Association Between Dietary and Serum Cholesterol and Cognitive Function Among the U.S. Elderly from NHANES 2011-2014

Abstract

Background:

The association between dietary or serum cholesterol and cognitive performance in older adults has not been well-established.

Objective:

This study aimed to investigate the potential association between dietary or serum cholesterol and cognitive performance in the elderly population.

Methods:

A cross-sectional analysis was conducted using data from the National Health and Nutrition Examination Survey (NHANES) 2011-2012 and 2013-2014. Diet and supplement cholesterol was estimated based on two non-consecutive 24-hour dietary recalls. Cognitive function was assessed using various statistical tests. Poor cognitive performance was defined as scores below the lowest quartile within age groups. Regression models were adjusted for demographic factors, and subgroup analyses were performed for non-Hispanic White (NHW) and non-Hispanic Black (NHB) individuals.

Results:

Among 759 participants aged 60 years and above, dietary cholesterol was only associated with dietary saturated fatty acids and serum high-density lipoprotein cholesterol. There was no evidence of an association between dietary cholesterol and cognitive function, except for NHB individuals, where dietary cholesterol showed a positive correlation with cognitive function. In the overall sample and NHW participants, there were consistent positive associations between serum total cholesterol and cognitive performance across statistical tests, while such associations were rare among NHB individuals. Although not statistically significant, NHB individuals had higher dietary/supplementary/total cholesterol intake compared with NHW individuals.

Conclusion:

Within the normal range, increasing serum cholesterol may be a potential factor to prevent or relieve cognitive dysfunction. However, ethnic differences should be taken into account when considering the association between cholesterol and cognitive performance.

Keywords

Alzheimer’s disease cholesterol cognitive function dietary NHANES

INTRODUCTION

As individuals age, cognitive dysfunction and dementia become increasingly common in the United States [1]. Alzheimer’s disease (AD) is the main type of dementia [2]. Progression from cognitive decline to AD is continuous and irreversible. Early intervention through lifestyle changes or rehabilitation therapy can be recommended for protection against AD, particularly during its prodromal stage known as mild cognitive impairment (MCI) [3]. The pathophysiology of AD is complex and involves various mechanisms such as oxidative stress, inflammatory response, and deposition of amyloid-β protein (Aβ) [4]. Unrecognized interactions may also exist among these mechanisms [5]. As the disease progresses, these pathological changes spread throughout different brain regions, affecting memory, thinking abilities, and other cognitive functions. The hippocampus, a brain region crucial for memory formation, is the earliest areas affected by AD. Gradually, other regions involved in language, judgment, and executive function also become affected [6]. Previous evidence suggested that an imbalance in dietary structure may be a modifiable risk factor for cognitive decline [7]. For instance, insufficient intake of carotenoids could accelerate the process of cognitive dysfunction [8].

Dietary cholesterol could have both positive and negative effects on the body, and several clinical studies have shown conflicting conclusions [9, 10]. On the one hand, cholesterol is crucial for the formation and maintenance of synaptic connections, supporting communication between neurons and cognitive processes [11]. Additionally, cholesterol plays a role in promoting neural recovery and preventing MCI during compensatory repair processes after damage to neural pathways [11]. In a long-term follow-up study by Ylilauri et al., it was found that intake of cholesterol and eggs does not increase the risk of dementia or AD in elderly Finnish men [9]. On the contrary, egg consumption was associated with positive performance in certain cognitive domains [9]. On the other hand, the relationship between cholesterol levels and cognitive decline is complex and condition-dependent. Cholesterol overload could lead to vascular and microvascular diseases that have adverse effects on cognition [12, 13]. Long-term exposure to high levels of cholesterol also promotes plaque formation in arteriosclerotic lesions leading to narrowed blood vessels with reduced blood flow affecting oxygenation and nutrient delivery to the brain [14]. This may result in cognitive deficits such as attention decline, memory impairment, or executive dysfunction [13]. Furthermore, excessive cholesterol affects Aβ production, aggregation and clearance within the brain, all hallmark pathological features of AD [15]. Clinical research conducted on Irish individuals has also confirmed a significant increase in cholesterol and sodium intake among AD patients with cognitive impairment [10].

Studies examining the link between serum cholesterol levels and cognitive decline often use predetermined thresholds based on population reference ranges or clinical guidelines. Generally speaking, researchers may consider a serum total cholesterol (TC) level above 6.22 mmol/L or low-density lipoprotein cholesterol (LDL-C) level above 4.14 mmol/L as indicative of high cholesterol [16]. The central nervous system (CNS) has its own independent system for lipoprotein supply and circulation [17]. Cholesterol is synthesized within the CNS, with little obtained from peripheral blood [18]. Cholesterol supplies the various needs of neurons and contributes to the composition of the myelin sheath [18, 19]. Low cholesterol levels may impair cognitive function by destroying the permeability of rat blood-brain barrier, making the CNS more vulnerable to external injury [20]. Moreover, increased cholesterol in neuronal membranes may also function as an antioxidant, protecting against oxygen free radicals [21]. A six-year nutritional survey found that patients with extremely low LDL-C, due to long-term diet adjustment or prescription medication, tended to have significantly worse reaction times and cognitive memory. The same study also found a significant association between low serum TC, low serum non-high-density lipoprotein (HDL) cholesterol (NHDLC), and slow visuomotor speed in male patients [22]. A longitudinal study over three years has also suggested that reducing cholesterol levels may lead to an increase in symptoms of depression [23]. The contradictory results of these studies may be influenced by various factors such as differences in subjects’ serum TC, hypercholesterolemia definition, and intake of cholesterol-lowering medications.

Our investigation specifically aimed to examine the association between dietary cholesterol intake, supplement cholesterol consumption, total cholesterol levels, serum cholesterol-related indicators, and cognitive performance, with the objective of identifying targeted interventions for the prevention and treatment of cognitive impairment in older adults. To accomplish this objective, we selected a sample of older adults that is representative at the national level from the National Health and Nutrition Examination Survey (NHANES) [24]. Moreover, given the limited literature on the influence of ethnicity on the association between cholesterol and cognitive function, we further analyzed whether these associations varied between non-Hispanic Blacks (NHB) and non-Hispanic Whites (NHW) participants. We hoped to explore effective interventions for preventing or treating cognitive impairment and AD in olderadults.

METHODS

Study design

The NHANES study aims to provide a representative sample of the U.S. civilian population. For this cross-sectional analysis, we used two cycles (2011-2012 and 2013-2014) of NHANES data that included information on dietary cholesterol, serum cholesterol, and cognitive performance [24]. We intended to analyze additional cycles but were unable to do so due to incomplete records of cognitive performance in those cycles. The CDC’s National Center for Health Statistics Ethics Review Board approved all NHANES programs, and all participants provided written informed consent [24].

The analysis presented in this study includes a summary sample of all people aged≥60 years old in the two survey periods. It includes complete cognitive performance scores, dietary and serum cholesterol-related information, as well as other important characteristics such as age, sex, ethnicity, income level, education level, smoking status, drinking, doctor hypercholesterolemia notification and prescription guidance. Of the 3,632 participants aged≥60 years old who took part in the surveys periods; only 759 had complete data that constituted our analysis dataset. All participants included in our study were informed by their doctors whether they had hypercholesterolemia or not which increased data accuracy and reliability.

Cholesterol in diet, supplement, and serum

During each cycle of NHANES, participants reported their detailed dietary intake for two 24-hour periods [25]. A nutrient analysis was conducted to determine energy, nutrient, and other food components consumed. An in-person interview was conducted during the NHANES visit to collect the first 24-hour recall and a telephone interview within three to ten days collected the second 24-hour recall. In our analyses, we calculated the average estimated dietary cholesterol intake (in micrograms) during the two recall periods or enrolled only one recall if there was no second recall available. Supplement use was also recorded during these two-day periods and included in total cholesterol intake calculations as the sum of dietary and supplement intake.

The laboratory indicators, which included TC, HDL-C, and LDL-C, were obtained from all enrolled participants. Serum samples were extracted in the morning and kept frozen at -30°C before being shipped to the National Center for Environmental Health for testing. TC was measured using two methods: LBXSCH represented the concentration measured by DxC800 with a timed-endpoint method, while LBXTC represented the concentration measured by a Roche Modular P chemistry analyzer with an enzymatic method. It should be noted that the LBXTC method is more recommended in clinical studies according to the guidelines provided by the NHANES official website (https://wwwn.cdc.gov/Nchs/Nhanes/2011-2012/BIOPRO_G.htm#LBXSCH).

Cognitive outcomes

Cognitive test results of participants aged 60 or above from NHANES 2011-2012 and 2013-2014 were recorded [8]. Trained examiners conducted assessments at the Mobile Examination Center using a three-part cognitive test consisting of Consortium to Establish a Registry for Alzheimer’s disease Word Learning (CERAD W-L) Test, Animal Fluency Test, and Digit Symbol Substitution Test (DSST). The CERAD W-L Test evaluates memory through immediate and delayed word recall. The Animal Fluency Test assesses semantic fluency and executive functioning by having participants name as many animals as possible in one minute. The DSST measures processing speed, attention, and working memory through symbol-number pairing. The CERAD W-L Test includes three consecutive learning tests and one delayed recall test according to published quality standards [26 –28]. Participants who scored below the minimum quartile on all three tests in their respective age groups were defined as having poor cognitive performance, We chose these cut-off points based on previous research and clinical practice experience [26, 27]. Cut-off points for each age group are listed in Supplementary Table 1.

Statistical analysis

We used EmpowerStats (version 4.0, http://www.empowerstats.com), a statistical analysis platform based on R software, for data analysis. EmpowerStats contains the underlying engine of SAS and can perform complex statistical analyses on weighted data. In addition to the target independent variables (cholesterol-related indicators in diet and serum) and target dependent variable (seven-item cognitive test scores), our study also involved multiple feature variables including age, gender, ethnicity, income level, education level, smoking status, drinking, doctor hypercholesterolemia notification and doctor prescription guidance. As this study evaluated results from seven-item tests examining cognitive function, Bonferroni adjustment was applied to adjust the significance level to p = 0.007 (0.05/7) when examining relationships between any test of cognitive function with target independent variables or feature variables [29].

Before regression modeling was conducted, we examined associations between each test score of cognitive function and each feature variable. We performed some data transformations where age, drinking and BMI as well as cholesterol-related indicators were continuous when testing Spearman correlation with cognitive scores, but were transformed into quartile or categorical variables for the Kruskal-Wallis test and Chi-squared test, respectively. As mentioned earlier in Chi-square tests the dependent variable (cognition test score) was treated as a binary outcome distinguishing normal cognition performance from poor cognition performance. Detailed information is provided in Table 1.

Table 1

Characteristics of enrolled participants from

Characteristic	All NHANES participants aged ≥60 years in 2011-2012 & 2013-2014 (n = 759)
	N	%
Age
60–69 y	424	55.86%
70–79 y	228	30.04%
80 + y	107	14.10%
Gender
Male	398	52.44%
Female	361	47.56%
Race/ethnicity
Mexican American	53	6.98%
Other Hispanic	65	8.56%
Non-Hispanic white	433	57.05%
Non-Hispanic Black	149	19.63%
Other/multiracial	59	7.77%
BMI
Underweight (<18.5)	8	1.05%
Normal (18.5 to < 25)	201	26.48%
Overweight (25 to < 30)	258	33.99%
Obese (30 or greater)	292	38.47%
Alcohol intake
<1 drinks/month	338	44.53%
1 to < 5 drinks/month	200	26.35%
5 to < 10 drinks/month	52	6.85%
10 + drinks/month	169	22.27%
Smoking status
<100 cigarettes/life	411	54.15%
>100 cigarettes/life	348	45.85%
Educational attainment
Less than 9th grade	64	8.43%
9–11th grade (12th grade with no diploma)	83	10.94%
High school graduate/GED	173	22.79%
Some college or AA	228	30.04%
College graduate or above	211	27.80%
Annual family income
< $20000	157	20.69%
> $20000	602	79.32%
Doctor hypercholesterolemia notification
Yes	448	59.03%
No	311	40.98%
Doctor prescription guidance
Yes	403	53.10%
No	356	46.90%
24-hour interview recall of dietary
Only one recall	45	5.93%
Both two recalls	714	94.07%
Serum cholesterol (LBXTX)
<6.22 mmol/L	674	88.80%
≥6.22 mmol/L	84	11.07%

	Mean	SD	Median	Q1-Q3
Age (y)	68.47	6.59	67.00	63.00, 73.00
BMI	29.13	6.40	27.70	24.60, 32.60
Alcohol intake	7.47	9.41	2.00	0.25, 12.86
CERAD W-L Trial 1 Score	5.16	1.68	5.00	4.00, 6.00
CERAD W-L Trial 2 Score	7.13	1.83	7.00	6.00, 8.00
CERAD W-L Trial 3 Score	7.87	1.74	8.00	7.00, 9.00
CERAD W-L Delayed Recall Score	6.49	2.26	7.00	5.00, 8.00
CERAD W-L Total Score (3 Recall trials)	20.16	4.58	21.00	17.00, 23.00
Animal Fluency: Total Score	18.87	5.54	19.00	15.00, 22.00
Digit Symbol: Score	55.26	17.09	55.00	44.00, 66.00
Dietary cholesterol intake (mg/day)	279.42	162.36	254.50	161.00, 366.50
Supplement cholesterol intake (mg/ day) (n = 122)	10.37	6.96	10.00	9.11, 11.63
Total cholesterol intake (mg/day) (n = 122)	268.31	141.74	248.00	160.50, 331.38
Cholesterol, total (mg/dl) (LBXSCH)	193.53	43.58	191.00	164.00, 217.00
Cholesterol, total (mg/dl) (LBXTC)	193.77	42.19	192.00	165.00, 220.00
HDL-cholesterol (mg/dl)	57.30	17.50	53.00	45.00, 67.00
LDL-Cholesterol (mg/dl)	111.49	36.30	111.00	87.00, 131.00

We utilized non-weighted Spearman correlation tests to explore associations between continuous form feature variables or target dependent variable and cognitive performance. To determine possible non-linear correlations, we further examined the data that were transformed into quartiles or categorical form, using Kruskal-Wallis tests or Chi-squared tests, respectively. To allow for potential confounding due to the strong association between fat and cholesterol, we first evaluated the correlations between dietary fat intake (total fat, saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids) and dietary cholesterol. We then examined the associations between cognitive scores and dietary fat intake.

To estimate correlations between targeted variable(s) with individual CERAD W-L Test score range: (0-10 points), CERAD W-L Total Test score range: (0-30 points), Animal Fluency Test score range: (0-39 points), and DSST score range: (0-105 points), we used a linear regression model (SURVEYREG procedure). We selected potential confounding variables using previous results from Pearson correlation and Kruskal-Wallis tests. Then, we adjusted for these selected variables in the linear model. Considering that age is the biggest confounding factor affecting cognition, linear regression was performed three times with 1) no covariates; 2) only setting age as a covariate; and 3) setting age, ethnicity, family income, education level and prescription guidance as covariates. Finally, we conducted stratified linear regression analysis to evaluate the associations between serum cholesterol-related indicators and cognitive performance separately among NHW and NHB, in which we excluded ethnicity and included only age, family income, education level, and prescription guidance as covariates. T-tests were used to compare differences in cholesterol intake between the population from NHW and NHB.

RESULTS

In the two NHANES cycles, 759 participants were screened for cognitive performance, serum cholesterol, and dietary cholesterol. The study also collected information on necessary covariates such as age, gender, ethnicity, income level, education level, smoking status, drinking, doctor hypercholesterolemia notification and doctor prescription guidance. Table 1 shows the characteristics of the participants with an even distribution in terms of gender. Most participants (55.86%) were aged between 60 to 69 years old at the time of examination. The majority of participants were NHW (nearly 57.05%), followed by NHB (19.63%). About one quarter of people had a normal or underweight BMI while one third was overweight and the rest obese. Two-thirds drank alcohol less than five times per month while an equal number smoked more than or less than 100 cigarettes/life. A high school education was most common among participants. 45 participants of the total 759 participants had only one recall.

Unweighted Spearman correlation analysis showed no association between cognitive test scores and dietary cholesterol intake but significant correlations existed with age, serum TC, serum HDL-C, and serum LDL-C (Supplementary Table 2). Although this was only true for some cognitive scores with LDL-C, Kruskal-Wallis test revealed significant differences in almost every cognitive tests among quartiles of serum TC but no difference in HDL-C and dietary cholesterol intake, when taking into account a Bonferroni adjustment (Supplementary Table 3). Kruskal-Wallis test results also showed significant differences in at least two cognitive tests for all subgroups, except for BMI, drinking, smoking status, and doctor hypercholesterolemia notification (Supplementary Table 4). Unweighted Spearman correlation analysis indicated that dietary cholesterol may be correlated with dietary saturated fatty acids and serum HDL-C (Pearson correlation coefficient: 0.08, p = 0.04 and -0.10, p = 0.01), but no correlation with total fat, monounsaturated fatty acid, polyunsaturated fatty acid and serum TC or LDL-C (Table 2). Dietary fat was almost unrelated to cognitive function (Supplementary Table 5).

Table 2

Unweighted Spearman coefficients (standard error (SE), p-value) for correlation between cholesterol intake and serum cholesterol related indicators or fat intake

Var 1	Var 2	N	Correlation	Std.err	95% CI low	95% CI upp	t	p
Dietary cholesterol	Cholesterol, total (LBXSCH)	759	–0.03	0.04	–0.10	0.05	–0.70	0.49
Dietary cholesterol	Cholesterol, total (LBXTC)	759	–0.02	0.04	–0.09	0.05	–0.54	0.59
Dietary cholesterol	HDL-cholesterol	759	–0.10	0.04	–0.17	–0.03	–2.78	0.01
Dietary cholesterol	LDL-Cholesterol	759	0.03	0.04	–0.04	0.10	0.74	0.46
Dietary cholesterol	Total fat	759	–0.02	0.04	–0.09	0.05	–0.5	0.62
Dietary cholesterol	Saturated fatty acid	735	0.08	0.04	<0.01	0.15	2.03	0.04
Dietary cholesterol	Monounsaturated fatty acid	759	–0.03	0.04	–0.10	0.05	–0.69	0.49
Dietary cholesterol	Polyunsaturated fatty acid	759	<0.01	0.04	–0.07	0.07	0.01	0.99

Participants were classified into groups of poor cognitive performance and normal cognitive performance based on cut-off points. Using Chi-square test, significant differences existed in the incidence of poor cognitive performance among groups of gender, ethnicity, education level and income level (Supplementary Table 6). Looking at differences in cognitive performance between the lowest quartile and higher quartile groups for various lipid indicators, the only significant difference was seen for serum TC (Supplementary Table 7). Many participants were on cholesterol-lowering medication, complicating the use of clinical cut-offs. Moreover, only 12% of participants exceeded these cut-offs, leading to imbalanced distribution. Therefore, we used quartiles to capture potential nonlinear associations and ensure balanced, robust analysis. The linear regression model for individual CERAD W-L Test scores indicated that higher cognitive test scores had no association with dietary cholesterol while significantly positive association existed between higher cognitive test scores and serum cholesterol-related indicators (Table 3). Similar results were observed in CERAD W-L Total Test score, Animal Fluency Test and DSST (Table 4).

Table 3

Adjusted* beta coefficients (standard error (SE), p-value) for score on CERAD Word Learning sub-test, for each mg/day increase in cholesterol intake, and each mg/dL increase in serum cholesterol, HDL, and LDL (p <0.05 is set as nominal significant)

	CERAD W-L Trial1 Score					CERAD W-L Trial 2 Score
	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol
Q2 vs. Q1	-0.22 (-0.68, 0.25) p=0.36	0.60 (0.17, 1.04) p=0.01^#	0.59 (0.15, 1.02) p=0.01	0.47 (0.01, 0.92) p=0.04	0.52 (0.09, 0.94) p=0.02	0.03 (-0.43, 0.48) p=0.90	0.66 (0.22, 1.10) p<0.01^#	0.42 (-0.02, 0.87) p=0.06	0.23 (-0.20, 0.66) p=0.30	0.22 (-0.22, 0.66) p=0.32
Q2 vs. Q1, Age-adjusted	-0.27 (-0.70, 0.16) p=0.23	0.42 (0.01, 0.83) p=0.05	0.42 (0.01, 0.82) p=0.05	0.48 (0.06, 0.90) p=0.02	0.37 (-0.02, 0.77) p=0.07	-0.02 (-0.45, 0.40) p=0.90	0.48 (0.05, 0.90) p=0.03	0.25 (-0.17, 0.67) p=0.24	0.25 (-0.16, 0.66) p=0.24	0.08 (-0.34, 0.49) p=0.72
Q2 vs. Q1, Fully-adjusted*	-0.36 (-0.77, 0.05) p=0.09	0.38 (-0.01, 0.77) p=0.06	0.38 (-0.03, 0.79) p=0.07	0.50 (0.09, 0.91) p=0.02	0.34 (-0.06, 0.73) p=0.10	-0.13 (-0.54, 0.27) p=0.52	0.41 (-0.00, 0.81) p=0.05	0.18 (-0.23, 0.58) p=0.39	0.28 (-0.12, 0.68) p=0.17	0.01 (-0.39, 0.42) p=0.95
Q3 vs. Q1	0.08 (-0.35, 0.51) p=0.72	0.80 (0.38, 1.22) p<0.01^#	0.46 (0.02, 0.89) p=0.04	0.08 (-0.42, 0.57) p=0.76	0.67 (0.24, 1.11) p<0.01^#	0.30 (-0.13, 0.73) p=0.17	0.61 (0.18, 1.04) p=0.01^#	0.28 (-0.17, 0.72) p=0.22	-0.07 (-0.55, 0.40) p=0.77	0.37 (-0.06, 0.80) p=0.09
Q3 vs. Q1, Age-adjusted	-0.01 (-0.41, 0.39) p=0.95	0.60 (0.20, 1.00) p<0.01^#	0.26 (-0.16, 0.68) p=0.23	0.13 (-0.33, 0.60) p=0.58	0.46 (0.03, 0.88) p=0.03	0.21 (-0.20, 0.62) p=0.32	0.42 (0.01, 0.83) p=0.05	0.08 (-0.35, 0.51) p=0.70	-0.01 (-0.46, 0.44) p=0.96	0.15 (-0.26, 0.56) p=0.48
Q3 vs. Q1, Fully-adjusted*	-0.19 (-0.59, 0.21) p=0.35	0.50 (0.09, 0.91) p=0.02	0.19 (-0.24, 0.62) p=0.39	0.08 (-0.37, 0.53) p=0.73	0.40 (-0.02, 0.83) p=0.06	0.03 (-0.37, 0.44) p=0.88	0.28 (-0.15, 0.70) p=0.20	-0.03 (-0.46, 0.40) p=0.90	-0.06 (-0.49, 0.38) p=0.79	0.05 (-0.36, 0.46) p=0.81
Q4 vs. Q1	0.13 (-0.33, 0.60) p=0.58	0.75 (0.35, 1.15) p<0.01^#	0.72 (0.29, 1.15) p<0.01^#	0.57 (0.12, 1.03) p=0.01	0.72 (0.29, 1.15) p<0.01^#	0.29 (-0.16, 0.75) p=0.21	0.80 (0.37, 1.23) p<0.01^#	0.71 (0.26, 1.17) p<0.01 #	0.45 (0.00, 0.90) p=0.05	0.66 (0.22, 1.09) p<0.01^#
Q4 vs. Q1, Age-adjusted	-0.04 (-0.48, 0.40) p=0.85	0.56 (0.18, 0.95) p<0.01^#	0.533 (0.11, 0.95) p=0.01	0.69 (0.28, 1.10) p<0.01^#	0.50 (0.07, 0.92) p=0.02	0.12 (-0.31, 0.55) p=0.59	0.62 (0.20, 1.04) p<0.01^#	0.53 (0.09, 0.97) p=0.02	0.57 (0.16, 0.99) p=0.01^#	0.43 (0.00, 0.86) p=0.05
Q4 vs. Q1, Fully-adjusted*	-0.14 (-0.57, 0.29) p=0.53	0.50 (0.11, 0.89) p=0.01	0.46 (0.03, 0.90) p=0.04	0.55 (0.13, 0.97) p=0.01	0.45 (0.02, 0.88) p=0.04	0.03 (-0.39, 0.45) p=0.89	0.50 (0.08, 0.92) p=0.02	0.41 (-0.02, 0.85) p=0.06	0.44 (0.03, 0.85) p=0.04	0.34 (-0.08, 0.77) p=0.11
	CERAD W-L Trial 3 Score					CERAD W-L Delayed Recall Score
	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol
Q2 vs. Q1	0.04 (-0.36, 0.43) p=0.85	0.76 (0.34, 1.18) p<0.01^#	0.53 (0.10, 0.95) p=0.02	0.26 (-0.19, 0.70) p=0.26	0.18 (-0.25, 0.60) p=0.41	-0.10 (-0.70, 0.49) p=0.73	1.07 (0.50, 1.63) p<0.01^#	0.85 (0.28, 1.43) p<0.01^#	0.45 (-0.16, 1.06) p=0.15	0.66 (0.06, 1.27) p=0.03
Q2 vs. Q1, Age-adjusted	-0.02 (-0.40, 0.36) p=0.92	0.62 (0.21, 1.03) p<0.01^#	0.40 (-0.02, 0.82) p=0.06	0.27 (-0.16, 0.70) p=0.23	0.06 (-0.34, 0.47) p=0.80	-0.21 (-0.76, 0.35) p=0.47	0.81 (0.26, 1.36) p<0.01^#	0.62 (0.06, 1.17) p=0.03	0.47 (-0.11, 1.06) p=0.12	0.45 (-0.12, 1.03) p=0.12
Q2 vs. Q1, Fully-adjusted*	-0.12 (-0.48, 0.23) p=0.49	0.55 (0.15, 0.95) p<0.01	0.33 (-0.08, 0.74) p=0.11	0.31 (-0.10, 0.72) p=0.14	0.00 (-0.39, 0.40) p=0.98	-0.33 (-0.87, 0.21) p=0.23	0.78 (0.24, 1.32) p<0.01^#	0.57 (0.03, 1.12) p=0.04	0.52 (-0.04, 1.09) p=0.07	0.40 (-0.16, 0.95) p=0.17
Q3 vs. Q1	0.22 (-0.17, 0.61) p=0.27	0.69 (0.29, 1.09) p<0.01^#	0.52 (0.11, 0.93) p=0.01	0.24 (-0.20, 0.67) p=0.29	0.47 (0.08, 0.86) p=0.02	0.18 (-0.38, 0.74) p=0.52	0.91 (0.31, 1.52) p<0.01^#	0.72 (0.12, 1.31) p=0.02	0.27 (-0.34, 0.88) p=0.38	0.92 (0.36, 1.49) p<0.01^#
Q3 vs. Q1, Age-adjusted	0.14 (-0.24, 0.51) p=0.47	0.57 (0.18, 0.97) p<0.01^#	0.40 (-0.00, 0.79) p=0.05	0.30 (-0.13, 0.72) p=0.17	0.32 (-0.06, 0.70) p=0.10	0.03 (-0.49, 0.55) p=0.90	0.68 (0.10, 1.26) p=0.02	0.48 (-0.10, 1.06) p=0.11	0.38 (-0.21, 0.96) p=0.21	0.64 (0.09, 1.18) p=0.02
Q3 vs. Q1, Fully-adjusted*	-0.04 (-0.41, 0.33) p=0.83	0.44 (0.03, 0.85) p=0.03	0.30 (-0.10, 0.70) p=0.14	0.25 (-0.16, 0.66) p=0.23	0.22 (-0.15, 0.60) p=0.24	-0.19 (-0.72, 0.34) p=0.48	0.57 (-0.02, 1.17) p=0.06	0.40 (-0.18, 0.99) p=0.17	0.37 (-0.21, 0.95) p=0.21	0.56 (0.02, 1.10) p=0.04
Q4 vs. Q1	0.26 (-0.17, 0.68) p=0.23	0.88 (0.49, 1.26) p<0.01^#	0.81 (0.42, 1.21) p<0.01^#	0.56 (0.14, 0.98) p=0.01	0.67 (0.29, 1.05) p<0.01^#	0.19 (-0.42, 0.80) p=0.54	1.23 (0.67, 1.79) p<0.01^#	1.20 (0.63, 1.77) p<0.01^#	0.71 (0.11, 1.31) p=0.02	0.99 (0.39, 1.60) p<0.01^#
Q4 vs. Q1, Age-adjusted	0.12 (-0.29, 0.52) p=0.58	0.75 (0.36, 1.13) p<0.01^#	0.68 (0.29, 1.07) p<0.01^#	0.68 (0.28, 1.07) p<0.01^#	0.51 (0.13, 0.89) p=0.01	-0.07 (-0.64, 0.50) p=0.81	0.99 (0.44, 1.53) p<0.01^#	0.96 (0.41, 1.51) p<0.01^#	0.91 (0.35, 1.47) p<0.01^#	0.70 (0.10, 1.29) p=0.02
Q4 vs. Q1, Fully-adjusted*	0.04 (-0.35, 0.43) p=0.83	0.65 (0.26, 1.04) p<0.01^#	0.59 (0.20, 0.98) p<0.01^#	0.57 (0.18, 0.95) p<0.01^#	0.43 (0.05, 0.81) p=0.03	-0.17 (-0.73, 0.38) p=0.54	0.89 (0.34, 1.44) p<0.01^#	0.86 (0.31, 1.42) p<0.01^#	0.79 (0.23, 1.34) p<0.01^#	0.62 (0.03, 1.21) p=0.04

*Fully adjusted models are adjusted for age, ethnicity, income, education, and prescription guidance. ^#p value <0.007 is set as significant (Bonferroni adjustment).

Table 4

Adjusted* beta coefficients (standard error (SE), p) for score on CERAD Word Learning sub-test, Animal Fluency and DSST, for each mg/day increase in cholesterol intake, and each mg/dL increase in serum cholesterol, HDL, and LDL (p <0.05 is set as nominal significant)

	CERAD W-L Total score					Animal Fluency score
	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol
Q2 vs. Q1	-0.15 (-1.30, 1.00) p=0.80	2.03 (0.89, 3.15) p<0.01^#	1.53 (0.40, 2.67) p=0.01	0.95 (-0.20, 2.10) p=0.11	0.92 (-0.22, 2.05) p=0.12	0.47 (-0.94, 1.87) p=0.52	1.06 (-0.34, 2.46) p=0.14	1.03 (-0.36, 2.43) p=0.15	0.68 (-0.74, 2.10) p=0.35	0.89 (-0.51, 2.30) p=0.21
Q2 vs. Q1, Age-adjusted	-0.32 (-1.38, 0.75) p=0.56	1.52 (0.44, 2.59) p=0.01^#	1.07 (-0.01, 2.15) p=0.05	1.00 (-0.08, 2.08) p=0.07	0.51 (-0.54, 1.57) p=0.34	0.34 (-0.98, 1.66) p=0.61	0.470 (-0.86, 1.80) p=0.49	0.51 (-0.81, 1.83) p=0.45	0.75 (-0.63, 2.13) p=0.29	0.45 (-0.87, 1.77) p=0.51
Q2 vs. Q1, Fully-adjusted*	-0.62 (-1.62, 0.38) p=0.22	1.34 (0.32, 2.35) p=0.01	0.89 (-0.15, 1.93) p=0.09	1.09 (0.05, 2.13) p=0.04	0.35 (-0.66, 1.37) p=0.49	0.08 (-1.18, 1.33) p=0.90	-0.03 (-1.31, 1.26) p=0.97	0.06 (-1.21, 1.33) p=0.92	0.77 (-0.54, 2.08) p=0.25	-0.01 (-1.28, 1.26) p=0.99
Q3 vs. Q1	0.60 (-0.51, 1.71) p=0.29	2.09 (0.98, 3.21) p<0.01^#	1.25 (0.11, 2.39) p=0.03	0.24 (-0.98, 1.45) p=0.70	1.51 (0.41, 2.61) p=0.01^#	0.43 (-0.90, 1.75) p=0.53	2.40 (0.99, 3.81) p<0.01^#	2.34 (0.89, 3.78) p<0.01^#	0.98 (-0.36, 2.33) p=0.15	1.82 (0.37, 3.26) p=0.01
Q3 vs. Q1, Age-adjusted	0.33 (-0.70, 1.37) p=0.53	1.59 (0.54, 2.64) p<0.01^#	0.74 (-0.36, 1.83) p=0.19	0.42 (-0.73, 1.56) p=0.47	0.92 (-0.13, 1.97) p=0.09	0.16 (-1.06, 1.39) p=0.79	1.72 (0.38, 3.05) p=0.01	1.68 (0.30, 3.05) p=0.02	1.14 (-0.16, 2.44) p=0.09	1.10 (-0.30, 2.50) p=0.12
Q3 vs. Q1, Fully-adjusted*	-0.20 (-1.22, 0.82) p=0.70	1.22 (0.14, 2.31) p=0.03	0.46 (-0.63, 1.55) p=0.41	0.27 (-0.83, 1.38) p=0.63	0.68 (-0.35, 1.71) p=0.20	-0.81 (-1.92, 0.31) p=0.16	0.89 (-0.44, 2.22) p=0.19	1.11 (-0.22, 2.43) p=0.10	0.90 (-0.32, 2.13) p=0.15	0.54 (-0.77, 1.85) p=0.42
Q4 vs. Q1	0.68 (-0.50, 1.87) p=0.26	2.43 (1.36, 3.49) p<0.01^#	2.25 (1.11, 3.38) p<0.01^#	1.58 (0.43, 2.73) p=0.01#	2.05 (0.94, 3.15) p<0.01^#	0.14 (-1.22, 1.50) p=0.84	1.37 (0.68, 3.32) p=0.03	2.02 (0.68, 3.36) p<0.01^#	1.86 (0.38, 3.34) p=0.01	2.07 (0.68, 3.47) p<0.01^#
Q4 vs. Q1, Age-adjusted	0.19 (-0.92, 1.30) p=0.73	1.93 (0.89, 2.96) p<0.01^#	1.74 (0.65, 2.84) p<0.01^#	1.94 (0.91, 2.98) p<0.01^#	1.44 (0.36, 2.52) p=0.01	-0.40 (-1.69, 0.90) p=0.55	0.92 (0.13, 2.61) p=0.12	1.43 (0.16, 2.69) p=0.03	2.19 (0.78, 3.60) p<0.01^#	1.36 (0.01, 2.72) p=0.05
Q4 vs. Q1, Fully-adjusted*	-0.07 (-1.12, 0.99) p=0.90	1.64 (0.60, 2.68) p<0.01^#	1.47 (0.37, 2.56) p=0.01	1.56 (0.51, 2.60) p<0.01^#	1.22 (0.15, 2.29) p=0.03	-0.74 (-1.91, 0.42) p=0.21	0.78 (-0.42, 1.99) p=0.20	0.89 (-0.31, 2.09) p=0.15	1.45 (0.15, 2.76) p=0.03	0.84 (-0.44, 2.11) p=0.20

	Digit Symbol Score
	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol
Q2 vs. Q1	0.65 (-3.83, 5.12) p=0.78	5.24 (1.21, 9.28) p=0.01	7.42 (3.57, 11.26) p<0.01^#	0.92 (-3.31, 5.15) p=0.67	5.86 (2.03, 9.70) p<0.01^#
Q2 vs. Q1, Age-adjusted	-0.11 (-4.10, 3.89) p=0.96	2.77 (-0.95, 6.48) p=0.15	5.22 (1.77, 8.67) p<0.01^#	1.16 (-2.60, 4.93) p=0.55	3.96 (0.53, 7.39) p=0.02
Q2 vs. Q1, Fully-adjusted*	-2.02 (-5.57, 1.52) p=0.26	1.88 (-1.59, 5.35) p=0.29	4.51 (1.17, 7.85) p=0.01	1.82 (-1.67, 5.30) p=0.31	3.39 (0.06, 6.72) p=0.05
Q3 vs. Q1	2.69 (-1.63, 7.00) p=0.22	9.69 (5.88, 13.49) p<0.01^#	7.91 (4.02, 11.80) p<0.01^#	2.74 (-1.61, 7.08) p=0.22	9.02 (5.00, 13.04) p<0.01^#
Q3 vs. Q1, Age-adjusted	1.42 (-2.23, 5.07) p=0.45	7.16 (3.82, 10.50) p<0.01^#	5.42 (1.97, 8.86) p<0.01^#	3.55 (-0.36, 7.46) p=0.08	6.18 (2.59, 9.77) p<0.01^#
Q3 vs. Q1, Fully-adjusted*	-1.67 (-4.93, 1.59) p=0.31	5.22 (1.92, 8.52) p<0.01^#	4.43 (1.18, 7.68) p=0.01	3.00 (-0.52, 6.51) p=0.09	5.32 (1.98, 8.66) p<0.01^#
Q4 vs. Q1	0.85 (-3.60, 5.29) p=0.71	6.78 (5.42, 13.09) p<0.01^#	9.76 (5.92, 13.59) p<0.01	5.39 (1.08, 9.69) p=0.01	8.47 (4.62, 12.31) p<0.01^#
Q4 vs. Q1, Age-adjusted	-1.51 (-5.45, 2.44) p=0.45	5.55 (3.32, 10.25) p<0.01^#	7.37 (3.90, 10.84) p<0.01^#	7.05 (3.33, 10.77) p<0.01^#	5.58 (2.04, 9.11) p<0.01^#
Q4 vs. Q1, Fully-adjusted*	-3.13 (-6.57, 0.30) p=0.07	5.50 (2.13, 8.86) p<0.01^#	6.35 (2.98, 9.71) p<0.01^#	5.13 (1.67, 8.59) p<0.01^#	5.03 (1.61, 8.44) p<0.01^#

*Fully adjusted models are adjusted for age, ethnicity, income, education, and prescription guidance. ^#p <0.007 is set as significant (Bonferroni adjustment).

Regarding the variables considered in our study, it is important to highlight that four specific variables—age, ethnicity, income level, and education level—exhibited significant correlation with cognitive function. This was consistently observed in unweighted Spearman correlation tests, Kruskal-Wallis tests, and Chi-squared tests. In light of these findings, we made the decision to include these variables as covariates in our regression analysis, in conjunction with the prescription guidance (Supplementary Tables 2, 4, and 6). After adjusting for individual age or age combined with other covariates in the regression model, similar patterns of results were still observed (results not shown).

We also examined the associations mentioned above among NHW and NHB participants. However, due to a small sample size of NHB participants (n = 149), we did not further investigate the relationship between supplement cholesterol or total cholesterol intake and cognitive function in different ethnic groups. Table 5 indicates that there was still no correlation between dietary cholesterol and cognitive function among quartile groups in NHW participants, whereas a positive correlation was found in CERAD W-L Trial 2 Test score in NWB participants. Notably, the correlation between dietary cholesterol and CERAD W-L Delayed Recall Test score or Animal Fluency Test score still held significance, although it didn’t reach the threshold after Bonferroni correction. When examining the effect estimates of serum indicators, it was discovered that NHW participants had strong correlations between serum TC and cognitive function in almost all tests except for CERAD W-L Trial 1, CERAD W-L Trial 2 and Animal Fluency Test score. However, in NHB participants, we did not observe any association between serum cholesterol-related indicators and cognitive function.

Despite not being statistically significant, there are some modest differences in cholesterol intake (mean (SE) 275.22±149.66 mg versus 287.66±198.33 mg), total cholesterol intake (mean (SE) was 266.87±134.16 mg versus 292.41±180.90), and supplement cholesterol intake (mean (SE) 10.17±6.19 mg versus 12.27±11.10 mg) between NHW and NHB participants as shown in Supplementary Table 8. Notably, we found NHW participants showed significantly better cognitive performance in CERAD W-L Total Test, Animal Fluency Test and DSST.

Table 5

Adjusted* beta coefficients (SE, p) for score on CERAD Word Learning sub-test, Animal Fluency test, and DSST, for each mg/day increase in dietary cholesterol intake and each mg/dL in serum cholesterol related indicators, stratified by race/ethnicity (p value <0.05 is set as nominal significant)

	Non-Hispanic white					Non-Hispanic black
	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol	Dietary cholesterol	Cholesterol, total (LBXSCH)	Cholesterol, total (LBXTC)	HDL-cholesterol	LDL-Cholesterol
CERAD W-L Trial 1 Score
Q2 vs. Q1	-0.37 (-0.81, 0.07) p=0.10	0.33 (-0.11, 0.76) p=0.14	0.34 (-0.09, 0.77) p=0.12	0.50 (0.07, 0.94) p=0.02	0.30 (-0.11, 0.72) p=0.15	0.07 (-0.69, 0.83) p=0.85	0.49 (-0.30, 1.29) p=0.22	0.80 (-0.00, 1.61) p=0.05	-0.05 (-0.93, 0.83) p=0.91	0.28 (-0.54, 1.10) p=0.50
Q3 vs. Q1	-0.22 (-0.63, 0.20) p=0.31	0.49 (0.06, 0.92) p=0.03	0.12 (-0.31, 0.54) p=0.59	-0.01 (-0.44, 0.42) p=0.97	0.43 (0.01, 0.85) p=0.05	0.09 (-0.73, 0.91) p=0.83	0.08 (-0.71, 0.87) p=0.84	0.47 (-0.27, 1.21) p=0.22	-0.40 (-1.22, 0.42) p=0.34	0.25 (-0.58, 1.07) p=0.56
Q4 vs. Q1	-0.16 (-0.59, 0.26) p=0.45	0.50 (0.07, 0.92) p=0.02	0.46 (0.03, 0.88) p=0.03	0.52 (0.10, 0.95) p=0.02	0.44 (0.01, 0.87) p=0.04	0.46 (-0.29, 1.22) p=0.23	0.05 (-0.71, 0.80) p=0.91	0.23 (-0.53, 0.99) p=0.55	-0.25 (-1.06, 0.57) p=0.55	0.46 (-0.32, 1.25) p=0.25
CERAD W-L Trial 2 Score
Q2 vs. Q1	-0.17 (-0.63, 0.29) p=0.48	0.45 (-0.00, 0.91) p=0.05	0.21 (-0.24, 0.66) p=0.36	0.29 (-0.16, 0.75) p=0.21	0.01 (-0.43, 0.44) p=0.98	0.14 (-0.65, 0.92) p=0.73	0.38 -0.48, 1.23) p=0.39	0.50 (-0.37, 1.38) p=0.26	0.66 (-0.29, 1.60) p=0.17	0.07 (-0.82, 0.96) p=0.88
Q3 vs. Q1	0.04 (-0.40, 0.48) p=0.86	0.27 (-0.18, 0.73) p=0.24	-0.09 (-0.54, 0.36) p=0.70	-0.15 (-0.60, 0.30) p=0.52	0.08 (-0.36, 0.52) p=0.73	0.24 (-0.61, 1.09) p=0.58	-0.01 (-0.86, 0.84) p=0.98	0.22 (-0.59, 1.03) p=0.60	0.57 (-0.31, 1.45) p=0.21	-0.04 (-0.94, 0.85) p=0.93
Q4 vs. Q1	<0.01 (-0.45, 0.45) p=0.99	0.55 (0.10, 1.00) p=0.02	0.45 (0.00, 0.90) p=0.05	0.42 (-0.03, 0.87) p=0.07	0.38 (-0.07, 0.83) p=0.10	1.36 (0.58, 2.15) p<0.01^#	0.45 (-0.37, 1.27) p=0.28	0.51 (-0.32, 1.34) p=0.23	0.71 (-0.17, 1.59) p=0.11	0.26 (-0.59, 1.12) p=0.54
CERAD W-L Trial 3 Score
Q2 vs. Q1	-0.11 (-0.54, 0.32) p=0.62	0.60 (0.18, 1.02) p=0.01^#	0.41 (-0.00, 0.82) p=0.05	0.27 (-0.15, 0.69) p=0.21	0.06 (-0.34, 0.46) p=0.76	0.32 (-0.44, 1.07) p=0.41	0.44 (-0.35, 1.23) p=0.28	0.55 (-0.26, 1.36) p=0.18	0.42 (-0.46, 1.29) p=0.35	-0.19 (-1.01, 0.63) p=0.65
Q3 vs. Q1	-0.01 (-0.42, 0.39) p=0.94	0.55 (0.13, 0.97) p=0.01	0.30 (-0.11, 0.71) p=0.16	0.13 (-0.29, 0.55) p=0.54	0.27 (-0.14, 0.68) p=0.19	0.32 (-0.50, 1.13) p=0.45	0.04 (-0.74, 0.83) p=0.92	0.22 (-0.53, 0.96) p=0.57	0.46 (-0.35, 1.28) p=0.27	0.15 (-0.68, 0.97) p=0.72
Q4 vs. Q1	0.04 (-0.38, 0.46) p=0.85	0.71 (0.30, 1.12) p<0.01^#	0.65 (0.24, 1.06) p<0.01^#	0.48 (0.07, 0.90) p=0.02	0.48 (0.07, 0.90) p=0.02	0.86 (0.11, 1.61) p=0.03	0.46 -0.30, 1.22) p=0.23	0.50 (-0.26, 1.27) p=0.20	0.42 (-0.40, 1.23) p=0.32	0.20 (-0.58, 0.99) p=0.62
CERAD W-L Delayed Recall Score
Q2 vs. Q1	-0.25 (-0.84, 0.34) p=0.41	0.83 (0.24, 1.41) p=0.01^#	0.65 (0.08, 1.23) p=0.03	0.51 (-0.07, 1.10) p=0.09	0.42 (-0.14, 0.98) p=0.14	0.37 (-0.55, 1.29) p=0.43	0.19 (-0.78, 1.17) p=0.70	0.15 (-0.85, 1.15) p=0.77	-0.19 (-1.26, 0.89) p=0.74	-0.24 (-1.25, 0.76) p=0.64
Q3 vs. Q1	-0.18 (-0.74, 0.39) p=0.54	0.60 (0.02, 1.18) p=0.04	0.35 (-0.23, 0.92) p=0.23	0.30 (-0.28, 0.89) p=0.31	0.52 (-0.05, 1.09) p=0.07	0.27 (-0.73, 1.26) p=0.60	-0.28 (-1.25, 0.68) p=0.57	-0.14 (-1.06, 0.79) p=0.77	-0.41 (-1.41, 0.60) p=0.43	0.23 (-0.78, 1.24) p=0.66
Q4 vs. Q1	-0.22 (-0.80, 0.36) p=0.45	0.91 (0.34, 1.49) p<0.01^#	0.89 (0.32, 1.46) p<0.01^#	0.77 (0.19, 1.34) p=0.01	0.61 (0.04, 1.19) p=0.04	1.03 (0.11, 1.95) p=0.03	0.24 (-0.70, 1.17) p=0.62	0.27 (-0.68, 1.22) p=0.58	-0.34 (-1.34, 0.67) p=0.51	0.07 (-0.90, 1.03) p=0.89
CERAD W-L Total Score (3 Recall trials)
Q2 vs. Q1	-0.65 (-1.79, 0.49) p=0.27	1.38 (0.26, 2.50) p=0.02	0.96 (-0.15, 2.07) p=0.09	1.07 (-0.05, 2.19) p=0.06	0.37 (-0.70, 1.44) p=0.50	0.52 (-1.44, 2.49) p=0.60	1.31 (-0.79, 3.41) p=0.22	1.86 (-0.28, 4.00) p=0.09	1.02 (-1.30, 3.35) p=0.39	0.16 (-2.02, 2.33) p=0.89
Q3 vs. Q1	-0.19 (-1.28, 0.90) p=0.73	1.32 (0.20, 2.43) p=0.02	0.33 (-0.78, 1.43) p=0.56	-0.03 (-1.15, 1.09) p=0.96	0.78 (-0.31, 1.87) p=0.16	0.65 (-1.48, 2.78) p=0.55	0.11 (-1.97, 2.19) p=0.92	0.90 (-1.07, 2.88) p=0.37	0.64 (-1.54, 2.81) p=0.57	0.35 (-1.83, 2.54) p=0.75
Q4 vs. Q1	-0.12 (-1.23, 0.99) p=0.83	1.75 (0.65, 2.86) p<0.01^#	1.56 (0.46, 2.66) p=0.01^#	1.42 (0.32, 2.53) p=0.01	1.30 (0.19, 2.42) p=0.02	2.69 (0.73, 4.65) p=0.01	0.96 (-1.05, 2.97) p=0.35	1.25 (-0.78, 3.27) p=0.23	0.88 (-1.28, 3.04) p=0.43	0.93 (-1.15, 3.01) p=0.38
Animal Fluency: Total Score
Q2 vs. Q1	-0.35 (-1.74, 1.04) p=0.62	-0.01 (-1.39, 1.37) p=0.99	-0.04 (-1.39, 1.32) p=0.96	1.08 (-0.29, 2.46) p=0.12	-0.03 (-1.34, 1.28) p=0.96	1.71 (-0.52, 3.95) p=0.14	-0.79 (-3.15, 1.58) p=0.52	-1.05 (-3.47, 1.37) p=0.40	0.52 (-2.11, 3.15) p=0.70	-2.18 (-4.58, 0.22) p=0.08
Q3 vs. Q1	-1.17 (-2.50, 0.15) p=0.08	1.02 (-0.36, 2.39) p=0.15	1.33 (-0.03, 2.69) p=0.06	1.24 (-0.13, 2.62) p=0.08	0.85 (-0.49, 2.19) p=0.21	1.18 (-1.24, 3.60) p=0.34	-1.84 (-4.19, 0.51) p=0.13	-1.17 (-3.41, 1.07) p=0.31	1.38 (-1.07, 3.84) p=0.27	-2.48 (-4.90, -0.07) p=0.05
Q4 vs. Q1	-1.06 (-2.41, 0.30) p=0.13	0.89 (-0.46, 2.25) p=0.20	0.98 (-0.37, 2.33) p=0.16	1.76 (0.40, 3.12) p=0.01	0.99 (-0.37, 2.36) p=0.15	3.05 (0.82, 5.28) p=0.01	0.23 (-2.03, 2.50) p=0.84	0.66 (-1.63, 2.96) p=0.57	0.35 (-2.09, 2.79) p=0.78	-0.11 (-2.42, 2.19) p=0.92
Digit Symbol: Score
Q2 vs. Q1	-2.71 (-6.46, 1.04) p=0.16	2.53 (-1.15, 6.20) p=0.18	5.37 (1.75, 8.99) p<0.01^#	1.75 (-1.95, 5.45) p=0.35	3.98 (0.49, 7.48) p=0.03	-1.85 (-7.70, 4.00) p=0.54	1.39 (-4.72, 7.50) p=0.66	2.42 (-3.81, 8.64) p=0.45	4.34 (-2.34, 11.01) p=0.21	-2.40 (-8.65, 3.85) p=0.45
Q3 vs. Q1	-2.30 (-5.88, 1.28) p=0.21	6.37 (2.72, 10.02) p<0.01^#	5.29 (1.66, 8.92) p<0.01^#	2.94 (-0.75, 6.63) p=0.12	6.42 (2.84, 9.99) p<0.01^#	0.36 (-5.96, 6.69) p=0.91	-0.02 (-6.09, 6.05) p=1.00	1.31 (-4.45, 7.07) p=0.66	5.96 (-0.27, 12.19) p=0.06	-2.27 (-8.57, 4.02) p=0.48
Q4 vs. Q1	-3.70 (-7.36, 0.04) p=0.05	6.37 (2.76, 9.99) p<0.01^#	7.09 (3.49, 10.69) p<0.01^#	5.64 (1.99, 9.29) p<0.01^#	5.71 (2.08, 9.34) p<0.01^#	0.87 (-4.96, 6.71) p=0.77	1.83 (-4.03, 7.68) p=0.54	3.82 (-2.08, 9.71) p=0.21	3.11 (-3.09, 9.31) p=0.33	1.60 (-4.40, 7.59) p=0.60

*Fully adjusted models are adjusted for age, income, education, and prescription guidance. ^#p value <0.007 is set as significant (Bonferroni adjustment).

DISCUSSION

Most research into cholesterol-associated cognitive impairment has focused on hypercholesterolemic patients or animal models [30 –32]. However, it is crucial to explore the association between dietary or serum cholesterol and cognitive dysfunction in the context of AD. We conducted a cross-sectional analysis using NHANES data from 2011 to 2014 to further investigate such association in older adults in the U.S. Of the participants included, 88.80% had non-hypercholesterolemia (serum cholesterol < 6.22 mmol/L). The median total intake of dietary cholesterol of our included participants was consistent with recommendations published by the U.S. for healthy diets that adults should limit their daily cholesterol intake to no more than 300 mg/day [33]. In addition, the median total score for the three individual CERAD W-L Test (with a maximum of 30 points) was 21. The Animal Fluency Test and DSST had median scores of 19 and 55 points, respectively. Overall, the average cognitive scores of our sample are compatible with normal cognitive function [34]. Our study showed that higher levels of serum TC, but not dietary cholesterol, were significantly associated with improved cognitive performance in memory, language, and executive function among individuals aged 60 years and above, highlighting a potential positive influence on AD-related cognitive decline.

Previous clinical studies reported contradictory conclusions for the effect of dietary lipids on cognitive function. Wardle et al found dietary interventions to control serum cholesterol had no adverse effects on mood (depression, anxiety, and hostility) and cognition among British population [35]. However, another retrospective study based on children’s cognition demonstrated that high saturated fat acid intake was associated with longer reaction time under task conditions and required greater amounts of cognitive flexibility. Meanwhile, an increase in saturated fat intake and dietary cholesterol was linked to higher switch costs [36]. Importantly, systematic review and meta-analysis also revealed that the relationship between dietary fat intake (total fat, saturated fatty acids, unsaturated fatty acids, cholesterol, etc.) and the risk of AD, dementia, and MCI is almost negligible [37, 38], which was consistent with the result of our cross-sectional study using Spearman correlation, Kruskal-Wallis, Chi-square, and regression analysis. Interestingly, our study found a positive association between cholesterol intake and CERAD W-L Test scores among NHB participants, which may be partly due to differences in ethnicity and patterns of dietary intake. An analysis of NHANES participants during 2007-2012 showed that NHB individuals were less likely to meet recommendations for dietary fiber and cholesterol intake after adjusting for sociodemographic indicators [39]. A similar result was found in our study that NHB individuals had relatively high cholesterol intake compared to NHW individuals, suggesting that exceeding a certain limit of cholesterol intake may cause qualitative changes that affect cognitive function. Moreover, we speculated that the lack of association between cholesterol intake and cognitive scores in NHW individuals may be related to higher baseline cognitive scores. Cognitive test score has a substantial ceiling effect, so that that smaller differences between individuals with higher scores may make it more difficult to observe changes [40]. Certainly, it is also possible that the small sample size of NHW participants in our study may have led to false positive results.

In the 1960 s, it was widely believed that dietary cholesterol regulated serum cholesterol levels and was a risk factor for cognitive impairment. However, recent studies have shown this hypothesis to be untenable. Clinical and epidemiological evidence has shown that increased consumption of dietary cholesterol has little impact on serum LDL:HDL ratios or on the incidence rate of cognitive impairment [41 –43]. Therefore, altering dietary recommendations to limit dietary cholesterol is unlikely to prevent cognitive decline. Our results also showed no association between dietary cholesterol and serum cholesterol-related indicators, except for HDL-C. Similarly, dietary fat significantly correlated with some cognitive test scores but only showed a mild correlation with dietary cholesterol. Higher levels of total-cholesterol intake may indicate better overall diet quality (e.g., rich in eggs/milk), healthy lifestyle choices, access to more nutritious food options rather than being solely responsible for high-fat diet or negative health outcomes. Furthermore, although 88.80% participants were not defined as hypercholesterolemia, over half of them admitted doctor hypercholesterolemia notification or prescription guidance. This suggests that understanding hypercholesterolemia and drug interventions may play a more critical role in regulating blood lipids than dietary intake.

However, in CERAD W-L Total Test, Animal Fluency Test, and DSST, NHB participants scored significantly lower on cognitive tests compared with NHW individuals. We believe that multiple factors contributed to this result. Firstly, racial disparities can lead to critical mutations in lipid metabolism genes. For example, a common variant of the LPL gene, rs328 (S447X), has been associated with HDL-C and triglycerides (TG). At this locus, stronger effects have been observed between European and African genetic backgrounds [44]. Additionally, disparities in education and income between ethnicity may contribute to the occurrence of cognitive differences in older adults [45]. Furthermore, lifestyle modifications are also beneficial in preventing cognitive impairment. A study found that NHB participants were more receptive to lifestyle changes, but less likely than NHW participants to receive medication prescriptions for high cholesterol [46]. All in all, these dietary patterns as well as other factors such as genetic, lifestyle and cultural differences may affect individual abilities regarding absorption or accumulation of cholesterol in blood and neural tissue. To further evaluate the observed association between race, cholesterol, and cognitive function, future studies need to include individuals from more diverse racial backgrounds.

A positive association between serum cholesterol and cognitive function was observed in both NHW and NHB as well as the overall sample. Higher serum TC (in the normal or slightly higher range) could help maintain or improve cognitive function. However, high levels of cholesterol in the blood have long been recognized as a risk factor for AD. A high-quality meta-analysis incorporating 34 relevant studies has also demonstrated a significant increase in the risk of developing AD in individuals with elevated serum TC levels [47]. Although cholesterol is involved in the formation of atherosclerosis and prevents Aβ cleaning in the nervous system, it also plays an important role in neuronal membrane formation, mediating trans-membrane signal transduction, regulating the release of neurotransmitters and maintaining the maturation and plasticity of synapses [48]. Some scholars believe that decreased serum TC may reflect ongoing cognitive impairment and nerve pathological processes which may increase the risk of cognitive impairment, dementia, and AD [49]. Some meta-analyses have also reported no evidence to support the association between serum TC in older adults and the risk of dementia or cognitive decline, even after conducting stratified analyses based on statin use or the presence of the APOE ɛ4 allele [50, 51]. Therefore, while serum cholesterol is known as a risk factor in the AD process, its utility during the early stages of cognitive impairment preceding AD may differ. Cholesterol in the human brain accounts for about 23–25% of total body cholesterol, concentrated mainly in myelin sheath and neuronal cell membranes [52]. The myelin sheath accelerates action transmission by wrapping axons and blocking interference between them. Cognitive deficits and neurodegeneration may be attributed to dysfunction of synaptic transmission caused by impaired synaptic transduction as well as defects in cholesterol biosynthesis [53]. Additionally, cholesterol helps maintain the stability of the blood-brain barrier and reduces damage to neurons from external harmful substances [20, 21]. The aforementioned evidence partially supports our research findings that there is a positive correlation between higher cholesterol levels within the normal range and improved cognitive function. However, it should be noted that the role of cholesterol in the progression of cognitive impairment to AD is complex. Further research is needed into the potential mechanisms underlying the observed relationship between cholesterol and AD. This can be achieved through high-quality randomized controlled trials and basic research in neurobiology. While a great number of pathways in which cholesterol can affect cognitive function are well-understood, there is still much to be learned. Investigating these mechanisms could provide insights into potential interventions blocking the progression of cognitive impairment to AD in the elderly population.

The cross-sectional design of our survey data was a major limitation, as it prevented us from determining whether a causal relationship existed between cholesterol and cognitive performance. Genetic factors and racial disparities strongly influence serum cholesterol and cognition, and the correlations we observed may be due to these factors, independent from the direct interaction between cholesterol and cognitive function. Additionally, sole 24-hour dietary reviews may only capture dietary intake over a short period but not fully reflect an individual’s typical eating habits or long-term diet. Moreover, while we adjusted for some confounding factors based on basic information from the NHANES database in our research, we have to acknowledge that it is hard to control for fasting or feeding status, cholesterol-lowering prescription type, or other medications that may affect lipid concentrations and cognitive function, such as psychotropic drugs, beta-blockers, and anticoagulants. Finally, omission of statistical interaction before separating by ethnic group may also be a potential limitation. Despite these limitations, our study had several advantages. We screened a large representative sample while controlling for potential confounding factors and collecting complete information on various cognitive scores. Although the strength of the associations we observed was inconsistent, even moderately effective interventions could reduce the economic and social burden associated with cognitive impairment and dementia. Furthermore, dietary interventions are relatively low-cost with minimal side effects, while serum indicators can be easily measured to predict future outcomes.

Conclusion

Our study indicates that there is no significant correlation between cholesterol intake and cognitive performance in most older adults in the U.S. population. However, diets that are high in cholesterol may act as protective measures against cognitive decline among NHB participants. In the overall sample and NHW participants, serum cholesterol levels within normal or slightly higher ranges were also positively associated with cognitive function. Our study highlighted the importance of considering racial differences in the association between cholesterol and cognitive performance.

Footnotes

ACKNOWLEDGMENTS

The authors have no acknowledgments to report.

FUNDING

This work was supported in part by the Natural Science Foundation of Nanjing University of Traditional Chinese Medicine (grant no. XZR2020072) and Nanjing Health Science and Technology Development Special Fund Project (YKK21118).

CONFLICT OF INTEREST

The authors have no conflict of interest to report.

DATA AVAILABILITY

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

The supplementary material is available in the electronic version of this article: .

References

Toepper

(2017) Dissociating normal aging from Alzheimer’s disease: A view from cognitive neuroscience. J Alzheimers Dis 57, 331–352.

Chan

, Wang

, Wu

, Liu

, Theodoratou

, Car

, Middleton

, Russ

, Deary

, Campbell

, Wang

, Rudan

(2013) Epidemiology of Alzheimer’s disease and other forms of dementia in China, 1990-2010: A systematic review and analysis. Lancet 381, 2016–2023.

Petersen

, Lopez

, Armstrong

, Getchius

TSD

, Ganguli

, Gloss

, Gronseth

, Marson

, Pringsheim

, Day

, Sager

, Stevens

, Rae-Grant

(2018) Practice guideline update summary: Mild cognitive impairment: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology 90, 126–135.

Serrano-Pozo

, Frosch

, Masliah

, Hyman

(2011) Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, a006189.

Calsolaro

, Edison

(2016) Neuroinflammation in Alzheimer’s disease: Current evidence and future directions. Alzheimers Dement 12, 719–732.

Braak

, Braak

(1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239–259.

Smith

, Blumenthal

(2016) Dietary factors and cognitive decline. J Prev Alzheimers Dis 3, 53–64.

Christensen

, Gleason

, Mares

(2020) Dietary carotenoids and cognitive function among US adults, NHANES 2011-2014. Nutr Neurosci 23, 554–562.

Ylilauri

, Voutilainen

, Lönnroos

, Mursu

, Virtanen

, Koskinen

, Salonen

, Tuomainen

, Virtanen

(2017) Association of dietary cholesterol and egg intake with the risk of incident dementia or Alzheimer disease: The Kuopio Ischaemic Heart Disease Risk Factor Study. Am J Clin Nutr 105, 476–484.

10.

Salerno-Kennedy

, Cashman

(2007) The relationship between nutrient intake and cognitive performance in people at risk of dementia. Ir J Med Sci 176, 193–198.

11.

Göritz

, Mauch

, Nägler

, Pfrieger

(2002) Role of glia-derived cholesterol in synaptogenesis: New revelations in the synapse-glia affair. J Physiol Paris 96, 257–263.

12.

Wendell

, Zonderman

, Metter

, Najjar

, Waldstein

(2009) Carotid intimal medial thickness predicts cognitive decline among adults without clinical vascular disease. Stroke 40, 3180–3185.

13.

Wright

, Festa

, Paik

, Schmiedigen

, Brown

, Yoshita

, DeCarli

, Sacco

, Stern

(2008) White matter hyperintensities and subclinical infarction: Associations with psychomotor speed and cognitive flexibility. Stroke 39, 800–805.

14.

Solomon

, Kåreholt

, Ngandu

, Winblad

, Nissinen

, Tuomilehto

, Soininen

, Kivipelto

(2007) Serum cholesterol changes after midlife and late-life cognition: Twenty-one-year follow-up study. Neurology 68, 751–756.

15.

Burns

, Duff

(2002) Cholesterol in Alzheimer’s disease and tauopathy. Ann N Y Acad Sci 977, 367–375.

16.

Grundy

, Cleeman

, Merz

, Brewer

Jr. , Clark

, Hunninghake

, Pasternak

, Smith

Jr. , Stone

(2004) Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III Guidelines. J Am Coll Cardiol 44, 720–732.

17.

Ladu

, Reardon

, Van Eldik

, Fagan

, Bu

, Holtzman

, Getz

(2000) Lipoproteins in the central nervous system. Ann N Y Acad Sci 903, 167–175.

18.

Dietschy

, Turley

(2004) Thematic review series: Brain Lipids. Cholesterol metabolism in the central nervous system during early development and in the mature animal. J Lipid Res 45, 1375–1397.

19.

Mauch

, Nägler

, Schumacher

, Göritz

, Müller

, Otto

, Pfrieger

(2001) CNS synaptogenesis promoted by glia-derived cholesterol. Science 294, 1354–1357.

20.

Wahler

, Swain

, Carson

, Bergasa

, Jones

(1993) Blood-brain barrier permeability is markedly decreased in cholestasis in the rat. Hepatology 17, 1103–1108.

21.

Joseph

, Denisova

, Villalobos-Molina

, Erat

, Strain

(1996) Oxidative stress and age-related neuronal deficits. Mol Chem Neuropathol 28, 35–40.

22.

Zhang

, Muldoon

, McKeown

(2004) Serum cholesterol concentrations are associated with visuomotor speed in men: Findings from the third National Health and Nutrition Examination Survey, 1988-1994. Am J Clin Nutr 80, 291–298.

23.

Nath

, Jeong

, Igarashi

, Proulx

, Aldwin

, Spiro

3rd (2015) Cholesterol and depressive symptoms in older men across time. Health Psychol Open 2, 2055102915592089.

24.

Centers for Disease Control and Prevention (CDC) (2013) National Health and Nutrition Examination Survey Data. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. Available at: https://www.cdc.gov/nchs/nhanes/

25.

Subar

, Kirkpatrick

, Mittl

, Zimmerman

, Thompson

, Bingley

, Willis

, Islam

, Baranowski

, McNutt

, Potischman

(2012) The Automated Self-Administered 24-hour dietary recall (ASA24): A resource for researchers, clinicians, and educators from the National Cancer Institute. J Acad Nutr Diet 112, 1134–1137.

26.

Wang

, Wang

, Hu

, Zhang

, Dong

, Zhang

(2021) Association of dietary vitamin D intake, serum 25(OH)D(3), 25(OH)D(2) with cognitive performance in the elderly. Nutrients 13, 3089.

27.

, Sun

, Zhang

(2019) Association of zinc, iron, copper, and selenium intakes with low cognitive performance in older adults: A cross-sectional study from National Health and Nutrition Examination Survey (NHANES). J Alzheimers Dis 72, 1145–1157.

28.

Dong

, Li

, Chen

, Li

, Wu

, Zhang

(2020) Association of dietary ω-3 and ω-6 fatty acids intake with cognitive performance in older adults: National Health and nutrition examination Survey (NHANES) 2011-2014. Nutr J 19, 25.

29.

Bland

, Altman

(1995) Multiple significance tests: The Bonferroni method. BMJ 310, 170.

30.

Han

, Kim

(2021) Are serum cholesterol levels associated with cognitive impairment and depression in elderly individuals without dementia?: A retrospective cohort study in South Korea. Int J Geriatr Psychiatry 36, 163–173.

31.

Husain

, Akhtar

, Abdin

, Islamuddin

, Shaharyar

, Najmi

(2018) Rosuvastatin ameliorates cognitive impairment in rats fed with high-salt and cholesterol diet via inhibiting acetylcholinesterase activity and amyloid beta peptide aggregation. Hum Exp Toxicol 37, 399–411.

32.

Abuelezz

, Hendawy

(2021) HMGB1/RAGE/TLR4 axis and glutamate as novel targets for PCSK9 inhibitor in high fat cholesterol diet induced cognitive impairment and amyloidosis. Life Sci 273, 119310.

33.

Spence

(2019) Nutrition and risk of stroke. Nutrients 11, 647.

34.

Welsh

, Butters

, Mohs

, Beekly

, Edland

, Fillenbaum

, Heyman

(1994) The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part V. A normative study of the neuropsychological battery. Neurology 44, 609–614.

35.

Wardle

, Rogers

, Judd

, Taylor

, Rapoport

, Green

, Nicholson Perry

(2000) Randomized trial of the effects of cholesterol-lowering dietary treatment on psychological function. Am J Med 108, 547–553.

36.

Khan

, Raine

, Drollette

, Scudder

, Hillman

(2015) The relation of saturated fats and dietary cholesterol to childhood cognitive flexibility. Appetite 93, 51–56.

37.

Zhu

, Chen

, Zhang

, Wu

, Zhao

(2021) Dietary fatty acids and risk for Alzheimer’s disease, dementia, and mild cognitive impairment: A prospective cohort meta-analysis. Nutrition 90, 111355.

38.

Engelhart

, Geerlings

, Ruitenberg

, Van Swieten

, Hofman

, Witteman

, Breteler

(2002) Diet and risk of dementia: Does fat matter?: The Rotterdam Study. Neurology 59, 1915–1921.

39.

Thompson

, Singleton

, Springfield

, Thorpe

Jr. , Odoms-Young

(2020) Differences in nutrient intake and diet quality between non-Hispanic Black and non-Hispanic white men in the United States. Public Health Rep 135, 334–342.

40.

Vieira

, Elkind

, Moon

, Rundek

, Boden-Albala

, Paik

, Sacco

, Wright

(2011) The metabolic syndrome and cognitive performance: The Northern Manhattan Study. Neuroepidemiology 37, 153–159.

41.

McNamara

(2014) Dietary cholesterol, heart disease risk and cognitive dissonance. Proc Nutr Soc 73, 161–166.

42.

Fernandez

, Webb

(2008) The LDL to HDL cholesterol ratio as a valuable tool to evaluate coronary heart disease risk. J Am Coll Nutr 27, 1–5.

43.

Herron

, Lofgren

, Sharman

, Volek

, Fernandez

(2004) High intake of cholesterol results in less atherogenic low-density lipoprotein particles in men and women independent of response classification. Metabolism 53, 823–830.

44.

Bentley

, Chen

, Shriner

, Doumatey

, Zhou

, Huang

, Mullikin

, Blakesley

, Hansen

, Bouffard

, Cherukuri

, Maskeri

, Young

, Adeyemo

, Rotimi

(2014) Gene-based sequencing identifies lipid-influencing variants with ethnicity-specific effects in African Americans. PLoS Genet 10, e1004190.

45.

Clouston

SAP

, Manganello

, Richards

(2017) A life course approach to health literacy: The role of gender, educational attainment and lifetime cognitive capability. Age Ageing 46, 493–499.

46.

Bacon

(2020) Racial/ethnic differences in treatment recommendations: Lifestyle changes and medication prescriptions for high cholesterol. Ethn Health 25, 273–288.

47.

Anstey

, Ashby-Mitchell

, Peters

(2017) Updating the evidence on the association between serum cholesterol and risk of late-life dementia: Review and meta-analysis. J Alzheimers Dis 56, 215–228.

48.

Silva

, Teixeira

, Remião

, Borges

(2013) Alzheimer’s disease, cholesterol, and statins: The junctions of important metabolic pathways. Angew Chem Int Ed Engl 52, 1110–1121.

49.

Benito-León

, Vega-Quiroga

, Villarejo-Galende

, Bermejo-Pareja

(2015) Hypercholesterolemia in elders is associated with slower cognitive decline: A prospective, population-based study (NEDICES). J Neurol Sci 350, 69–74.

50.

Peters

, Xu

, Antikainen

, Beckett

, Gussekloo

, Jagger

, Jukema

, Keinanen-Kiukaanniemi

, Rydén

, Skoog

, Staessen

, Thijs

, Trompet

, Tully

, Tzourio

, Anstey

(2021) Evaluation of high cholesterol and risk of dementia and cognitive decline in older adults using individual patient meta-analysis. Dement Geriatr Cogn Disord 50, 318–325.

51.

Anstey

, Lipnicki

, Low

(2008) Cholesterol as a risk factor for dementia and cognitive decline: A systematic review of prospective studies with meta-analysis. Am J Geriatr Psychiatry 16, 343–354.

52.

Björkhem

, Meaney

(2004) Brain cholesterol: Long secret life behind a barrier. Arterioscler Thromb Vasc Biol 24, 806–815.

53.

Petrov

, Kasimov

, Zefirov

(2016) Brain cholesterol metabolism and its defects: Linkage to neurodegenerative diseases and synaptic dysfunction. Acta Naturae 8, 58–73.