Blood Triglyceride and High-Density Lipoprotein Levels Are Associated with Plasma Amyloid-β Transport: A Population-Based Cross-Sectional Study

Abstract

Background:

Studies have found that blood lipids are associated with plasma amyloid-β (Aβ) levels, but the underlying mechanism is still unclear. Two Aβ transporters, soluble form of low-density lipoprotein receptor related protein-1 (sLRP1) and soluble receptor of advanced glycation end products (sRAGE), are crucial in peripheral Aβ transport.

Objective:

The aim was to investigate the effects of lipids on the relationships between plasma Aβ and transporter levels.

Methods:

This study included 1,436 adults aged 40 to 88 years old. Blood Aβ, sLRP1, sRAGE, and lipid levels were measured. Univariate and multivariate analyses were used to analyze the relationships between lipids and plasma Aβ, sLRP1, and sRAGE.

Results:

After adjusting for all possible covariates, high-density lipoprotein (HDL-c) was positively associated with plasma Aβ₄₂ and sRAGE (β= 6.158, p = 0.049; β= 121.156, p < 0.001, respectively), while triglyceride (TG) was negatively associated with plasma Aβ₄₀, Aβ₄₂, and sRAGE (β= –48.389, p = 0.017; β= –11.142, p = 0.020; β= –147.937, p = 0.003, respectively). Additionally, positive correlations were found between plasma Aβ and sRAGE in the normal TG (Aβ₄₀: β= 0.034, p = 0.005; Aβ₄₂: β= 0.010, p = 0.001) and HDL-c groups (Aβ₄₀: β= 0.023, p = 0.033; Aβ₄₂: β= 0.008, p = 0.002) but not in the high TG and low HDL-c groups.

Conclusion:

Abnormal levels of TG and HDL-c are associated with decreased Aβ and sRAGE levels. Positive correlations between plasma Aβ and sRAGE were only found in the normal TG and HDL-c groups but not in the high TG and low HDL-c groups. These results indicated that dyslipidemia contributing to plasma Aβ levels might also be involved in peripheral Aβ clearance.

Keywords

Alzheimer’s disease amyloid-β high-density lipoprotein soluble low-density lipoprotein receptor-related protein-1 soluble receptor of advanced glycation end products triglyceride

INTRODUCTION

The number of patients with dementia is increasing rapidly because of societal aging. Alzheimer’s disease (AD) is the primary cause of cognitive impairment, accounting for up to 75%of all dementia cases and leading to a costly public health burden in all countries worldwide [1]. The main pathological hallmarks of AD are amyloid-β (Aβ) plaque deposition and neurofibrillary tangles composed of tau phosphorylation [2]. Aβ₄₀ and Aβ₄₂, the most important components of senile plaques, are generated from the cleavage of amyloid-β protein precursor (AβPP) by sequential β- and γ-secretases [3]. Aβ can be cleared from the central nervous system (CNS) to the blood via transporters through the blood-brain barrier (BBB), while reverse transport of peripheral Aβ across the BBB into the brain can occur as well [4, 5].

There is a complex dynamic equilibrium between the amyloid burden in the brain and plasma Aβ [6]. A study found that plasma Aβ levels are closely associated with brain Aβ accumulation [7], and the clearance of peripheral Aβ can reflect Aβ plaque deposition in the brain [8]. When transferred into the periphery, 70–90%of plasma Aβ is bound by the soluble form of low-density lipoprotein receptor related protein-1 (sLRP1) and is delivered to the organs responsible for clearance (liver and kidney) for systemic elimination [9, 10]. Moreover, the soluble receptor of advanced glycation end products (sRAGE) can bind plasma Aβ as a decoy receptor and can prevent it from binding to the membrane RAGE [11, 12].

It has been suggested that dyslipidemia is closely correlated with the risk of AD and plays an important role in the metabolism of brain Aβ [13 –16]. Previous studies have found that dyslipidemia is associated with plasma Aβ levels [17 –19]. However, the mechanism by which blood lipids affect plasma Aβ is still unclear. Considering that sLRP1 and sRAGE are crucial in peripheral Aβ transport, we performed a cross-sectional study to investigate the relationships between blood lipids and plasma Aβ transport in a community population. We hypothesized that blood lipids may also contribute to plasma Aβ levels through the effect on sLRP1 and sRAGE levels and subsequently contribute to the pathogenesis of AD not only from the central but also from the peripheral perspective.

MATERIALS AND METHODS

Participants

The present study was a cross-sectional study, and a random cluster sampling method was used to select the samples. A total of 1,903 subjects aged 40–88 years old were recruited from Qubao village, a suburb of Xi’an in northwest China. Study inclusion criteria were as follows: participants were local residents who had lived in Qubao for more than three years and signed the informed consent form to participate in this study. Study exclusion criteria were as follows: participants who had no blood samples drawn; participants who had hemolytic blood specimens; participants who had long-term use of lipid-lowering drugs; participants with blood measurements of Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE below the lower limit of detection; and participants who refused to get blood samples taken. After referring to all exclusion criteria, 1,436 participants were finally included in this study (Fig. 1). The studies in this report were conducted under approval by the Medical Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University. All participants were required to sign a written informed consent form before participating in the study.

Fig. 1

Flow chart. Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products.

Questionnaire survey and demographic information

Standardized questionnaires were used to collect participants’ sociodemographic information and medical histories. Demographic information included the following: sex, age, years of education, smoking (yes or no), drinking (yes or no), and physical exercise level. Physical examinations included height, weight, systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse rate. Medical histories provided information regarding hypertension (yes or no), diabetes (yes or no), heart disease (yes or no), dyslipidemia including the use of lipid-lowering drugs (yes or no), cerebrovascular disease including stroke and transient ischemic attack (yes or no), and other physical diseases. Body mass index (BMI) was defined as a person’s weight in kilograms divided by the square of his or her height in meters (kg/m²). Mean arterial pressure (MAP) was calculated as 1/3 SBP and 2/3 DBP. The definitions of hypertension and diabetes were consistent with those described in our previous article [20].

Blood biomarkers and APOE genotyping

Ten milliliters of fasting venous blood was collected from participants between 8 am and 10 am and placed in a red-cap nonanticoagulant tube and a purple-cap anticoagulant tube. Within 2 h of collection, the anticoagulant blood of the purple-cap was centrifuged at a centrifugation force of 1,500×g for 10 min. The centrifuged blood samples were separated into plasma and blood cells and stored in a freezer at –80%for future use. Red-cap nonanticoagulant blood samples were sent to the biochemistry laboratory of the First Affiliated Hospital of Xi’an Jiaotong University. Blood lipids, including high-density lipoprotein (HDL-c), low-density lipoprotein (LDL-c), total cholesterol (TC), triglyceride (TG), and fasting blood glucose (FBG), were determined by using a biochemical automatic analyzer (C501, Roche, Sweden). According to the 2016 Chinese guidelines for the management of dyslipidemia in adults [21], a low blood HDL level (< 1.04 mmol/L, n = 101) and/or high blood LDL level (≥3.37 mmol/L, n = 327) and/or high blood TC level (≥5.18 mmol/L, n = 549) and/or high blood TG level (≥1.70 mmol/L, n = 429) was considered as dyslipidemia. Plasma Aβ₄₀ and Aβ₄₂ levels were detected using commercial competitive inhibition enzyme-linked immunosorbent assay (ELISA) kits SEB010Hu and SEA645Hu (Cloud-Clone Corp., Wuhan, China), and the levels of sLRP1 and sRAGE were measured by sandwich ELISA kits CEA864Hu and CEA946Hu (Cloud-Clone Corp., Wuhan, China), respectively. Briefly, the specific amino acid sequences on the immunogenic parts were Asp672∼Val711 for Aβ₄₀, Asp672∼Ala713 for Aβ₄₂, Lys25~Lys189 for sLRP1, and Ala23~Ser317 for sRAGE. The intra–assay coefficients of variation (CV) were < 10%, and the inter–assay CV was < 12%. The lower limits of quantitation were 5.23 pg/mL for Aβ₄₀, 1.39 pg/mL for Aβ₄₂, 1.29 ng/mL for sLRP1 and 1.24 pg/mL for sRAGE. The detection method of the APOE genotype has been described in our previous article [22].

Statistical analyses

The differences in normally distributed data (age, years of education, MAP, BMI, pulse rate, TC, HDL-c, LDL-c, Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE) between groups were analyzed using unpaired Student’s t-tests and expressed as the mean (standard deviation), with post hoc comparison using the false discovery rate (FDR) test for multiple comparisons. FBG and TG presented skewed distributions and were analyzed using the Mann–Whitney U test and expressed as medians (quartiles). Chi-square tests and percentages were used for categorical data. Participants were first classified into a hyperlipidemia group and a normal lipid group, and the differences in baseline information were compared. Plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE levels were also compared between the dyslipidemia group and the normal lipid group.

Then, simple Pearson’s correlation analysis was used to analyze the relationships between blood lipid parameters and plasma Aβ, sLRP1, and sRAGE levels in all participants. For multivariate analysis, multiple linear regression analysis (MLR) was used to analyze the relationship between plasma Aβ₄₀, Aβ₄₂, sLRP1, sRAGE, and lipid levels. In the MLR models, plasma concentrations of Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE were the dependent variables. Blood lipid levels and other covariates were independent variables. Considering that other confounding variables might have an impact on plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE levels, we compared plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE levels in different subgroups of covariates to discover possible confounders.

Simple Pearson’s correlation and MLR analyses were also used to investigate the correlations between plasma Aβ and sLRP1 and sRAGE levels in the total population of participants. The confounding factors, including sex, age, years of education, smoking, drinking, physical activity, medical history, FBG, MAP, and pulse rate, were adjusted. Finally, to explore the potential effects of blood lipids on the relationships between plasma Aβ and Aβ transporters, we performed the above analysis in the abnormal lipid group and in the normal lipid group. GraphPad Prism version 5.0 (GraphPad Software, Inc., San Diego) and SPSS version 18.0 (SPSS Inc., IBM, Chicago) were used to draw graphs and perform statistical analysis, respectively. All statistical tests were two-sided, and a p value of < 0.05 was considered as statistically significant.

RESULTS

Demographic and clinical characteristics of samples

Among the total 1,436 participants ranging in age from 40 to 88 years old (mean 59.2±9.7 years), 835 (58%) were women. A total of 387 subjects refused to be tested for the APOE genotype, and 1,049 participants had APOE genotyping, among which 144 were APOE ɛ4 carriers and 905 were noncarriers. As shown in Table 1, 60%(867) met the criteria of dyslipidemia compared with 40%(569) having normal blood lipids. Participants with dyslipidemia had higher levels of FBG, MAP, BMI, TC, TG, and LDL-c (for all: p < 0.001) than those with normal blood lipids. The results also showed that there was a higher ratio of females (p = 0.001), physical inactivity (p = 0.020), APOE ɛ4 carriers (p = 0.004), and presence of hypertension and diabetes (for all: p < 0.001) in this population as well as a lower incidence of smoking (p = 0.005).

Table 1

Demographic and clinical characteristics of the study participants

	Total (n = 1,436)	Normal blood lipids (n = 569)	Dyslipidemia (n = 867)	p
Age, y (mean±SD)	59.2±9.7	58.6±9.8	59.6±9.6	0.062
Gender, female, n (%)	835 (58.1)	300 (52.7)	535 (61.7)	0.001
Education, y (mean±SD)	6.3±3.5	6.4±3.4	6.1±3.6	0.103
Drinking, n (%)	201 (14.0)	81 (14.2)	120 (13.8)	0.876
Smoking, n (%)	483 (33.6)	216 (38.0)	267 (30.8)	0.005
Lack of activity, n (%)	177 (12.3)	56 (9.8)	121 (14.0)	0.020
Cardiovascular disease, n (%)	130 (9.1)	46 (8.1)	84 (9.7)	0.300
Hypertension, n (%)	714 (49.7)	247 (43.4)	467 (53.9)	< 0.001
Diabetes mellitus, n (%)	192 (13.4)	42 (7.4)	150 (17.3)	< 0.001
Cerebrovascular disease, n (%)	165 (11.5)	63 (11.1)	102 (11.8)	0.687
MAP, mmHg (mean±SD)	99.7±12.4	98±11	101±13	< 0.001
Pulse rate, (mean±SD)	71±9	70.9±9.5	71.6±9.5	0.153
FBG, mmol/l, median(quartile)	5.01 (4.63, 5.53)	4.95 (4.57, 5.38)	5.04 (4.68, 5.67)	< 0.001
BMI, kg/m² (mean±SD)	25.30±3.22	24.60±2.99	25.75±3.28	< 0.001
TC, mmol/L (mean±SD)	4.99±0.99	4.38±0.54	5.40±1.01	< 0.001
TG, median(quartile)	1.33 (0.99, 1.87)	1.06 (0.82, 1.33)	1.67 (1.22, 2.33)	< 0.001
LDL-c, mmol/L (mean±SD)	2.86±0.74	2.43±0.49	3.15±0.73	< 0.001
HDL-c, mmol/L (mean±SD)	1.40±0.29	1.41±0.24	1.40±0.32	0.834
APOE ɛ4 carriers, n (%)	144 (13.7)	39 (9.8)	105 (16.1)	0.004

Among the total 1,436 participants, 387 subjects refused to be tested for the APOE genotype and finally 1,049 participants had APOE genotyping, among which 144 participants were APOE ɛ4 carriers and 905 were non-carriers. Unpaired Student’s t-test and mean±SD were used to compare the difference of the approximately normally distributed continuous variables between normal blood lipids and dyslipidemia, Mann-Whitney U test and median (quartile) were used for the skew distributional data, and χ² and percentage were used for categorical variables. MAP, mean arterial pressure; FBG, fasting blood glucose; BMI, body mass index; APOE, apolipoprotein E; TC, total cholesterol; TG, triglyceride; LDL-c, low-density lipoprotein; HDL-c, high-density lipoprotein; SD, standard deviation.

Plasma Aβ and Aβ transporter levels in the dyslipidemia group and normal lipid group

To investigate the relationships between blood lipid parameters and plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE, we first divided the total population into groups according to blood lipid levels. In the total population of participants, the mean values of plasma Aβ₄₀, Aβ₄₂, sRAGE, and sLRP1 were 245.86 pg/ml, 50.31 pg/ml, 811.96 pg/ml, and 5278.43 ng/ml, respectively. As shown in Table 2, compared with the total normal lipid group, the concentrations of plasma sRAGE were lower in the total dyslipidemia group (p = 0.004). In the high TG group, the levels of plasma Aβ₄₀, Aβ₄₂, and sRAGE were lower and sLRP1 levels were higher than those in the normal TG group (p = 0.022; p = 0.022; p < 0.001; p = 0.026, respectively). After multiple comparisons across the four outcomes (Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE), the results mentioned above were still significant. The concentrations of plasma Aβ₄₀ and sRAGE were lower in the low HDL-c group than in the normal HDL-c group (p = 0.041 and p = 0.029, respectively). Although the significant difference disappeared after multiple comparisons, there was still a tendency toward a correlation between plasma Aβ₄₀/sRAGE levels and HDL-c (p = 0.082; p = 0.08, respectively). The levels of plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE were not significantly different between the high TC group and the normal TC group or between the high LDL-c group and the normal LDL-c group (Table 2).

Table 2

Plasma Aβ and Aβ transporters in the normal blood lipids and dyslipidemia groups

Subgroups	Aβ₄₀, pg/ml mean (SD)	p	Adj. p	Cohen’s d	Aβ₄₂, pg/ml mean (SD)	p	Adj. p	Cohen’s d	sLRP1, ng/ml	p	Adj. p	Cohen’s d	sRAGE, pg/ml mean (SD)	p	Adj. p	Cohen’s d
Dyslipidemia (n = 867)	241.73 (113.81)	0.129	0.195	0.080	49.01 (26.69)	0.410	0.410	0.044	5,388.22 (4,563.37)	0.146	0.195	–0.077	761.82 (323.55)	0.004	0.016	0.153
Total normal (n = 569)	251.02 (112.45)				50.21 (27.00)				5,016.57 (4,982.23)				811.96 (322.08)
High TC (n = 549)	247.14 (115.78)	0.647	0.996	–0.024	49.48 (26.87)	0.996	0.996	< 0.001	5,337.33 (4,374.16)	0.544	0.996	–0.032	765.82 (323.79)	0.948	0.996	–0.003
Normal TC (n = 887)	244.31 (111.83)				49.49 (26.79)				5,181.21 (4,947.68)				764.67 (321.92)
High TG (n = 429)	234.42 (107.98)	0.022	0.026	0.126	46.99 (26.59)	0.022	0.026	0.122	5,669.17 (4,636.95)	0.026	0.026	–0.118	699.06 (313.97)	< 0.001	< 0.001	0.268
Normal TG (n = 1,007)	250.05 (115.26)				50.55 (26.85)				5,059.14 (4,767.52)				792.96 (322.18)
High LDL-c (n = 327)	248.57 (114.15)	0.565	0.840	–0.030	50.11 (26.08)	0.630	0.840	–0.025	5,273.65 (4,633.73)	0.887	0.887	–0.008	745.56 (319.10)	0.213	0.840	0.066
Normal LDL-c (n = 1,109)	244.46 (113.12)				49.30 (27.03)				5,231.23 (4,767.33)				770.87 (323.44)
Low HDL-c (n = 101)	223.23 (109.58)	0.041	0.082	0.108	47.33 (24.71)	0.401	0.515	0.044	4,945.06 (4,522.11)	0.515	0.515	0.034	696.96 (345.16)	0.029	0.082	0.115
Normal HDL-c (n = 1,335)	247.08 (113.47)				49.65 (26.97)				5,263.35 (4,752.28)				770.17 (320.34)

Unpaired Student’s t-test and mean (SD) were used to compare the difference of plasma Aβ and Aβ transporter levels in the subgroups of dyslipidemia versus total normal lipids levels; high TC (≥5.18 mmol/L) versus normal TC; high TG (≥1.70 mmol/L) versus normal TG; high LDL-c (≥3.37 mmol/L) versus normal LDL-c; low HDL-c (< 1.04 mmol/L) versus normal HDL-c, with post hoc comparison using the false discovery rate (FDR) test for multiple comparisons. Cohen’s d is used to describe the effect size of differences between variables. Dyslipidemia: a low blood HDL level and/or high blood LDL level and/or high blood TC level and/or high blood TG level was considered dyslipidemia. Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products; SD, standard deviation.

The effects of the covariates on plasma Aβ and Aβ transporter levels

To explore potential confounders for subsequent multivariate analysis, the differences in plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE levels among covariates were compared. As depicted in Table 3, the data show that participants who smoked had a higher level of plasma Aβ₄₂ than those who did not (p = 0.022), the concentrations of plasma Aβ₄₀ and Aβ₄₂ were lower in APOE ɛ4 carriers (p = 0.046; p = 0.027, respectively), and other covariates had no effect on plasma Aβ₄₀ and Aβ₄₂ levels. Age, sex, education, presence of diabetes mellitus, smoking, and drinking had significant effects on plasma sLRP1 and sRAGE levels; CHD and intensity of physical activity had effects on plasma sRAGE levels, but not sLRP1. Hypertension had a significant effect on plasma sLRP1 levels, but not sRAGE. Cerebrovascular disease and APOE ɛ4 had no effect on plasma sLRP1 and sRAGE levels.

Table 3

Plasma Aβ and Aβ transporter levels in the subgroups of the covariates

	Subgroups	Aβ₄₀, ng/ml mean (SD)	Aβ ₄₂, ng/ml mean (SD)	sLRP1, pg/ml mean (SD)	sRAGE, ng/ml mean (SD)
Age, y	< 60 (n = 785)	245.54 (114.45)	49.32 (26.43)	4816.22 (4509.97)^**	683.71 (302.15)^**
	≥60 (n = 651)	245.22 (112.04)	49.69 (27.29)	5752.50 (4949.20)	862.90 (319.25)
Gender	Male (n = 601)	245.59 (116.84)	51.08 (27.29)	4590.45 (4720.70)^**	801.84(313.89)^**
	Female (n = 835)	245.25 (110.81)	48.34 (26.42)	5708.09 (4693.76)	738.66 (326.23)
Formal education	Illiteracy (n = 194)	246.36 (108.64)	49.73 (26.86)	6056.66 (4712.93)^**	877.26 (341.55)^**
	Primary school (n = 430)	249.60 (111.30)	50.34 (27.55)	5775.21 (5152.65)	812.96 (318.52)
	High school or above (n = 812)	242.95 (115.52)	48.98 (26.43)	4762.71 (4458.87)	713.01 (309.70)
Hypertension	Yes (n = 714)	241.50 (109.65)	49.00 (26.60)	5536.00 (4815.80)^*	766.86(324.48)
	No (n = 722)	249.27 (116.82)	49.96 (27.03)	4950.32 (4640.31)	763.38 (320.79)
DM	Yes (n = 192)	244.20 (114.19)	51.68 (27.10)	6084.15 (4935.87)^*	713.15 (333.09)^*
	No (n = 1244)	245.58 (113.24)	49.15 (26.76)	5112.02 (4692.97)	773.15 (320.24)
CHD	Yes (n = 130)	239.67 (117.66)	46.53 (27.63)	5876.07 (4897.82)	863.06 (327.22)^**
	No (n = 1306)	245.97 (112.92)	49.78 (26.72)	5177.54 (4716.37)	755.34 (320.54)
Stroke or TIA	Yes (n = 165)	236.59 (113.70)	46.36 (24.53)	5908.24 (4912.26)	802.66 (309.92)
	No (n = 1271)	246.53 (113.27)	49.89 (27.08)	5154.67 (4707.39)	760.23 (323.92)
Lack of activity	Yes (n = 177)	239.24 (110.40)	48.09 (26.86)	5520.31 (4553.17)	868.11 (375.88)^**
	No (n = 1259)	246.26 (113.75)	49.68 (26.81)	5201.54 (4761.12)	750.60 (311.72)
Smoking	Yes (n = 483)	248.29 (117.94)	51.77 (28.01)^*	4654.83 (4797.16)^*	815.17 (318.65)^**
	No (n = 953)	243.94 (110.97)	48.33 (26.13)	5537.03 (4678.88)	739.74 (321.67)
Drinking	Yes (n = 201)	241.87 (116.97)	49.72 (25.37)	4454.30 (4637.82)^*	814.25 (327.40)^*
	No (n = 1235)	245.97 (112.76)	49.45 (27.05)	5369.25 (4740.85)	757.14 (321.15)
APOE ɛ4 carriers	Yes (n = 144)	228.11 (112.26)^*	45.67 (24.46)^*	5085.10 (4338.04)	795.76 (317.49)
	No (n = 905)	248.69 (115.28)	51.05 (27.45)	5309.29 (4866.83)	778.45 (324.86)

Among the total 1,436 participants, 387 subjects refused to be tested for the APOE genotype and finally 1,049 participants had APOE genotyping. Unpaired Student’s t-test and mean (SD) were used for plasma Aβ to compare the difference between the subgroups of the covariates. CHD, coronary heart disease; DM, diabetes mellitus; TIA, transient ischemic attack; SD, standard deviation; Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products; APOE: apolipoprotein E; ^*p < 0.05, ^**p < 0.001.

Relationships between blood lipid parameters and plasma Aβ and Aβ transporter levels in all participants

To further investigate the connections between blood lipid parameters and plasma Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE levels, correlation and MLR analyses were used. Simple Pearson’s correlation analysis showed that among the 1436 participants, blood log-transformed TG was negatively correlated with plasma Aβ₄₀, Aβ₄₂, and sRAGE levels (r = –0.111, p < 0.001; r = –0.094, p < 0.001; r = –0.194, p < 0.001, respectively) and positively correlated with plasma sLRP1 levels (r = 0.063, p = 0.017). Blood concentrations of HDL-c were positively correlated with plasma Aβ₄₀, sLRP1, and sRAGE levels (r = 0.079, p = 0.003; r = 0.055, p = 0.037; r = 0.171, p < 0.001, respectively). Blood concentrations of LDL-c were negatively correlated with plasma sRAGE levels (r = –0.063, p = 0.017). No significant association was found between blood concentrations of TC and plasma Aβ₄₀/Aβ₄₂/sLRP1/sRAGE levels (Table 4).

Table 4

Simple Pearson’s correlation analysis of the relationships between serum lipid parameters and plasma Aβ, sLRP1, sRAGE levels in total participants

Variables	Aβ₄₀ levels		Aβ₄₂ levels		sLRP1 levels		sRAGE levels
	r	p	r	p	r	p	r	p
TC levels	0.007	0.805	–0.007	0.793	0.036	0.173	–0.030	0.257
Log-TG levels	–0.111	< 0.001	–0.094	< 0.001	0.063	0.017	–0.194	< 0.001
LDL-c levels	0.014	0.608	0.012	0.648	0.014	0.602	–0.063	0.017
HDL-c levels	0.079	0.003	0.048	0.072	0.055	0.037	0.171	< 0.001

Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products; TC, total cholesterol; Log-TG, log-transformed triglyceride; LDL-c, low-density lipoprotein; HDL-c, high-density lipoprotein.

To eliminate the influence of the covariates, MLR analysis was performed in the 1,046 participants with APOE genotyping. After adjusting for age, gender, education years, smoking history, drinking history, intensity of physical activity, APOE genotype, BMI, log-transformed FBG, MAP, pulse rate, history of transient ischemic attack or stroke, dyslipidemia, and heart disease, significant negative associations were found between blood log-transformed TG and plasma Aβ₄₀, Aβ₄₂, and sRAGE levels (β= –48.389, p = 0.017; β= –11.142, p = 0.020; β= –147.937, p = 0.003, respectively), while blood concentrations of HDL-c were positively associated with plasma Aβ₄₂ and sRAGE levels (β= 6.158, p = 0.049; β= 121.156, p < 0.001, respectively). However, the positive connections between TG and plasma sLRP1 levels, as well as between HDL-c and plasma Aβ₄₀/sLRP1 levels found in the simple Pearson’s correlation, disappeared in the MLR models. In addition, the negative connection between LDL-c and plasma sRAGE levels found in the simple Pearson’s correlation also disappeared in Model 2 of the MLR analysis. No significant association was found between blood concentrations of TC/LDL-c and plasma Aβ₄₀/Aβ₄₂/sLRP1/sRAGE levels (Table 5).

Table 5

Multivariate analysis of the relationships between serum lipid parameters and plasma Aβ, sLRP1, sRAGE levels

Model	Variables	Aβ₄₀ levels			Aβ₄₂ levels			sLRP1 levels			sRAGE levels
		β	p	$f_{X}^{2}$	β	p	$f_{X}^{2}$	β	p	$f_{X}^{2}$	β	p	$f_{X}^{2}$
0	TC levels	–1.824	0.608	< 0.001	–0.283	0.735	< 0.001	140.677	0.341	0.001	–7.432	0.456	0.001
1		–1.864	0.608	< 0.001	0.008	0.992	< 0.001	–31.826	0.831	< 0.001	–12.385	0.199	0.002
2		1.462	0.726	< 0.001	0.146	0.882	< 0.001	–127.922	0.451	0.001	–0.590	0.954	< 0.001
0	Log-TG levels	–54.804	0.001	0.010	–9.809	0.013	0.006	1,794.441	0.010	0.006	–266.235	< 0.001	0.031
1		–55.949	0.001	0.010	–9.054	0.024	0.005	1,248.522	0.075	0.003	–272.352	< 0.001	0.036
2		–48.389	0.017	0.005	–11.142	0.020	0.005	409.188	0.622	< 0.001	–147.937	0.003	0.010
0	LDL-c levels	–0.504	0.916	< 0.001	–0.317	0.779	< 0.001	9.290	0.047	< 0.001	–29.229	0.030	0.005
1		–0.416	0.931	< 0.001	–0.182	0.872	< 0.001	–73.721	0.708	< 0.001	–33.641	0.008	0.007
2		5.587	0.308	0.001	–0.051	0.969	< 0.001	–263.854	0.237	0.002	–14.321	0.288	0.001
0	HDL-c levels	21.095	0.084	0.003	3.851	0.180	0.002	772.954	0.128	0.002	201.219	< 0.001	0.033
1		23.085	0.072	0.003	5.840	0.053	0.003	–58.167	0.912	< 0.001	210.554	< 0.001	0.037
2		16.935	0.202	0.001	6.158	0.049	0.004	543.667	0.315	0.001	121.156	< 0.001	0.010

Multiple linear regression was used in the 1046 participants with APOE genotyping. Aβ₄₀, Aβ₄₂, sLRP1, and sRAGE level were dependent variable. Model 0: Univariate linear regression model with only the variable of interest (TC or Log-TG or LDL-c or HDL-c). Model 1: adjusted for age, gender, and education years. Model 2: adjusted for age, gender, education years, smoking, drinking, intensity of physical activity, APOE genotype, body mass index, log-transformed fasting blood glucose, mean arterial pressure, pulse rate, transient ischemic attack or stroke, dyslipidemia, and heart disease. $f_{X}^{2}$ was used to describe the effect size in MLR analysis. $f_{X}^{2}$ = ( $R_{XA}^{2}$ - $R_{A}^{2}$ )/(1 - $R_{XA}^{2}$ ). $R_{XA}^{2}$ is the coefficient of determination for the model with both the variable of interest and the covariates to be adjusted. $R_{A}^{2}$ is the coefficient of determination for the model with only the covariates to be adjusted (without the variable of interest). Effect sizes of $f_{X}^{2}$ = 0.1², 0.25², and 0.4² are by convention considered as small, medium, and large. APOE, apolipoprotein E; Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products; TC, total cholesterol; Log-TG, log-transformed triglyceride; LDL-c, low-density lipoprotein; HDL-c, high-density lipoprotein.

Relationships between the plasma levels of Aβ and Aβ transporters in all participants

The above MLR analysis found that blood TG and HDL-c were related to both Aβ and Aβ transporter levels. Therefore, the effects of blood TG and HDL-c on the relationships between Aβ and Aβ transporters were explored. First, simple Pearson’s correlation and MLR analyses were used to explore whether there were correlations between plasma Aβ and sLRP1 or sRAGE in the 1436 participants. Simple Pearson’s correlation analysis showed that plasma Aβ₄₂ was positively correlated with sRAGE but not sLRP1 (r = 0.061, p = 0.022; r = 0.004, p = 0.872, respectively), and no correlation was found between plasma Aβ₄₀ and sLRP1 or sRAGE (Fig. 2). After adjusting for all possible confounding factors, plasma Aβ₄₂ was still positively associated with sRAGE but not sLRP1 (β= 0.007, p = 0.004; β< 0.001, p = 0.641, respectively). A positive association was also found between Aβ₄₀ and sRAGE but not sLRP1 (β= 0.023, p = 0.025; β< 0.001, p = 0.561, respectively) (Table 6).

Fig. 2

Correlations between plasma Aβ and Aβ transporters in total participants. Simple Pearson’s correlations of plasma sLRP1 with Aβ₄₀ and Aβ₄₂ levels (A, B); plasma sRAGE with Aβ₄₀ and Aβ₄₂ levels (C, D). Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products.

Table 6

The relationships between plasma levels of Aβ and sLRP1, sRAGE in total participants

Model	Variables	Aβ₄₀ levels			Aβ₄₂ levels
		β	p	$f_{X}^{2}$	β	p	$f_{X}^{2}$
0	sLRP1 levels	< 0.001	0.472	< 0.001	< 0.001	0.872	< 0.001
1		< 0.001	0.572	< 0.001	< 0.001	0.623	< 0.001
2		< 0.001	0.561	< 0.001	< 0.001	0.641	< 0.001
0	sRAGE levels	0.016	0.086	0.002	0.005	0.022	0.004
1		0.021	0.031	0.003	0.006	0.014	0.004
2		0.023	0.025	0.004	0.007	0.004	0.005

Multiple linear regression was used in the 1436 participants. Aβ₄₀, Aβ₄₂ were dependent variable. Model 0: Univariate linear regression model with only the variable of interest (sLRP1 or sRAGE). Model 1: adjusted for age, gender, and education years. Model 2: adjusted for age, gender, education years, smoking, drinking, intensity of physical activity, log-transformed fasting blood glucose, mean arterial pressure, pulse rate, transient ischemic attack or stroke, and heart disease. $f_{X}^{2}$ was used to describe the effect size in MLR analysis. $f_{X}^{2}$ = ( $R_{XA}^{2}$ - $R_{A}^{2}$ )/(1 - $R_{XA}^{2}$ ). $R_{XA}^{2}$ is the coefficient of determination for the model with both the variable of interest and the covariates to be adjusted. $R_{A}^{2}$ is the coefficient of determination for the model with only the covariates to be adjusted (without the variable of interest). Effect sizes of $f_{X}^{2}$ = 0.1², 0.25², and 0.4² are by convention considered as small, medium, and large. Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products.

The effects of blood lipid levels on the relationships between plasma levels of Aβ and sRAGE

As mentioned above, plasma Aβ was correlated with sRAGE but not sLRP1 (Table 6). Moreover, MLR analysis found that blood TG and HDL-c were related to both Aβ and sRAGE levels but not sLRP1 levels (Table 5). Therefore, only the effects of blood TG and HDL-c on the relationships between Aβ and sRAGE were explored, and no relationships were found between Aβ and sLRP1. As shown in Table 7, stratified MLR analyses were performed according to blood TG and HDL-c status. After adjusting for all possible confounding factors mentioned above, MLR models indicated that in the normal TG and HDL-c groups, plasma sRAGE was positively associated with Aβ₄₂ (β_TG = 0.010, p = 0.001; β_HDLhboxc = 0.008, p = 0.002) and Aβ₄₀ levels (β_TG = 0.034, p = 0.005; β_HDLhboxc = 0.023, p = 0.033), and this correlation disappeared in the high TG and low HDL-c groups.

Table 7

The relationships between plasma levels of Aβ and sRAGE according to blood lipids levels

Variables	Subgroups	Model	Aβ₄₀ levels			Aβ₄₂ levels
			β	p	$f_{X}^{2}$	β	p	$f_{X}^{2}$
High TG	Yes	0	–0.016	0.354	0.002	–0.002	0.567	0.001
		1	–0.015	0.421	0.001	–0.001	0.841	< 0.001
		2	–0.019	0.321	0.002	< 0.001	0.934	0.001
	No	0	0.025	0.028	0.005	0.007	0.006	0.008
		1	0.031	0.010	0.007	0.008	0.006	0.007
		2	0.034	0.005	0.008	0.010	0.001	0.01
Low HDL-c	Yes	0	–0.013	0.686	0.002	–0.005	0.528	0.004
		1	–0.009	0.793	0.001	–0.004	0.589	0.017
		2	0.011	0.787	0.001	–0.010	0.343	0.004
	No	0	0.017	0.072	0.002	0.006	0.012	0.005
		1	0.023	0.027	0.004	0.007	0.008	0.005
		2	0.023	0.033	0.003	0.008	0.002	0.006

Multiple linear regression was used. Aβ₄₀, Aβ₄₂ were dependent variable. Model 0: univariate linear regression model with only the variable of interest (sRAGE). Model 1: adjusted for age, gender, and education years. Model 2: adjusted for age, gender, education years, smoking, drinking, intensity of physical activity, log-transformed fasting blood glucose, mean arterial pressure, pulse rate, transient ischemic attack or stroke, and heart disease. $f_{X}^{2}$ was used to describe the effect size in MLR analysis. $f_{X}^{2}$ = ( $R_{XA}^{2}$ - $R_{A}^{2}$ )/(1 - $R_{XA}^{2}$ ). $R_{XA}^{2}$ is the coefficient of determination for the model with both the variable of interest and the covariates to be adjusted. $R_{A}^{2}$ is the coefficient of determination for the model with only the covariates to be adjusted (without the variable of interest). Effect sizes of $f_{X}^{2}$ = 0.1², 0.25², and 0.4² are by convention considered as small, medium, and large. Aβ, amyloid-β; sLRP1, soluble low-density lipoprotein receptor-related protein-1; sRAGE, soluble receptor for advanced glycation end products; TG, triglyceride; HDL-c, high-density lipoprotein.

DISCUSSION

In this cross-sectional study, abnormal levels of blood TG and HDL-c were associated with decreased plasma Aβ and sRAGE levels in participants not using lipid-lowering drugs. Moreover, positive correlations between plasma Aβ₄₀, Aβ₄₂, and sRAGE were only found in the normal HDL-c and TG groups, but not in the high TG and low HDL-c groups, and there was no correlation between plasma Aβ₄₀, Aβ₄₂, and sLRP1 levels.

Previous studies have proposed a strong connection between blood cholesterol levels and the metabolism of Aβ in the brain. One study found a rapid aggregation of Aβ in the brain in rabbits that were fed a 2%cholesterol-enriched diet [23]. Another study also demonstrated that intracellular cholesterol seemed to increase Aβ levels in the brains of transgenic mice overexpressing AβPP [13]. The treatment of cultured cells with statins resulted in reduced Aβ production [15]. There was also evidence showing that cholesterol-lowering agent treatment can reduce the prevalence of AD [14] and decrease Aβ accumulation in the CNS [24]. The mechanisms of hyperlipidemia affecting Aβ metabolism have been confirmed as follows: increased cholesterol in the membrane can promote the binding of AβPPs to lipid rafts where amyloidogenic processing of AβPP largely occurs. Intracellular cholesterol seems to increase the activities of BACE1 and γ-secretase along with the generation of Aβ. The increased cholesterol levels can promote the transformation of soluble Aβ monomer to a toxic β-sheet-rich structure.

However, it should be pointed out that the effects of hyperlipidemia on brain Aβ metabolism may involve not just the changes in Aβ levels in the CNS. Considering the dynamic equilibrium between the central and peripheral circulatory systems, the correlation between blood lipids and plasma Aβ may provide additional evidence that hyperlipidemia is involved in the etiology of AD. Using cross-sectional analysis, Abdullah et al. [25] reported that blood Aβ₄₂ levels were positively correlated with HDL among statin nonusers. A 5-year prospective study in 440 elderly persons showed that participants with the highest third of TC or LDL-C at baseline showed lower plasma Aβ₄₂ at 5 years [18]. There were also reports that blood Aβ₄₀ and Aβ₄₂ levels were increased as a result of accumulated cholesterol [26]. All of these findings indicate that in addition to the direct effects of hyperlipidemia on brain Aβ aggregation and deposition, blood lipids may also affect peripheral Aβ transport and be involved in the cycling of Aβ between the brain and plasma.

In the current study, we found that participants with abnormal levels of blood TG/HDL-c had lower plasma Aβ levels. Plasma sLRP1 acts as a ‘sink’ for Aβ by binding 70–90%of peripheral Aβ which in turn prevents free plasma Aβ re-entry into the brain [9]. In addition, sRAGE also serves as a decoy receptor for Aβ that can inhibit the binding of Aβ to RAGE and promote peripheral Aβ clearance through blood circulation to the liver and other organs [27]. Therefore, sLRP1 and sRAGE are of great importance in the regulation of plasma Aβ. Previous studies have reported that sLRP1 and sRAGE levels in AD are lower than those in normal healthy controls [9, 28]. In the present study, we assume that lower levels of plasma sRAGE in the high TG and low HDL-c groups may indicate poor peripheral clearance of Aβ via uncoupled sRAGE and plasma Aβ (no correlations were found between plasma Aβ and sRAGE in the high TG and low HDL-c groups). Many studies have demonstrated that there is a complex equilibrium of Aβ between the central and peripheral circulatory systems [6 , 29]. In particular, recent studies suggested a similar amyloid-associated alteration between Aβ in plasma and in CSF [30], indicating that plasma Aβ is closely related to cerebral Aβ deposition. Therefore, we assumed that abnormal blood TG and HDL-c levels may cause peripheral Aβ transport and clearance disorders by lowering sRAGE levels. When the clearance of peripheral Aβ is reduced, to maintain dynamic equilibrium, more peripheral Aβ will be transported into the brain through the BBB, eventually leading to reduced plasma Aβ levels and increased cerebral Aβ deposition. A previous study [31] also reported that plasma Aβ₄₂ showed significant negative correlations with Aβ-positron emission tomography (PET) burden and had an accuracy approximately equal to 90%to discriminate Aβ (+) subjects from Aβ (–) subjects when using Pittsburgh Compound B-PET (PIB-PET) as a standard of truth, which further confirmed our hypothesis.

The present study found that sRAGE was decreased in the high TG and low HDL-c groups. However, the exact mechanism remains unclear, and the following explanations are possible. The soluble form of RAGE can be generated from two different mechanisms: esRAGE is produced by alternative splicing of RAGE mRNA [32], whereas cleaved RAGE (cRAGE) is generated from the cleavage of native membrane-bound RAGE mediated by proteolysis [33]. Consistent with our current study, accumulating evidence suggests that sRAGE is involved in dyslipidemia and related diseases. Hidenori et al. [32] found that plasma esRAGE levels are inversely associated with quantitatively determined atherosclerosis in the carotid and femoral arteries. Falconi et al. [34] also found that plasma sRAGE levels in patients with coronary artery disease were significantly lower than those in age-matched healthy controls. A large number of studies have shown that lipid metabolism is often accompanied by increased oxidative stress [35, 36]. Since the alternative splicing of RAGE mRNA is highly sensitive to redox balance, the expression of sRAGE might decrease in an environment with increased oxidative stress [37]. Moreover, to downregulate oxidative stress, sRAGE may bind to AGEs by competing with RAGE, which subsequently decreases the levels of sRAGE [38]. In addition, in response to increased oxidative stress, the exposure of RAGE to its ligands might be elevated, leading to increased RAGE expression via a positive feedback mechanism. At the same time, the activation of RAGE inhibits its self-cleavage and downregulates the expression of sRAGE [39]. These findings indicated that abnormal blood lipid levels might also reduce plasma sRAGE levels by increasing oxidative stress.

To date, the underlying mechanisms for the altered correlations between sRAGE and plasma Aβ in abnormal TG/HDL-c groups and normal controls have not been determined. As this study reported, the levels of sRAGE were lower in individuals with abnormal TG/HDL-c levels, and we assume that lower levels of plasma sRAGE may lead to a drop in sRAGE-bound Aβ, which might partially explain the lost correlation between sRAGE and plasma Aβ in the abnormal TG/HDL-c groups. Furthermore, the binding of plasma Aβ to sRAGE might, to a great extent, be affected by the level of oxidization of sRAGE. Since oxidative damage to proteins is one of the earliest events and persists during the progression of AD [40], the levels of oxidized sRAGE will be investigated in future studies to explore the exact mechanisms of their interaction with plasma Aβ and cerebral amyloidosis.

There were limitations in that many correlations and effect sizes appeared to be small in our study. However, the statistical significance was not always consistent with clinically meaningful. Sometimes, a correlation appeared to be very small but of potential clinical significance. Amyloid plaques are the pathological hallmark of AD. Aβ is the major species found in cerebral amyloid plaques and is mainly distributed in the brain [41]. Previous studies have shown that concentrations of Aβ in the central pool (CSF) were at least 5–15 times higher than those in the peripheral pool (plasma) [42, 43]. Additionally, Aβ₄₂ was quantitatively the minor form of Aβ produced, and the hemodilution effect through the high-volume circulatory system also efficiently reduced systemic Aβ levels. Therefore, we assumed that the relatively small difference in plasma Aβ between blood lipid groups may reflect a large difference in amyloid plaque burden in the brain. However, due to the lack of data on brain Aβ deposition to support our speculation, further studies are needed to determine the relationships of plasma Aβ and cerebral Aβ deposition using PIB-PET or CSF.

In summary, in this community-based cross-sectional study, abnormal levels of blood TG and HDL-c were associated with decreased plasma Aβ, sLRP1, and sRAGE concentrations. Positive correlations between plasma Aβ and sRAGE were only found in the normal TG and HDL-c groups but not in the high TG and low HDL-c groups. These results indicated that dyslipidemia contributing to plasma Aβ levels might also be involved in peripheral Aβ clearance.

Some limitations are noteworthy. First, it is difficult to explain the causal relationship between blood lipids and plasma Aβ transport given the cross-sectional study design. Many correlations and effect sizes of this study appear to be small. Therefore, longitudinal cohort studies and randomized, double-blind, placebo-controlled trials are necessary to validate our results in the future. Second, the accumulation of Aβ in the brain has not been measured. PIB-PET imaging will be used to investigate the effects of peripheral Aβ clearance on brain Aβ deposition. Third, oxidized sRAGE and oxidized LDL, which play a major role in oxidative stress, and other Aβ-binding proteins such as albumin or apolipoproteins should be investigated in future studies to explore the exact mechanisms of their interaction with plasma Aβ transport and cerebral amyloidosis.

Footnotes

ACKNOWLEDGMENTS

We were thankful for the cooperation of all participants in our study. This work was supported by the Nature Science Foundation of China (No.81771168); and the Key Research & Development Programs of Shaanxi Province (No. 2018ZDXM-SF-052); and Clinical Research Award of the First Affiliated Hospital of Xi’an Jiaotong University (No. XJTU1AF-CRF-2018-004); and the Fundamental Research Funds for the Central Universities (No. xzy022020044).

Authors’ disclosures available online ().

References

Takizawa

, Thompson

, van Walsem

, Faure

, Maier

(2015) Epidemiological and economic burden of Alzheimer’s disease: A systematic literature review of data across Europe and the United States of America. J Alzheimers Dis 43, 1271–1284.

Bondi

, Edmonds

, Salmon

(2017) Alzheimer’s disease: Past, present, and future. J Int Neuropsychol Soc 23, 818–831.

Nathalie

, Jean-Noël

(2008) Processing of amyloid precursor protein and amyloid peptide neurotoxicity. Curr Alzheimer Res 5, 92–99.

Shibata

, Yamada

, Kumar

, Calero

, Bading

, Frangione

, Holtzman

, Miller

, Strickland

, Ghiso

, Zlokovic

(2000) Clearance of Alzheimer’s amyloid-ss(1-40) peptide from brain by LDL receptor-related protein-1 at the blood-brain barrier. J Clin Invest 106, 1489–1499.

Roberts

, Elbert

, Kasten

, Patterson

, Sigurdson

, Connors

, Ovod

, Munsell

, Mawuenyega

, Miller-Thomas

, Moran

, Cross

, 3rd Derdeyn

, Bateman

(2014) Amyloid-β efflux from the central nervous system into the plasma. Ann Neurol 76, 837–844.

Giedraitis

, Sundelöf

, Irizarry

, Gårevik

, Hyman

, Wahlund

, Ingelsson

, Lannfelt

(2007) The normal equilibrium between CSF and plasma amyloid beta levels is disrupted in Alzheimer’s disease. Neurosci Lett 427, 127–131.

Devanand

, Schupf

, Stern

, Parsey

, Pelton

, Mehta

, Mayeux

(2011) Plasma Aβ and PET PiB binding are inversely related in mild cognitive impairment. Neurology 77, 125–131.

DeMattos

, Bales

, Cummins

, Dodart

, Paul

, Holtzman

(2001) Peripheral anti-A beta antibody alters CNS and plasma A beta clearance and decreases brain A beta burden in a mouse model of Alzheimer’s disease. Proc Natl Acad Sci U S A 98, 8850–8855.

Sagare

, Deane

, Bell

, Johnson

, Hamm

, Pendu

, Marky

, Lenting

, Wu

, Zarcone

, Goate

, Mayo

, Perlmutter

, Coma

, Zhong

, Zlokovic

(2007) Clearance of amyloid-beta by circulating lipoprotein receptors. Nat Med 13, 1029–1031.

10.

Tamaki

, Ohtsuki

, Iwatsubo

, Hashimoto

, Yamada

, Yabuki

, Terasaki

(2006) Major involvement of low-density lipoprotein receptor-related protein 1 in the clearance of plasma free amyloid beta-peptide by the liver. Pharm Res 23, 1407–1416.

11.

Raucci

, Cugusi

, Antonelli

, Barabino

, Monti

, Bierhaus

, Reiss

, Saftig

, Bianchi

(2008) A soluble form of the receptor for advanced glycation endproducts (RAGE) is produced by proteolytic cleavage of the membrane-bound form by the sheddase a disintegrin and metalloprotease 10 (ADAM10). FASEB J 22, 3716–3727.

12.

Cai

, Liu

, Wang

, Qin

, Zhou

, Xiao

, Chang

, Yan

, Zhao

(2016) Role of RAGE in Alzheimer’s disease. Cell Mol Neurobiol 36, 483–495.

13.

Refolo

, Malester

, LaFrancois

, Bryant-Thomas

, Wang

, Tint

, Sambamurti

, Duff

, Pappolla

(2000) Hypercholesterolemia accelerates the Alzheimer’s amyloid pathology in a transgenic mouse model. Neurobiol Dis 7, 321–331.

14.

Rockwood

, Kirkland

, Hogan

, MacKnight

, Merry

, Verreault

, Wolfson

, McDowell

(2002) Use of lipid-lowering agents, indication bias, and the risk of dementia in community-dwelling elderly people. Arch Neurol 59, 223–227.

15.

Fassbender

, Simons

, Bergmann

, Stroick

, Lutjohann

, Keller

, Runz

, Kuhl

, Bertsch

, von Bergmann

, Hennerici

, Beyreuther

, Hartmann

(2001) Simvastatin strongly reduces levels of Alzheimer’s disease beta-amyloid peptides Abeta 42 and Abeta 40 in vitro and in vivo. Proc Natl Acad Sci U S A 98, 5856–5861.

16.

Llorente-Cortés

, Otero-Viñas

, Berrozpe

, Badimon

(2004) Intracellular lipid accumulation, low-density lipoprotein receptor-related protein expression, and cell survival in vascular smooth muscle cells derived from normal and atherosclerotic human coronaries. Eur J Clin Invest 34, 182–190.

17.

Smith

, Betteridge

(2004) Plasma beta-amyloid (A beta) 40 concentration, lipid status and age in humans. Neurosci Lett 367, 48–50.

18.

Blasko

, Kemmler

, Jungwirth

, Wichart

, Weissgram

, Jellinger

, Tragl

, Fischer

(2011) Prospective study on association between plasma amyloid beta-42 and atherosclerotic risk factors. J Neural Transm (Vienna) 118, 663–672.

19.

Burgess

, McIsaac

, Naus

, Chan

, Tansley

, Yang

, Miao

, Ross

, van Eck

, Hayden

, van Nostrand

, St George-Hyslop

, Westaway

, Wellington

(2006) Elevated plasma triglyceride levels precede amyloid deposition in Alzheimer’s disease mouse models with abundant A beta in plasma. Neurobiol Dis 24, 114–127.

20.

Jiang

, Shang

, Li

, Chen

, Dang

, Wang

, Huo

, Deng

, Wang

, Qu

(2018) Pulse pressure is associated with plasma amyloid-β transport dysfunction. J Hypertens 36, 569–579.

21.

Joint committee issued Chinese guideline for the management of dyslipidemia in adults (2016) 2016 Chinese guidelines for the management of dyslipidemia in adults. Chin J Cardiol 44, 833–853.

22.

Wei

, Gao

, Jiang

, Shang

, Chen

, Dang

, Wang

, Huo

, Wang

, Qu

(2020) The apolipoprotein E ɛ4 allele-dependent relationship between serum lipid levels and cognitive function: A population-based cross-sectional study. Front Aging Neurosci 12, 44.

23.

Chen

, Wagener

, Morgan

, Hui

, Ghribi

, Geiger

(2010) Endolysosome mechanisms associated with Alzheimer’s disease-like pathology in rabbits ingesting cholesterol-enriched diet. J Alzheimers Dis 22, 1289–1303.

24.

Refolo

, Pappolla

, LaFrancois

, Malester

, Schmidt

, Thomas-Bryant

, Tint

, Wang

, Mercken

, Petanceska

, Duff

(2001) A cholesterol-lowering drug reduces beta-amyloid pathology in a transgenic mouse model of Alzheimer’s disease. Neurobiol Dis 8, 890–899.

25.

Abdullah

, Luis

, Paris

, Ait-ghezala

, Mouzon

, Allen

, Parrish

, Mullan

, Ferguson

, Wood

, Crawford

, Mullan

(2009) High serum Abeta and vascular risk factors in first-degree relatives of Alzheimer’s disease patients. Mol Med 15, 95–100.

26.

Liu

, An

, Ma

, Feng

, Wang

, Lu

, Xiao

(2018) High-cholesterol diet results in elevated amyloid-β and oxysterols in rats. Mol Med Rep 17, 1235–1240.

27.

Yamagishi

, Matsui

(2010) Soluble form of a receptor for advanced glycation end products (sRAGE) as a biomarker. Front Biosci (Elite Ed) 2, 1184–1195.

28.

Emanuele

, D’Angelo

, Tomaino

, Binetti

, Ghidoni

, Politi

, Bernardi

, Maletta

, Bruni

, Geroldi

(2005) Circulating levels of soluble receptor for advanced glycation end products in Alzheimer disease and vascular dementia. Arch Neurol 62, 1734–1736.

29.

Fandos

, Pérez-Grijalba

, Pesini

, Olmos

, Bossa

, Villemagne

, Doecke

, Fowler

, Masters

, Sarasa

(2017) Plasma amyloid β 42/40 ratios as biomarkers for amyloid β cerebral deposition in cognitively normal individuals. Alzheimers Dement (Amst) 8, 179–187.

30.

Ovod

, Ramsey

, Mawuenyega

, Bollinger

, Hicks

, Schneider

, Sullivan

, Paumier

, Holtzman

, Morris

(2017) Amyloid β concentrations and stable isotope labeling kinetics of human plasma specific to central nervous system amyloidosis. Alzheimers Dement 13, 841–849.

31.

Nakamura

, Kaneko

, Villemagne

, Kato

, Doecke

, Doré

, Fowler

, Li

, Martins

, Rowe

, Tomita

, Matsuzaki

, Ishii

, Arahata

, Iwamoto

, Ito

, Tanaka

, Masters

, Yanagisawa

(2018) High performance plasma amyloid-β biomarkers for Alzheimer’s disease. Nature 554, 249–254.

32.

Koyama

, Shoji

, Yokoyama

, Motoyama

, Mori

, Fukumoto

, Emoto

, Shoji

, Tamei

, Matsuki

, Sakurai

, Yamamoto

, Yonekura

, Watanabe

, Yamamoto

, Nishizawa

(2005) Plasma level of endogenous secretory RAGE is associated with components of the metabolic syndrome and atherosclerosis. Arterioscler Thromb Vasc Biol 25, 2587–2593.

33.

Tam

, Shiu

, Leng

, Bucala

, Betteridge

, Tan

(2011) Enhanced expression of receptor for advanced glycation end-products is associated with low circulating soluble isoforms of the receptor in Type 2 diabetes. Clin Sci (Lond) 120, 81–89.

34.

Falcone

, Emanuele

, D’Angelo

, Buzzi

, Belvito

, Cuccia

, Geroldi

(2005) Plasma levels of soluble receptor for advanced glycation end products and coronary artery disease in nondiabetic men. Arterioscler Thromb Vasc Biol 25, 1032–1037.

35.

Stocker

, Keaney

Jr. , (2004) Role of oxidative modifications in atherosclerosis. Physiol Rev 84, 1381–1478.

36.

Rizzo

, Kotur-Stevuljevic

, Berneis

, Spinas

, Rini

, Jelic-Ivanovic

, Spasojevic-Kalimanovska

, Vekic

(2009) Atherogenic dyslipidemia and oxidative stress: A new look. Transl Res 153, 217–223.

37.

Devangelio

, Santilli

, Formoso

, Ferroni

, Bucciarelli

, Michetti

, Clissa

, Ciabattoni

, Consoli

, Davì

(2007) Soluble RAGE in type 2 diabetes: Association with oxidative stress. Free Radic Biol Med 43, 511–518.

38.

Huang

, Que

, Shen

(2015) Correlation of the plasma levels of soluble RAGE and endogenous secretory RAGE with oxidative stress in pre-diabetic patients. J Diabetes Complications 29, 422–426.

39.

Lindsey

, Cipollone

, Abdullah

, McGuire

(2009) Receptor for advanced glycation end-products (RAGE) and soluble RAGE (sRAGE): Cardiovascular implications. Diab Vasc Dis Res 6, 7–14.

40.

Butterfield

, Reed

, Newman

, Sultana

(2007) Roles of amyloid beta-peptide-associated oxidative stress and brain protein modifications in the pathogenesis of Alzheimer’s disease and mild cognitive impairment. Free Radic Biol Med 43, 658–677.

41.

Younkin

(1995) Evidence that A beta 42 is the real culprit in Alzheimer’s disease. Ann Neurol 37, 287–288.

42.

Fagan

, Mintun

, Mach

, Lee

, Dence

, Shah

, Larossa

, Spinner

, Klunk

, Mathis

(2006) Inverse relation between in vivo amyloid imaging load and cerebrospinal fluid Abeta42 in humans. Ann Neurol 59, 512–519.

43.

Toledo

, Vanderstichele

, Figurski

, Aisen

, Petersen

, Weiner

, Jack

Jr. , Jagust

, Decarli

, Toga

, Toledo

, Xie

, Lee

, Trojanowski

, Shaw

(2011) Factors affecting Abeta plasma levels and their utility as biomarkers in ADNI. Acta Neuropathol 122, 401–413.