreased cognitive functions, which may be associated with incidence of dementia. Future studies should focus on folate dosing to increase the availability of THF and to reduce serum THF levels.

Conclusion

Our study evaluates the relationship between serum THF and cognitive functions, and it shows that serum THF levels are associated with worse performance on cognitive tests. In particular, poor learning ability and memory, as determined by CERAD-WL and CERAD-DR scores, are associated with high levels of serum THF. But their relationship needs to be discussed in conjunction with function/processing speed and categorical verbal fluency. Serum THF concentrations greater than or equal to 2.8 nmol/L are associated with decreased executive function/processing speed and global cognitive function. The association between increased serum THF and worse categorical verbal fluency is dependent on serum THF status (≥1.1 nmol/L), age (<70 years old), and BMI (>30). This study identifies potential risk factors for decreased cognitive functions that may aid in early diagnosis and treatment of dementia.

Footnotes

ACKNOWLEDGMENTS

The present study was supported by the Traditional Chinese Medicine Bureau of Guangdong Province, China [20221348]; the Natural Science Foundation of Guangdong Province [2022A151510450]; the Shenzhen Science and Technology Innovation Committee subject [JCYJ20210324123614040]; Bao’an TCM Development Foundation [2020KJCX-KTYJ-131, 2020KJCX-KTYJ-133 and 2020KJCX-KTYJ-134]; and Sanming Project of Medicine in Shenzhen [SZZYSM202106009].

Authors’ disclosures available online ().

ETHICS STATEMENT

The studies involving human participants were reviewed and approved by the Research Ethics Committee of the National Public Health Institute. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin ( accessed on 09 December 2021).

DATA AVAILABILITY STATEMENT

The data are available to the public ( accessed on 09 December 2021).

References

Goodman

, Lochner

, Thambisetty

, Wingo

, Posner

, Ling

(2017) Prevalence of dementia subtypes in United States Medicare fee-for-service beneficiaries, 2011-2013. Alzheimers Dement 13, 28–37.

World Health Organization (2020) Global Dementia Observatory. https://www.who.int/data/gho/data/themes/global-dementia-observatory-gdo, Last updated September 21, 2020, Accessed on December 09 2021.

Zukotynski

, Kuo

, Mikulis

, Rosa-Neto

, Strafella

, Subramaniam

, Black

(2018) PET/CT of Dementia. AJR Am J Roentgenol 211, 246–259.

Lei

, Cheng

, Frangi

, Wei

, Yu

, Liang

, Mai

, Duan

, Nong

, Li

, Su

, Wang

, Zhao

, Deng

, Zhang

(2021) Auto-weighted centralised multi-task learning via integrating functional and structural connectivity for subjective cognitive decline diagnosis. Med Image Anal 74, 102248.

Wang

, Wang

, Liu

, Mao

, Dong

, Li

, Huang

, Lei

, Chu

, Wang

, Gao

(2022) Random forest model in the diagnosis of dementia patients with normal mini-mental state examination scores. J Pers Med 12, 37.

Bédard

, Kergoat

, Leclerc

(2015) Systematic review of vision-related quality of life questionnaires for older institutionalised seniors with dementia. Ophthalmic Physiol Opt 35, 377–387.

Martínez

, Vernooij

, Fuentes Padilla

, Zamora

, Bonfill Cosp

, Flicker

(2017) 18F PET with florbetapir for the early diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 11, CD012216.

Kokkinou

, Beishon

, Smailagic

, Noel-Storr

, Hyde

, Ukoumunne

, Worrall

, Hayen

, Desai

, Ashok

, Paul

, Georgopoulou

, Casoli

, Quinn

, Ritchie

(2021) Plasma and cerebrospinal fluid Abeta42 for the differential diagnosis of Alzheimer’s disease dementia in participants diagnosed with any dementia subtype in a specialist care setting. Cochrane Database Syst Rev 2, CD010945.

Feldman

, Jacova

, Robillard

, Garcia

, Chow

, Borrie

, Schipper

, Blair

, Kertesz

, Chertkow

(2008) Diagnosis and treatment of dementia: 2. Diagnosis. CMAJ 178, 825–836.

10.

Creavin

, Wisniewski

, Noel-Storr

, Trevelyan

, Hampton

, Rayment

, Thom

, Nash

, Elhamoui

, Milligan

, Patel

, Tsivos

, Wing

, Phillips

, Kellman

, Shackleton

, Singleton

, Neale

, Watton

, Cullum

(2016) Mini-Mental State Examination (MMSE) for the detection of dementia in clinically unevaluated people aged 65 and over in community and primary care populations. Cochrane Database Syst Rev 2016, CD011145.

11.

Huskisson

, Maggini

, Ruf

(2007) The influence of micronutrients on cognitive function and performance. J Int Med Res 35, 1–19.

12.

Taylor

, Mahnken

, Sullivan

(2020) NHANES 2011-2014 reveals cognition of US older adults may benefit from better adaptation to the Mediterranean diet. Nutrients 12, 1929.

13.

Wang

, Zhu

, Xing

, Jia

, Tang

(2021) B vitamins and prevention of cognitive decline and incident dementia: A systematic review and meta-analysis. Nutr Rev 80, 931–949.

14.

Bailey

, Ayling

(2009) The extremely slow and variable activity of dihydrofolate reductase in human liver and its implications for high folic acid intake. Proc Natl Acad Sci U S A 106, 15424–15429.

15.

Abali

, Skacel

, Celikkaya

, Hsieh

(2008) Regulation of human dihydrofolate reductase activity and expression. Vitam Horm 79, 267–292.

16.

Ramos

, Allen

, Mungas

, Jagust

, Haan

, Green

, Miller

(2005) Low folate status is associated with impaired cognitive function and dementia in the Sacramento Area Latino Study on Aging. Am J Clin Nutr 82, 1346–1352.

17.

Min

, Min

(2016) The folate-vitamin B12 interaction, low hemoglobin, and the mortality risk from Alzheimer’s disease. J Alzheimers Dis 52, 705–712.

18.

Nguyen

, Gracely

, Lee

(2013) Serum folate but not vitamin B-12 concentrations are positively associated with cognitive test scores in children aged 6-16 years. J Nutr 143, 500–504.

19.

Diaz

, Custodio

, Montesinos

, Lira

, Herrera-Perez

, Pintado-Caipa

, Cuenca-Alfaro

, Gamboa

, Lanata

(2021) Thyroid dysfunction, Vitamin B12, and folic acid deficiencies are not associated with cognitive impairment in older adults in Lima, Peru. Front Public Health 9, 676518.

20.

Bailey

, Jun

, Murphy

, Green

, Gahche

, Dwyer

, Potischman

, McCabe

, Miller

(2020) High folic acid or folate combined with low vitamin B-12 status: Potential but inconsistent association with cognitive function in a nationally representative cross-sectional sample of US older adults participating in the NHANES. Am J Clin Nutr 112, 1547–1557.

21.

Morris

, Evans

, Bienias

, Tangney

, Hebert

, Scherr

, Schneider

(2005) Dietary folate and vitamin B12 intake and cognitive decline among community-dwelling older persons. Arch Neurol 62, 641–645.

22.

Centers for Disease Control and Prevention (CDC), National Center for Health Statistics (NCHS) (2020) National Health and Nutrition Examination Survey Data.

23.

Fazili

, Whitehead

Jr , Paladugula

, Pfeiffer

(2013) A high-throughput LC-MS/MS method suitable for population biomonitoring measures five serum folate vitamers and one oxidation product. Anal Bioanal Chem 405, 4549–4560.

24.

Pfeiffer

, Fazili

, McCoy

, Zhang

, Gunter

(2004) Determination of folate vitamers in human serum by stable-isotope-dilution tandem mass spectrometry and comparison with radioassay and microbiologic assay. Clin Chem 50, 423–432.

25.

Fazili

, Pfeiffer

(2004) Measurement of folates in serum and conventionally prepared whole blood lysates: Application of an automated 96-well plate isotope-dilution tandem mass spectrometry method. Clin Chem 50, 2378–2381.

26.

Fazili

, Pfeiffer

(2013) Accounting for an isobaric interference allows correct determination of folate vitamers in serum by isotope dilution-liquid chromatography-tandem MS. J Nutr 143, 108–113.

27.

Morris

, Heyman

, Mohs

, Hughes

, van Belle

, Fillenbaum

, Mellits

, Clark

(1989) The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer’s disease. Neurology 39, 1159–1165.

28.

Alam

, Singleton

, Aguiñaga

, Chodzko-Zajko

, Jahan

, Oke

, Schwingel

(2021) Is acculturation associated with the cognitive performance of older Hispanics? J Alzheimers Dis 85, 535–544.

29.

Brody

, Kramarow

, Taylor

, McGuire

(2019) Cognitive performance in adults aged 60 and over: National Health and Nutrition Examination Survey, 2011-2014. Natl Health Stat Report, pp 1-23.

30.

Arab

, Ang

(2015) A cross sectional study of the association between walnut consumption and cognitive function among adult US populations represented in NHANES. J Nutr Health Aging 19, 284–290.

31.

Liu

, Qiao

, Zhuang

, Xu

, Chen

, Lai

, Wang

(2021) Choline intake correlates with cognitive performance among elder adults in the United States. Behav Neurol 2021, 2962245.

32.

Beydoun

, Beydoun

, Gamaldo

, Teel

, Zonderman

, Wang

(2014) Epidemiologic studies of modifiable factors associated with cognition and dementia: Systematic review and meta-analysis. BMC Public Health 14, 643.

33.

Langa

, Levine

(2014) The diagnosis and management of mild cognitive impairment: A clinical review. JAMA 312, 2551–2561.

34.

Botelho

, Lyra

, Proença

, Godinho

, Mendes

, Machado

(2020) Relationship between blood and standard biochemistry levels with periodontitis in Parkinson’ disease patients: Data from the NHANES 2011-2012. J Pers Med 10, 69.

35.

Lyra

, Machado

, Proença

, Mendes

, Botelho

(2021) Tooth loss and blood pressure in Parkinson’s disease patients: An exploratory study on NHANES data. Int J Environ Res Public Health 18, 5032.

36.

Livingston

, Huntley

, Sommerlad

, Ames

, Ballard

, Banerjee

, Brayne

, Burns

, Cohen-Mansfield

, Cooper

, Costafreda

, Dias

, Fox

, Gitlin

, Howard

, Kales

, Kivimäki

, Larson

, Ogunniyi

, Orgeta

, Ritchie

, Rockwood

, Sampson

, Samus

, Schneider

, Selbæk

, Teri

, Mukadam

(2020) Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet 396, 413–446.

37.

Baiano

, Barone

, Trojano

, Santangelo

(2020) Prevalence and clinical aspects of mild cognitive impairment in Parkinson’s disease: A meta-analysis. Mov Disord 35, 45–54.

38.

Vandenbroucke

, von Elm

, Altman

, Gøtzsche

, Mulrow

, Pocock

, Poole

, Schlesselman

, Egger

(2007) Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): Explanation and elaboration. Ann Intern Med 147, W163–W194.

39.

Johnson

, Dohrmann

, Burt

, Mohadjer

(2014) National health and nutrition examination survey: Sample design, 2011-2014. Vital Health Stat 2, 1–33.

40.

Liu

, Wang

, Lu

, Li

, Gong

, Sheng

, Hu

, Peng

, Sun

(2013) The effects of intraoperative cryoprecipitate transfusion on acute renal failure following orthotropic liver transplantation. Hepatol Int 7, 901–909.

41.

U. S. Centers for Disease Control Prevention, https://www.cdc.gov/nchs/nhanes/irba98.htm, last updated May 2021, Accessed on December 9, 2021.

42.

Liu

, Zhao

, Cheng

, Yu

, Cheng

, Fan

, Chen

(2018) AST to ALT ratio and arterial stiffness in non-fatty liver Japanese population: A secondary analysis based on a cross-sectional study. Lipids Health Dis 17, 275.

43.

Bahous

, Cosín-Tomás

, Deng

, Leclerc

, Malysheva

, Ho

, Pallàs

, Kaliman

, Bedell

, Caudill

, Rozen

(2019) Early manifestations of brain aging in mice due to low dietary folate and mild MTHFR deficiency. Mol Neurobiol 56, 4175–4191.

44.

Jiang

, Wang

, Lu

, Chen

, Teng

, Li

, Xie

, Zhao

(2021) Brain imaging changes and related risk factors of cognitive impairment in patients with heart failure. Front Cardiovasc Med 8, 838680.

45.

Kreusler

, Vogel

, Willenberg

, Baber

, Dietz

, Körner

, Ceglarek

, Kiess

(2021) Folate and cobalamin serum levels in healthy children and adolescents and their association with age, sex, BMI and socioeconomic status. Nutrients 13, 546.

46.

Gates

, Mills

(2005) Presbycusis. Lancet 366, 1111–1120.

47.

, Zhou

, Jiang

, Wang

, Zhang

, Li

, Xu

, Liu

, Wang

, Zhu

(2021) Dose-dependent variation of synchronous metabolites and modules in a yin/yang transformation model of appointed ischemia metabolic networks. Front Neurosci 15, 645185.

48.

Zheng

, Chen

, Ma

, Wang

, Liang

, Ma

(2021) B vitamins supplementation can improve cognitive functions and may relate to the enhancement of transketolase activity in a rat model of cognitive impairment associated with high-fat diets. Curr Med Sci 41, 847–856.

49.

Bai

, Fan

, Li

, Dong

, Gao

, Fu

, Huang

, Liu

(2021) Effects of folic acid combined with DHA supplementation on cognitive function and amyloid-β-related biomarkers in older adults with mild cognitive impairment by a randomized, double blind, placebo-controlled trial. J Alzheimers Dis 81, 155–167.

50.

McNulty

, Ward

, Hoey

, Hughes

, Pentieva

(2019) Addressing optimal folate and related B-vitamin status through the lifecycle: Health impacts and challenges. Proc Nutr Soc 78, 449–462.

51.

Murphy

MSQ

, Muldoon

, Sheyholislami

, Behan

, Lamers

, Rybak

, White

, Harvey

ALJ

, Gaudet

, Smith

, Walker

, Wen

, MacFarlane

(2021) Impact of high-dose folic acid supplementation in pregnancy on biomarkers of folate status and 1-carbon metabolism: An ancillary study of the Folic Acid Clinical Trial (FACT). Am J Clin Nutr 113, 1361–1371.

52.

Barron

, Ogretmen

, Krupenko

(2021) Dietary folic acid alters metabolism of multiple vitamins in a CerS6- and sex-dependent manner. Front Nutr 8, 758403.