Cerebrospinal Fluid 7-Ketocholesterol Level is Associated with Amyloid-β 42 and White Matter Microstructure in Cognitively Healthy Adults

Abstract

Background:

Abnormal cholesterol metabolism changes the neuronal membrane and may promote amyloidogenesis. Oxysterols in cerebrospinal fluid (CSF) are related to Alzheimer’s disease (AD) biomarkers in mild cognitive impairment and dementia. Cholesterol turnover is important for axonal and white matter (WM) microstructure maintenance.

Objective:

We aim to demonstrate that the association of oxysterols, AD biomarkers, and WM microstructure occurs early in asymptomatic individuals.

Methods:

We studied the association of inter-individual variability of CSF 24-hydroxycholesterol (24-OHC), 27-hydroxycholesterol (27-OHC), 7-ketocholesterol (7-KC), 7β-hydroxycholesterol (7β-OHC), amyloid-β₄₂ (Aβ₄₂), total-tau (t-tau), phosphorylated-tau (p-tau), neurofilament (NfL), and WM microstructure using diffusion tensor imaging, generalized linear models and moderation/mediation analyses in 153 healthy adults.

Results:

Higher 7-KC levels were related to lower Aβ₄₂, indicative of greater AD pathology (p = 0.041)_. Higher 7-KC levels were related to lower fractional anisotropy (FA) and higher mean (MD), axial (AxD), and radial (RD) diffusivity. 7-KC modulated the association between AxD and NfL in the corpus callosum splenium (B = 39.39, p = 0.017), genu (B = 68.64, p = 0.000), and fornix (B = 10.97, p = 0.000). Lower Aβ₄₂ levels were associated to lower FA and higher MD, AxD, and RD in the fornix, corpus callosum, inferior longitudinal fasciculus, and hippocampus. The association between AxD and Aβ₄₂ was moderated by 7K-C (p = 0.048).

Conclusion:

This study adds clinical evidence to support the role of 7K-C on axonal integrity and the involvement of cholesterol metabolism in the Aβ₄₂ generation process.

Keywords

Amyloid-β cerebrospinal fluid diffusion tensor imaging oxysterols white matter

INTRODUCTION

Oxysterols have been proposed as the link between brain cholesterol metabolism and Alzheimer’s disease (AD) [1]. In AD, oxidative stress enhances the formation of oxysterols, which in turn increase pro-inflammatory mediator production and exacerbate neuronal damage, connecting all these processes in a vicious circle [2 –4]. Moreover, oxysterols play a fundamental role in amyloid-β protein precursor (AβPP) processing and amyloid-β₄₂ (Aβ₄₂) generation as they induce changes in cell membrane dynamics [5] that are related to the amyloidogenic processing in lipid rafts [6, 7]. While experimental work suggests that oxysterols contribute to amyloidogenesis [6 –8], human studies have not shown an association between circulating oxysterols and cerebrospinal fluid (CSF) Aβ₄₂ levels [9]. Higher 24-hydroxycholesterol (24OH-C) and 27-hydroxycholesterol (27OH-C), have been related to higher CSF total-tau (t-tau) and phosphorylated-tau (p-tau) levels in mild cognitive impairment (MCI) and AD, but no in healthy people [10, 11].

To analyze the axolemma and myelin integrity, which are highly dependent on cholesterol metabolism and turnover, the assessment of white matter (WM) microstructure using diffusion tensor imaging (DTI) is a valuable in vivo neuroimaging technique [12]. Tensor-derived measurements include fractional anisotropy (FA) and radial, mean, and axial diffusivities (RD, MD, AxD) [13]. Despite the importance of cholesterol metabolism in the axolemma and myelin, few studies have analyzed the effect of cholesterol mis-metabolism on the microstructure of WM [14 –16], and to our knowledge there is no previous work investigating the effect of oxysterols on WM microstructure. It is well established that patients with AD or MCI show lower FA and higher MD, AxD, and RD values especially in areas related to cognition, such as the superior and inferior longitudinal fasciculus, hippocampus, or corpus callosum [17, 18]. Moreover other studies have shown WM microstructure changes in cognitively healthy people with positive amyloid biomarkers, especially in the fornix and the uncinate fascicle [19, 20].

To better understand the relationship between oxysterol homeostasis, AD pathology, and neuronal integrity, we aimed to examine: 1) the association between CSF oxysterols and AD biomarkers; 2) the association between CSF oxysterols and WM microstructure; 3) the effect of the interaction between oxysterols and WM microstructure on AD biomarkers and 4) the involvement of axonal damage. For the present study we focus on the two major oxysterols synthetized enzymatically, 24OH-C and 27OH-C; and two synthetized non-enzymatically, 7β-hydroxycholesterol (7βOH-C) and 7-ketocholesterol (7K-C). We hypothesized that inter-individual variability of CSF oxysterols is related to changes in DTI indexes and is associated with lower Aβ₄₂ and/or higher t-tau or p-tau levels. Furthermore, to investigate the occurrence of axonal damage, we have also looked at neurofilament light chain (NfL).

MATERIALS AND METHODS

Participants

This study is a cross-sectional analysis of baseline data from the Gipuzkoa Alzheimer Project (GAP). The GAP study is an ongoing longitudinal study on preclinical and prodromal phases of AD. A cohort of 410 non-demented volunteers (aged 40–80) was recruited from the community through local media advertisements and presentations at the local Alzheimer’s Association. Baseline visits took place between June 2011 and January 2013. Exclusion criteria were dementia and any neurologic, systemic, or psychiatric disorder causing cognitive impairment. All participants completed a comprehensive clinical and neuropsychological evaluation, anthropometric and cardiovascular risk assessment, blood workup, apolipoprotein E (APOE) genotyping for the ɛ2/ɛ3/ɛ4 polymorphism, as well as brain magnetic resonance imaging (MRI). Lumbar puncture for CSF biomarkers was optional. For the present study, GAP participants deemed clinically normal based on a global Clinical Dementia Rating (CDR) score of 0 [21] with available CSF samples (n = 153) were selected. For the WM microstructure analyses, 26 participants were excluded due to incidental findings and/or presence of MRI artifacts identified in the DTI sequences by visual inspection and noted non-usable for this study.

The study was approved by the local Ethics and Clinical Research Committee, and all participants gave written informed consent to participate.

Magnetic resonance imaging

Data acquisition

Whole brain scans were obtained using a Siemens 3T Magnetom TrioTim scanner (Siemens, Erlangen, Germany) in combination with a 32-channel head coil. To increase the inter-participant homogeneity of the image acquisition, the AutoAlign Head LS software tool (Siemens) was used. Diffusion-weighted images were obtained using an echo planar imaging (EPI) sequence with the following specifications: TR 9300 ms, TE 92 ms, voxel size: 1.7 mm isotropic, 71 consecutive slices, acquisition matrix 122×122 (FOV 208 mm), 6/8 partial Fourier, 64 diffusion directions with b-value 1000 s/mm², and one image with no diffusion weighting. The bandwidth was 1640 Hz/pixel.

Image processing and analyses

Before image processing, visual quality check was performed, mainly in order to identify and exclude those DWI volumes affected by severe artefacts.

DICOM images were converted into 4D compressed NIfTI files and diffusion gradient directions were extracted with dcm2nii (part of MRIcron package, http://www.mccauslandcenter.sc.edu/mricro/mricron/). Then 4D brain volumes were processed and analyzed by Tract-Based Spatial Statistics (TBSS), part of the FSL toolbox (FMRIB Software Library, Version 5.0.5; FMRIB, Oxford, United Kingdom) [22]. After image conversion, off-resonance effects and subject movement in diffusion imaging were corrected by Eddy-Current correction tool [23] and scalar DTI maps (FA, MD, AxD, and RD) were created by fitting a tensor model to the raw diffusion data using FDT (FMRIB’s Diffusion Toolbox). Then FA images were brain-extracted using Brain Extraction Tool (BET) [24], with b = 0 as the reference volume. All participants FA data were then aligned into a study-specific FA template, using the nonlinear registration tool FNIRT [25], which uses a b-spline representation of the registration warp field [26]. The resulting warp-fields were then applied to MD, AxD, and RD images. Due to the fact that in this study the participants are middle to older age (40–75 years), we created a study-specific FA template instead of using the standard template (FMRIB58_FA) provided by FSL software, which is based on 58 participants from 20 to 50 years. Before voxel-wise statistical analyses, all participant’s FA data were obtained and projected onto a mean FA tract skeleton, which represents the centers of all tracts common to the group [27].

CSF biomarker measurements

CSF was obtained and collected following international consensus recommendations as described previously [28]. Samples were aliquoted and stored in polypropylene tubes at –80°C and shipped on dry ice to the Clinical Neurochemistry Laboratory in Gothenburg for analysis. Biomarker concentrations were measured by board-certified laboratory technicians using commercial assays as previously described [29] (MSD: Aβ_42, Aβ₄₀; Fujirebio Europe INNOTEST: t-tau and p-tau; and UmanDiagnostics: NfL). Using the cut-off value provided by the laboratory (Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden, Dr. H. Zetterberg), participants with Aβ₄₂/Aβ₄₀ <0.08 pg/mL, were considered amyloid positive (Aβ+).

Quantification of CSF oxysterols by LC-MS/MS

Human CSF samples (400μl) spiked with internal standards (1 ng 24-OHCd7, 0.25 ng 25-OHCd6, 4 ng 27-OHCd6, 0.5 ng 7β-OHCd7, 15 ng 7-KCd7) was mixed with 1600μl ice-cold methanol containing 4 mg/ml butyl hydroxytoluene (BHT) and incubated for 10 min in ice. Samples were then centrifuged (14,000×g at 4°C for 10 min) and the methanolic supernatant was diluted with acidified water up to 12.5% of methanol. Oxysterol enrichment and quantification was performed as described by Dias et al., 2018 [30]. Briefly, oxysterols were extracted using Oasis HLB Prime cartridges (bed wt. 30 mg, 1 ml, 96-well) and analyzed by liquid chromatography UltiMate 3000 HPLC (Dionex, Thermo Scientific Ltd., Hemel Hempstead, UK) on-line coupled to the ESI-QqLIT-MS/MS mass spectrometer (QTrap 5500, AB Sciex, Warrington, UK). Multiple reaction monitoring with transitions of 367.2/161 for 24-OHC, 367.4/147 for 25-OHC, 385.4/161 for 27-OHC, 385.4/81 for 7β-OHC, and 401.4/95 for 7-KC were used to collect data. Data were examined using Analyst Software 1.7.2 (AB Sciex, Warrington, UK).

APOE genotype

APOE genotype was determined using 1-stage polymerase chain reaction as previously described [31]. Participants were classified as APOE ɛ4 carriers (APOE4+) if they had at least one ɛ4 allele, or as non-carriers (APOE4–).

White matter hyperintensities

Individual global white matter hyperintensity (WMH) was semiquantified by the Fazekas scale by an experienced neuroradiologist on fluid attenuation inversion recovery MRI sequences. Fazekas scale ranges between 0 and 3 (0 = no WMH; 1 = focal/punctate lesions; 2 = beginning confluent lesions; 3 = confluent-diffuse lesions) [32].

Statistical analysis

To assess the appropriateness of parametric statistics, the Kolmogorov-Smirnov test was used to examine the normality of oxysterol distribution. Frequency distributions were calculated for categorical variables, and means (standard deviations) or median (percentile₂₅-percentile₇₅) were calculated for continuous variables.

Association between oxysterols and AD biomarkers

To test the hypothesis that inter-individual variability of CSF oxysterols is associated to lower Aβ₄₂ and/or higher t-tau or p-tau levels, generalized linear models (GLM) were conducted. The dependent variable was CSF biomarkers (Aβ₄₂, t-tau, p-tau) and the predictors 24-OHC, 27-OHC, 7-KC and 7β-OHC. The analysis was adjusted by age, sex, and APOE genotype.

Association between oxysterols and WM microstructure

All DTI indexes were correlated with age, sex, and WMH; therefore all analyses were adjusted for these confounders. To test the hypothesis that inter-individual variability of CSF oxysterols is related to changes in WM microstructure as indexed by the DTI measurements (FA, MD, RD, AxD), whole-brain voxel-wise statistical analyses were performed with a nonparametric permutation inference tool (randomise, part of the FSL toolbox). For these analyses, 5000 permutations were generated and Threshold-Free Cluster Enhancement (TFCE) was conducted as a thresholding option [33]. All the resulting statistical maps were corrected for multiple comparisons with family wise error (FWE), thresholded at p < 0.05 and only clusters with at least 100 contiguous voxels were defined as significant clusters. Resulting significant clusters and regions were labelled using the atlas available through FSL toolbox (ICBM-DTI-81 white matter labels atlas).

Based on previous results linking WM microstructure in certain areas with cognitive performance [34 –36], the subsequent analyses were performed in seven regions of interest (ROI): corpus callosum, inferior and superior longitudinal fasciculus, cingulate gyrus, hippocampus, uncinate fasciculus, fornix (column + body), and fornix cres/stria terminalis. Individual DTI mean values were extracted in these regions from the voxels with significant findings in the previous whole-brain voxel-wise analyses. These regions were labeled using the atlas available through FSL toolbox (ICBM-DTI-81 white matter labels atlas). The DTI indexes were multiplied by 1000 in the analyses [37].

Effect of the interaction between oxysterols and WM microstructure on AD biomarkers

To test the hypothesis that the association between oxysterols and WM affects the AD pathology (biomarkers), first the association between amyloid and WM microstructure was investigated. In order to ensure that the possible association between amyloid and WM microstructure occurs in the same regions where the oxysterol showed a significant effect, we conducted the analysis with the DTI indexes obtained from the areas (voxels) where the oxysterol had shown a significant effect. GLMs were conducted. The dependent variable was the mean of the DTI measures of each region and the predictor was CSF amyloid.

If oxysterol in one hand and amyloid on the other were associated to WM microstructure, the effect of the interaction oxysterol*WM microstructure on AD pathology was investigated. GLMs were designed with data from the region where the oxysterol and amyloid showed a significant effect. Amyloid was the dependent variable and the interaction oxysterol*WM microstructure was predictive variable. If an interaction was significant, mediation and moderation analyses were performed. Moderation is shown up by a significant interaction effect.

Involvement of axonal damage

To study to what extent axonal damage is involved in the association between WM microstructure, oxysterols and AD pathology, new GLMs were conducted. We analyzed whether the interactions oxysterol*DTI and AD biomarkers*DTI had any effect on CSF NfL, as a biomarker of axonal damage. If an interaction was significant, mediation and moderation analyses were performed.

Statistical analyses were conducted in SPSS version 20. For mediation and moderation analysis we applied the PROCESS macro for SPSS (version 3) by Andrew F. Hayes (http://www.afhayes.com).

Data availability

Data from this work not provided in this article will be shared at the request of any other investigator for purposes of replicating procedures and results.

RESULTS

Descriptive data for the whole sample (n = 153, CDR score = 0) and for the sample with MRI are shown in Table 1. Participant age range was 40–75, 67 (43.8%) were women and mean MMSE score was 29. 25 participants (16.6%) were Aβ+.

Table 1

Sample characteristics

Characteristic	All sample (n = 153)	Sample with DTI analysis (n = 127)	p ^a
Age, y	57.86 (6.80)	57.80 (7.03)	0.664
Sex (% female)	67 (43.8)	78 (61.4)	0.005
Education, y	14 (11–17)	14.00 (11–17)	0.525
MMSE, score	29 (28–30)	29.00 (28–30)	0.719
APOEɛ4 carrier	36 (23.5)	27 (21.3)	0.202
Fazekas scale, n (%)			0.953
0	86 (56.6)	71 (55.9)
1	52 (34.2)	44 (34.6)
2	9 (5.9)	8 (6.3)
3	5 (3.3)	4 (3.1)
24-OHC, ng/dl	1.75 (1.36–2.67)	1.73 (1.36–2.67)	0.973
27-OHC, ng/ml	2.28 (1.24)	2.29 (1.10)	0.924
7-KC, ng/dl	13.60 (6.44–24.45)	13.12 (6.36–20.46)	0.066
7β-OHC, ng/dl	2.70 (1.80–4.69)	2.75 (1.80–4.84)	0.653
Aβ₄₂, pg/ml	478.7 (152.2)	478.31 (154.02)	0.939
t-tau, pg/ml	268.4 (201.9–336.8)	270.8 (194.2–353.6)	0.451
p-tau, pg/ml	38.50 (30–46)	39.00 (30–47)	0.678
NfL, pg/ml	501.9 (397.2–630.8)	501.9 (391.1–630.8)	0.730

Significant effects are shown in bold. Mean (standard deviation) of measures, median (p25–p75) and in categorical variables, n (%). 24OH-C, 24-hydroxycholesterol; 27OH-C, 27-hydroxycholesterol; 7K-C, 7-ketocholesterol; 7βOH-C, 7β-hydroxycholesterol; MMSE, Mini-Mental Status Examination; APOE, Apolipoprotein E; NfL, neurofilament light; t-tau, total-tau; p-tau, phosphorylated tau; Aβ₄₂, amyloid-β_1–42. ^aComparison between participants with and without DTI analysis.

Association between oxysterols and AD biomarkers

To study the association between oxysterol levels and CSF AD biomarkers, GLMs were conducted among 153 healthy participants controlling for age, sex, and APOE genotype. Aβ₄₂/Aβ₄₀ ratios, but not 7-KC, 24-OHC, 27-OHC, and 7β-OHC levels, were significantly different between APOE+ and APOE–. Higher 7-KC was associated with lower Aβ₄₂, indicative of greater AD pathology (B = –1.41; p = 0.041) (Table 2). When participants with Fazekas 2 and 3 were excluded for the analysis, the association between 7-KC and Aβ₄₂ remains (B = –1.43; p = 0.048). There were no effects of 7-KC on t-tau or p-tau. The other oxysterols, 24-OHC, 27-OHC, and 7β-OHC, did not show any significant effect on any CSF biomarker.

Table 2

Relationship between CSF oxysterols and AD CSF biomarkers

	Aβ₄₂		t-tau		p-tau
	B (95% CI)	p	B 95% CI)	p	B (95% CI)	p
Model 1
24OH-C	–6.50 (–23.62–10.62)	0.457	–1.35 (–13.02–10.32)	0.821	0.06 (–1.35–1.47)	0.939
Age	1.35 (–2.23–4.92)	0.460	6.76 (4.33–9.19)	0.000	0.77 (0.48–1.07)	0.000
Sex: F	–4.39 (–52.61–43.84)	0.858	–7.42 (–40.26–25.42)	0.658	0.08 (–3.89–4.06)	0.967
Sex: M	0^a		0^a		0^a
APOE4+	–52.10 (–108.91–4.70)	0.072	21.37 (–17.14–59.88)	0.277	2.58 (–2.07–7.23)	0.277
APOE4–	0^a		0^a		0^a
Model 2
27OH-C	5.86 (–13.85–25.57)	0.560	1.98 (–11.43–15.39)	0.772	0.91 (–0.70–2.53)	0.266
Age	1.31 (–2.29–4.90)	0.476	6.78 (4.34–9.22)	0.000	0.80 (0.50–1.09)	0.000
Sex: F	–5.68 (–54.74–43.38)	0.820	–6.52 (–39.89–26.84)	0.702	0.06 (–3.96–4.09)	0.975
Sex: M	0^a		0^a		0^a
APOE4+	–54.15 (–112.04–3.74)	0.067	23.42 (–15.78–62.63)	0.242	2.82 (–1.89–7.54)	0.241
APOE4–	0^a		0^a		0^a
Model 3
7K-C	–1.41 (–2.77–0.05)	0.041	–0.46 (–1.39–0.48)	0.338	–0.08 (–0.19–0.03)	0.170
Age	1.44 (–2.08–4.96)	0.422	6.79 (4.38–9.20)	0.000	0.79 (0.50–1.08)	0.000
Sex: F	–3.17 (–51.10–44.76)	0.897	–6.33 (–39.23–26.57)	0.706	0.17 (–3.80–4.14)	0.934
Sex: M	0^a		0^a		0^a
APOE4+	–54.87 (–111.25–1.51)	0.056	19.96 (–18.58–58.51)	0.310	2.42 (–2.23–7.06)	0.308
APOE4–	0^a		0^a		0^a
Model 4
7β-C	1.57 (–6.00–9.13)	0.685	–1.97 (–6.91–2.97)	0.435	–0.25 (–0.85–0.35)	0.417
Age	0.75 (–2.87–4.37)	0.685	6.96 (4.52–9.40)	0.000	0.80 (0.50–1.09)	0.000
Sex: F	–5.33 (–54.64–43.99)	0.832	–9.79 (–43.17–23.59)	0.566	–0.22 (–3.82–4.27)	0.913
Sex: M	0^a		0^a		0^a
APOE4+	–58.87 (–116.71–1.04)	0.046	24.78 (–14.11–63.68)	0.212	2.75 (–1.83–7.58)	0.231
APOE4–	0^a		0^a		0^a

Significant effects are shown in bold. B, beta coefficient; CI, confidence interval; APOE4+, apolipoprotein E ɛ4 carrier; APOE4–, apolipoprotein E ɛ4 non-carrier; 24OH-C, 24-hydroxycholesterol; 27OH-C, 27-hydroxycholesterol; 7K-C, 7-ketocholesterol; 7βOH-C, 7β-hydroxycholesterol; Aβ₄₂, amyloid-β_1–42; t-tau, total-tau; p-tau, phosphorylated tau. ^aSet to zero because this parameter is redundant.

Association between oxysterols and WM microstructure

To study the correlation between oxysterols and WM microstructure, whole-brain voxel-wise correlational analysis was conducted among 127 healthy participants controlling for age, sex, and WMH. Descriptive data for the subsample with available MRI included more women than the all sample (p = 0.005) (Table 1).

The analysis detected a significant association between 7-KC and FA (the higher 7K-C, the lower FA) (Fig. 1). Most of these voxels were located in the corpus callosum. Significant associations were also seen in the following regions: left anterior and right posterior corona radiata, left external capsule, right hippocampus, and right superior longitudinal fascicule. We did not observe significant association between higher 7K-C and higher FA. Analyses also revealed a significant correlation between 7-KC and MD, AxD, and RD (Fig. 1) (the higher 7-KC, the higher MD, AxD, and MD) distributed throughout the whole brain WM in both hemispheres. We did not observe significant association between higher 7-KC and lower MD, AxD, or RD. Table 3 shows the distribution and proportion of voxels where a significant association between 7-KC and each DTI measure was detected in every ROI.

Fig. 1

Association between 7K-C and white matter microstructure. Data adjusted for age, sex, and white matter hyperintensities (FWE-corrected, thresholded at p < 0.05, n = 127). Results are shown as an overlay on the FSL standard T1 template brain (coordinates according to MINI152 template). 7K-C, 7-ketocholesterol; L, left; R, right, FA, fractional anisotropy; MD, mean diffusivity; AxD, axial diffusivity; RD, radial diffusivity.

There was no significant effect for 24-OHC, 27-OHC, and 7-βOHC in any of the DTI measurements.

Effect of the interaction between oxysterols and WM microstructure on AD biomarkers

As 7-KC had a significant effect on Aβ₄₂ and WM microstructure, the association between Aβ₄₂ and DTI indexes of WM microstructure was analyzed in the voxels where 7-KC showed a significant effect (Table 3). GLMs were adjusted for age, sex, WMH, and APOE genotype.

Table 3

Distribution and proportion of significant voxels for the association between 7K-C and ROIs DTI indexes

Region (No of voxels)^a		Number of significant voxels (%)^b
		FA	MD	AxD	RD
CC	Genu (8851)	429 (4.85)	470 (5.31)	98 (1.11)	2 (0.02)
	Body (13711)	247 (1.80)	658 (4.80)	0	21 (0.15)
	Splenium (12729)	223 (1.75)	524 (4.12)	116 (0.91)	387 (3.04)
Fornix	Column + body (659)	0	50 (7.59)	53 (8.04)	47 (7.13)
	Cres/stria terminalis Right (1124)	0	36 (3.20)	16 (1.42)	42 (3.74)
	Cres/stria terminalis Left (1125)	0	68 (6.04)	55 (4.89)	61 (5.42)
ILF	Right (2228)	0	298 (13.38)	231 (10.37)	255 (11.45)
	Left (2231)	0	258 (11.56)	240 (10.76)	35 (1.57)
SLF	Right (6607)	118 (1.79)	578 (8.75)	332 (5.02)	617(9.34)
	Left (6605)	0	447 (6.77)	4 (0.06)	397 (6.01)
GC	Right (2342)	0	31 (1.32)	0	0
	Left (2751)	16 (0.58)	42 (1.53)	0	0
Hippoc.	Right (1236)	239 (19.34)	229 (18.53)	28 (2.27)	247 (19.98)
	Left (1155)	0	120 (10.39)	0	0
FU	Right (380)	0	38 (10.00)	0	0
	Left (376)	0	0	0	0

Number of significant voxels (FWE-corrected and p < 0.05). CC, corpus callosum; ILF, inferior longitudinal fasciculus; SLF, superior longitudinal fasciculus; GC, gyrus cingulum; Hippoc., hippocampus; UF, uncinated fasciculus. ^aIn parenthesis, the number of voxels according to ICBM-DTI-81 white matter labels atlas. ^bIn parenthesis, the proportion of significant voxels with respect to the total number of voxels in this area.

Higher Aβ₄₂ was associated with higher FA on the splenium of the corpus callosum and AxD of the right hippocampus; lower MD on the body of the corpus callosum, fornix (column + body) and left inferior longitudinal fasciculus; lower AxD on the genu and splenium of the corpus callosum and fornix (column + body); and lower RD on the fornix (column + body) (Fig. 2). The βA+ group (n = 25) had higher MD, AxD, and RD values in the fornix (column + body) than βA–group (n = 126) (MD: βA+, median = 1.75; βA–, median = 1.42; p = 0.001/AxD: βA+, median = 2.32; βA–, median = 2.03; p = 0.003/RD: βA+, median = 1.50; βA–, median = 1.11; p = 0.002) (Fig. 2).

Fig. 2

Association between Aβ₄₂ and white matter microstructure. Analyses were performed in the voxels were 7-ketocholesterol showed significant effect. Data adjusted for age, sex, white matter hyperintensities, and APOE genotype. Aβ_42, amyloid-β_1–42; FA, fractional anisotropy; MD, mean diffusivity; AxD, axial diffusivity; RD, radial diffusivity.

The effect of the 7-KC and its interaction with the WM microstructure indexes (FA*7-KC; MD*7-KC; AxD*7-KC; RD*7-KC) on the level of Aβ₄₂ was studied. Higher MD, AxD, and RD of the fornix, and higher AxD of the splenium and genu of the corpus callosum, right hippocampus, and left inferior longitudinal fasciculus were related to lower Aβ₄₂, sign of greater AD pathology (Table 4). In these models, 7-KC and the interactions between 7-KC and DTI measures were not significant. In the left inferior longitudinal fasciculus model, AxD, 7-KC, and 7-KC*AxD interaction showed a significant effect. Moderation analyses showed that the relationship between AxD and Aβ₄₂ was moderated by 7-KC level (B = 27.66, p = 0.048).

Table 4

Effect of 7K-C and ROIs WM microstructure on Aβ₄₂ levels

	FRNX (MD)		FRNX (AxD)		FRNX (RD)
	B (95% CI)	p ^a	B (95% CI)	p ^a	B (95% CI)	p ^a
7K-C	–1.06 (–5.45–3.32)	0.635	–0.90 (–6.97–5.17)	0.771	–1.04 (–4.67–2.59)	0.575
MD	–121.19 (–207.72–34.66)	0.006
7K-C*MD	0.16 (–2.10–2.42)	0.890
AxD			–131.74 (–230.42–33.06)	0.009
7K-C*AxD			0.07 (–2.40–2.54)	0.955
RD					–112.62 (–192.35–32.89)	0.006
7K-C*RD					0.16 (–2.00–2.32)	0.884
	CC-S (FA)		CC-S (AxD)		CC-B (MD)
	B (95% CI)	p ^a	B(95% CI)	p ^a	B (95% CI)	p ^a
7K-C	4.66 (–35.40–44.71)	0.820	–17.89 (–43.85–8.07)	0.177	5.73 (–18.63–30.08)	0.645
FA	1.09 (–0.39–2.57)	0.148
7K-C*FA	–0.01 (–0.06–0.05)	0.796
AxD			–472.22 (–900.19–44.26)	0.031
7K-C*AxD			11.02 (–5.74–27.77)	0.198
MD					–457.53 (–1498.56–583.51)	0.389
7K-C*MD					–8.56 (–40.36–23.24)	0.598
	CC-G (AxD)		ILF L (AxD)		Hippoc. R (AxD)
	B (95% CI)	p ^a	B (95% CI)	p ^a	B (95% CI)	p ^a
7K-C	–2.24 (–29.75–25.28)	0.873	–38.73 (–74.93–2.52)	0.036	–0.13 (–8.35–8.10)	0.976
AD	–440.17 (–813.01–67.33)	0.021	–722.00 (–1287.77–156.23)	0.012	248.42 (57.91–438.94)	0.011
7K-C*AxD	0.95 (–16.40–18.31)	0.914	27.67 (1.22–54.11)	0.040	–1.65 (–9.62–6.33)	0.686

Significant effects are shown in bold. B, beta coefficient; CI, confidence interval; Aβ₄₂, amyloid-β_1–42; CC-S, corpus callosum-splenium; CC-B, corpus callosum-body; CC-G, corpus-callosum-genu; ILF L, inferior longitudinal fasciculus left; FRNX, fornix (column + body); Hippoc. R, hippocampus right; FA, fractional anisotropy; MD, mean diffusivity; AxD, axial diffusivity; RD, radial diffusivity. ^aGeneralized linear models adjusted for sex, age, white matter hyperintensities, and APOE genotype.

Involvement of axonal damage

As AxD is related to axonal damage, the effect of the interaction between 7K-C*AxD and Aβ₄₂*AxD on NfL for ROI was analyzed. The 7-KC*AxD interaction showed a significant effect on NfL in the fornix and in the splenium and genu of the corpus callosum (Table 5). In these regions, the relationship between AxD and NfL was moderated by 7-KC (Fornix: B = 10.97, p = 0.000; splenium: B = 39.39, p = 0.017; Genu: B = 68.64, p = 0.000). The Aβ₄₂*AxD showed no significant effect on NfL in any region.

Table 5

Effect of 7K-C, Aβ₄₂, and ROIs axial diffusivity on NfL levels

	FRNX		CC-S
	B (95% CI)	p ^a	B(95% CI)	p ^a
Model 1
AxD	–202.83 (–377.69–27.97)	0.023	–633.11 (–1425.09–158.87)	0.117
7K-C	–24.07 (–34.83–13.32)	0.000	–58.49 (–106.53–10.45)	0.017
AxD *7K-C	10.98 (6.60–15.35)	0.000	39.40 (8.39–70.41)	0.013
Model 2
AxD	331.43 (–121.02–783.88)	0.151	–456.89 (–2281.61–1367.83)	0.624
Aβ₄₂	1.17 (–0.84–3.18)	0.254	–2.13 (–7.69–3.43)	0.452
AxD*Aβ_1–42	–0.42 (–1.36–0.53)	0.388	1.60 (–2.16–5.37)	0.404
	CC-G		ILF L
	B (95% CI)	p ^a	B (95% CI)	p ^a
Model 1
AxD	–855.54 (–1523.47–187.60)	0.012	–447.54 (–1515.27–620.18)	0.411
7K-C	–106.22 (–155.50–56.93)	0.000	20.29 (–48.03–88.62)	0.561
AxD *7K-C	68.64 (37.55–99.73)	0.000	–12.48 (–62.38–37.43)	0.624
Model 2
AxD	–216.69 (–1753.82–1320.44)	0.782	1016.47 (–1347.23–3380.17)	0.399
Aβ₄₂	–1.28 (–5.60–3.05)	0.563	3.79 (–2.51–10.10)	0.238
AxD*Aβ₄₂	1.01 (–1.85–3.86)	0.490	–2.75 (–7.55–2.05)	0.261
	Hippoc. R
	B (95% CI)MODEL	p ^a
Model 1
AxD	–45.98 (–413.78–321.81)	0.806
7K-C	4.90 (–10.99–20.78)	0.546
AxD *7K-C	–2.14 (–17.54–13.25)	0.785
Model 2
AxD	–19.19 (–962.30–923.92)	0.968
Aβ₄₂	0.27 (–1.42–1.97)	0.750
AxD*Aβ₄₂	–0.07 (–1.89–1.76)	0.942

Significant effects are shown in bold. B, beta coefficient; CI, confidence interval; Aβ₄₂, amyloid-β_1–42; CC-S, corpus callosum-splenium; CC-G, corpus-callosum-genu; ILF L, inferior longitudinal fasciculus left; FRNX, fornix (column + body); Hippoc. R, hippocampus right, AxD, axial diffusivity; NfL, neurofilament light; 7K-C, 7-ketocholesterol. ^aGeneralized linear models adjusted for sex, age, white matter hyperintensities, and APOE genotype.

DISCUSSION

We have investigated the association between oxysterols, CSF AD biomarkers, and WM microstructure in a population of cognitively healthy middle-aged participants. According to our results, higher 7-KC levels are related to lower Aβ₄₂, indicative of greater AD pathology, and higher 7-KC levels are associated with lower FA and higher MD, AxD, and RD in multiple WM regions, a sign of WM damage. Moreover, 7-KC modulated the association between AxD and axonal loss, as measured by NfL, in the splenium and genu of the corpus callosum and fornix. Interestingly, in the voxels where 7-KC showed a significant effect on FA, MD, AxD, and RD, lower Aβ₄₂ levels were related to lower FA and higher MD, AxD, and RD values in the fornix, corpus callosum, inferior longitudinal fasciculus, and hippocampus. Furthermore, participants who were Aβ+had significantly higher MD, AxD, and RD values in the fornix than the Aβ–group.

Association between oxysterols and AD biomarkers

Previous case-control studies on CSF oxysterol levels have reported that 24-OHC and 27-OHC levels are higher in patients with AD or MCI [10,38, 10,38]. Other studies, analyzing patients with AD or MCI and controls all together, have shown that higher 24-OHC CSF concentrations were related to higher t-tau and p-tau levels, but no relation was described with Aβ₄₂ [9,10, 9,10]. However, these correlations were significant only in the AD or MCI group, but not in healthy people. It has also been shown that higher 24-OHC levels are associated with higher sAβPPα and sAβPPβ levels in patients with AD, but not with Aβ₄₂ [39]. Our results confirm the absence of association between 24-OHC and 27-OHC and CSF AD biomarkers in cognitively healthy adults. This suggests that if 24-OHC and 27-OHC are involved in AD pathophysiology this would not represent an early event. Although 24-OHC and 27-OHC are the oxysterols most widely considered to be potentially implicated in AD pathogenesis, the possible involvement of oxysterols resulting from cholesterol autoxidation is now emerging. In fact, in a recent systematic analysis of oxysterols in postmortem human AD brains, the level of some of the oxysterols deriving from cholesterol autoxidation were higher in patients in the Braak I-II stages than in controls, and there were no differences in the levels of the two oxysterols of enzymatic origin [40]. 7-KC is a major product of reactive oxygen species-caused oxidation of cholesterol [41]. We found that higher 7-KC levels were associated with lower Aβ_42 . This result is in line with previous model lipid membrane studies reporting a relation between 7-KC and Aβ₄₂ [42,43, 42,43]_. One study demonstrated that partial substitution of cholesterol with 7-KC in the model lipid membrane enhances Aβ₄₂ insertion into the lipid bilayer, by decreasing intramolecular cohesive interactions [43]. In the same line, it has been shown that 7-KC renders lipid bilayer less condensed and more fluid than cholesterol, thus accelerating Aβ₄₂ association with the bilayer [6], specially the protofibrillar Aβ₄₂ [42]. Moreover, a recent model lipid membrane study demonstrated that cholesterol and 7-KC have different effects on membrane-mediated aggregation of Aβ₄₂. While cholesterol inhibited the nucleation step and accelerated fibrillar Aβ₄₂ growth, the partial substitution of membrane cholesterol with 7-KC slightly increased the nucleation phase and remarkably decreased fibril elongation [44]. The oligomers or protofibrillar Aβ₄₂ formed in the nucleation step are reportedly more toxic than soluble monomers and mature fibrils [45]. These papers suggested that cholesterol and 7-KC can modulate interaction of Aβ₄₂ with cell membranes by influencing the fibrillation of the peptide. While the significance of the association we found between 7-KC and Aβ₄₂ is marginal, our work adds clinical evidence to support these experimental studies on the relation between 7-KC and Aβ₄₂.

Association between oxysterols and WM microstructure

Regarding WM microstructure, we have shown that higher 7-KC was associated with lower FA and higher MD, AxD, and RD, indicative of WM damage [46] in multiple ROIs, but especially in the superior and inferior longitudinal fasciculus and hippocampus. These areas are considered late myelination regions. Several studies have described that late myelination region fibers are more vulnerable than fibers from early myelination regions [47, 48]. Our study suggests that correlates of this vulnerability may involve 7-KC. In any case, to our knowledge, this might be the first study investigating the effect of oxysterols on WM microstructure. Therefore, our findings should be confirmed with other studies.

Effect of the interaction between oxysterols and WM microstructure on AD biomarkers

With respect to Aβ₄₂, we found a relation between lower CSF Aβ₄₂, sign of greater AD pathology, and lower FA and higher MD, AxD, or RD in areas related to amyloid deposition [49] such as fornix, corpus callosum, inferior longitudinal fasciculus, and hippocampus in those voxels where 7-KC showed a significant effect on the DTI indexes. Previous studies have shown lower FA and higher MD in patients with AD, especially in areas such as corpus callosum, fornix, superior, and inferior longitudinal fasciculus or cingulum [17 , 50]. Similar changes in WM microstructure are also found in cognitively healthy persons with abnormal Aβ₄₂ levels, especially in the uncinate fasciculus and fornix [19, 20]. Persons with a genetic AD show greater changes on the fornix in comparison with other regions such as the hippocampus [51]. These findings are in line with our results. In our study, Aβ+ participants have significantly higher MD, AxD, and RD values in the fornix than the Aβ–group. This result, together with the literature mentioned above, suggests that fornix may be selectively more vulnerable than other WM regions in the early stage of the disease. In this regard, some studies go further and suggest that microstructural degradation of the fornix, which precedes hippocampal atrophy, may serve as a new imaging marker for subjects with early-stage MCI [36].

We have found a relation between Aβ₄₂, 7K-C, and WM microstructure in ROIs. Axon integrity depends mainly on the axolemma structure which in turn depends on cholesterol turnover. Our work would suggest that cholesterol metabolism and integration of 7-KC in the axolemma could change the structure of the membrane, alter WM microstructure and may promote amyloidogenesis. Our results showed that 7K-C was not a moderator or mediator of the association between Aβ₄₂ and WM microstructure in ROIs. This observation could be interpreted as meaning that although the 7K-C increase would favor amyloid pathology, amyloid may have an independent effect on the WM microstructure. The fact that it was observed only in regions directly related to the brain areas where pathological amyloid deposits take place [49] would support this view.

Involvement of axonal damage

Neurofilament proteins have been proposed as a marker of neuroaxonal damage [52]. We found that 7-KC modulates the relation between NfL and AxD. The interpretation of AxD index is complex. Some early experimental studies in selected mouse models of WM abnormalities demonstrated that decreased AxD may represent reliable in vivo surrogate markers of axonal damage [53, 54]. However in some pathological conditions or diseases, when axonal loss, axonal injury, and demyelination coexist, the decreased anisotropic diffusion seems to enhance axial diffusivity and attempts to distinguish axonal damage can lead to misinterpretation of the results [46, 55]. Our results demonstrated that interindividual 7K-C levels affect the anisotropic diffusion. 7-KC induces oxidative stress [56] which in turn induces activation of microglia and astrocytes with a consequent increase of pro-inflammatory mediator production and exacerbating neuronal damage [2]. Our results suggest that the WM microstructure changes are related to axonal damage and altered cholesterol metabolism and increased levels of 7-KC. Future studies should look also at myelin proteins in CSF and neuroinflammation markers.

Limitations

The GAP study has allowed us to analyze a rigorously phenotyped cohort with strict clinical and cognitive criteria to exclude participants with cognitive impairment. However, the study must also be interpreted in the light of its limitations. An important one is that DTI metrics are only an indirect measure of WM microstructure and the influence of crossing fibers makes the interpretation more complicated in all DTI studies. However, we have performed a high resolution 3T-DWI with 65 directions which has allowed us to realize robust estimates of tensor derived-properties [57]. The significant value of the association between 7-KC and Aβ₄₂ is marginal. Our results may contribute to the understanding of the role of lipid metabolism in the association between oxidative stress and AD on one hand, and changes in the microstructure of the WM on the other hand. However, our findings should be confirmed in larger samples including participants with clinical expression of the AD process (MCI or very mild dementia). We did not find any effect of 24-OHC, 27-OHC, or 7β-OHC on WM microstructure. This should also be investigated further in symptomatic patients. This is a cross-sectional study, which cannot characterize how the association between 7-KC, Aβ₄₂, and WM microstructure changes over time. Longitudinal follow-up data on the GAP study cohort are now being collected and will be the subject of further analysis to confirm these cross-sectional findings.

Conclusion

This study shows that interindividual variability of CSF 7-KC levels in healthy adults are associated with Aβ₄₂ levels adding clinical evidence to support experimental studies on the potential effect of 7-KC in the Aβ₄₂ aggregation process and membrane structure. Furthermore, this study reports the effect of 7-KC on WM microstructure and its modulating effect on the relation between axonal damage and WM microstructure.

Footnotes

ACKNOWLEDGMENTS

The authors gratefully acknowledge the volunteers (a Basque Cohort) who participate in the Gipuzkoa Alzheimer Project (GAP) for their generous collaboration. This work has been possible thanks to the support of the computing infrastructure of the i2BASQUE academic network ().

This work was supported by the Department of Economic Promotion, Rural Areas and Territorial Balance of the Provincial Government of Gipuzkoa (124/16); the Department of Health of the Basque Government (2016111096); and by the Carlos III Institute of Health (PI15/00919, PN de I + D + I 2013-2016). Predoctoral fellowship grant (Programa Predoctoral, de Formación de Personal Investigador no doctor, RBFI-2015-1-0231) was received from the Basque Government (AIJ). It was undertaken at CITA Alzheimer Foundation, Centre for Research and Advanced Therapies for Alzheimer’s disease (), which is supported by the Ministry of Economy and Competitiveness of Spain, the Basque Country Government, Kutxa-Fundazioa and anonymous private sponsors. HZ is a Wallenberg Academy Fellow supported by grants from the Swedish Research Council (#2018-02532), the European Research Council (#681712) and Swedish State Support for Clinical Research (#ALFGBG-720931).

Authors’ disclosures available online ().

References

Zarrouk

, Debbabi

, Bezine

, Karym

, Badreddine

, Rouaud

, Moreau

, Cherkaoui-Malki

, El Ayeb

, Nasser

, Hammami

, Lizard

(2018) Lipid biomarkers in Alzheimer’s disease. Curr Alzheimer Res 15, 303–312.

Gamba

, Testa

, Gargiulo

, Staurenghi

, Poli

, Leonarduzzi

(2015) Oxidized cholesterol as the driving force behind the development of Alzheimer’s disease. Front Aging Neurosci 7, 119.

Hughes

, Rosano

, Evans

, Kuller

(2013) Brain cholesterol metabolism, oxysterols, and dementia. J Alzheimers Dis 33, 891–911.

Dias

IHK

, Mistry

, Fell

, Reis

, Spickett

, Polidori

, Lip

GYH

, Griffiths

(2014) Oxidized LDL lipids increase β-amyloid production by SH-SY5Y cells through glutathione depletion and lipid raft formation. Free Radic Biol Med 75, 48–59.

Wnętrzak

, Chachaj-Brekiesz

, Janikowska-Sagan

, Dynarowicz-Latka

(2019) Influence of 7α-hydroxycholesterol on sphingomyelin and sphingomyelin/phosphatidylcholine films –The Langmuir monolayer study complemented with theoretical calculations. Biochim Biophys Acta Biomembr 1861, 861–870.

Phan

HTT

, Hata

, Morita

, Yoda

, Hamada

, Vestergaard

, Takagi

(2013) The effect of oxysterols on the interaction of Alzheimer’s amyloid beta with model membranes. Biochim Biophys Acta 1828, 2487–2495.

Famer

, Meaney

, Mousavi

, Nordberg

, Björkhem

, Crisby

(2007) Regulation of α- and β-secretase activity by oxysterols: Cerebrosterol stimulates processing of APP via the α-secretase pathway. Biochem Biophys Res Commun 359, 46–50.

Marwarha

, Raza

, Prasanthi

JRP

, Ghribi

(2013) Gadd153 and NF-κB crosstalk regulates 27-hydroxycholesterol-induced increase in BACE1 and β-amyloid production in human neuroblastoma SH-SY5Y cells. PLoS One 8, e70773.

Popp

, Meichsner

, Kölsch

, Lewczuk

, Maier

, Kornhuber

, Jessen

, Lütjohann

(2013) Cerebral and extracerebral cholesterol metabolism and CSF markers of Alzheimer’s disease. Biochem Pharmacol 86, 37–42.

10.

Leoni

, Solomon

, Lövgren-Sandblom

, Minthon

, Blennow

, Hansson

, Wahlund

L-O

, Kivipelto

, Björkhem

(2013) Diagnostic power of 24S-hydroxycholesterol in cerebrospinal fluid: Candidate marker of brain health. J Alzheimers Dis 36, 739–47.

11.

Shafaati

, Solomon

, Kivipelto

, Björkhem

, Leoni

(2007) Levels of ApoE in cerebrospinal fluid are correlated with tau and 24S-hydroxycholesterol in patients with cognitive disorders. Neurosci Lett 425, 78–82.

12.

Tournier

J-D

, Mori

, Leemans

, Morgan

(2011) Diffusion tensor imaging and beyond. Magn Reson Med 65, 1532–1556.

13.

Alexander

, Lee

, Lazar

, Field

(2007) Diffusion tensor imaging of the brain. Neurotherapeutics 4, 316–329.

14.

Warstadt

, Dennis

, Jahanshad

, Kohannim

, Nir

, Mcmahon

, De Zubicaray

, Montgomery

, Henders

, Martin

, Whitfield

Jr CRJ , Bernstein

, Weiner

, Toga

, Wright

, Thompson

(2014) Serum cholesterol and variant in cholesterol-related gene CETP predict white matter microstructure and for the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Neurobiol Aging 35, 2504–2513.

15.

Ryu

C-W

, Coutu

J-P

, Greka

, Rosas

, Jahng

G-H

, Rosen

, Salat

(2017) Differential associations between systemic markers of disease and white matter tissue health in middle-aged and older adults. J Cereb Blood Flow Metab 37, 3568–3579.

16.

Williams

, Leritz

, Shepel

, Mcglinchey

, Milberg

, Rudolph

, Lipsitz

, Salat

(2013) Interindividual variation in serum cholesterol is associated with regional white matter tissue integrity in older adults. Hum Brain Mapp 34, 1826–1841.

17.

Kumar

, Singh

, Verma

, Mishra

(2018) Proteomics based identification of differential plasma proteins and changes in white matter integrity as markers in early detection of mild cognitive impaired subjects at high risk of Alzheimer’s disease. Neurosci Lett 676, 71–77.

18.

K-C

, Luo

, Zeng

Q-Z

, Xu

X-J

, Huang

P-Y

, Shen

Z-J

, Xu

J-J

, Zhou

, Zhang

M-M

(2018) Distinct patterns of interhemispheric connectivity in patients with early- and late-onset Alzheimer’s disease. Front Aging Neurosci 10, 261.

19.

Vipin

, Ng

, Ji

, Shim

, Lim

JKW

, Pasternak

, Zhou

, Alzheimer’s Disease Neuroimaging Initiative (2019) Amyloid burden accelerates white matter degradation in cognitively normal elderly individuals. Hum Brain Mapp 40, 2065–2075.

20.

Al-Janabi

, Brown

, Bahrani

, Abner

, Barber

, Gold

, Goldstein

, Murphy

, Nelson

, Johnson

, Shaw

, Smith

, Trojanowski

, Wilcock

, Jicha

(2018) Distinct white matter changes associated with cerebrospinal fluid amyloid-β1-42 and hypertension. J Alzheimers Dis 66, 1095–1104.

21.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology 43, 2412–2414.

22.

Jenkinson

, Beckmann

, Behrens

TEJ

, Woolrich

, Smith

(2012) FSL. Neuroimage 62, 782–790.

23.

Andersson

JLR

, Sotiropoulos

(2016) An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage 125, 1063–1078.

24.

Smith

(2002) Fast robust automated brain extraction. Hum Brain Mapp 17, 143–155.

25.

Andersson

JLR

, Jenkinson

, Smith

(2007), Non-linear registration aka Spatial normalisation FMRIB Technical Report TR07JA2.

26.

Rueckert

, Sonoda

, Hayes

, Hill

DLG

, Leach

, Hawkes

(1999) Nonrigid registration using free-form deformations: Application to breast MR images. IEEE Trans Med Imaging 18, 712–721.

27.

Smith

, Jenkinson

, Johansen-Berg

, Rueckert

, Nichols

, Mackay

, Watkins

, Ciccarelli

, Cader

, Matthews

, Behrens

TEJ

(2006) Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data. Neuroimage 31, 1487–1505.

28.

Duits

, Martinez-Lage

, Paquet

, Engelborghs

, Lleo

, Hausner

, Molinuevo

, Stomrud

, Farotti

, Ramakers

, Tsolaki

, Skarsgard

, Astrand

, Wallin

, Vyhnalek

, Holmber-Clausen

, Forlenza

O V

, Ghezzi

, Ingelsson

, Hoff

, Roks

, De

, Papma

, Izagirre

, Taga

, Struyfs

, Alcolea

, Frolich

, Balasa

, Minthon

, Twisk

, Persson

, Zetterberg

, van der Flier

, Teunissen

, Scheltens

, Blennow

(2016) Performance and complications of lumbar puncture in memory clinics: Results of the multicenter lumbar puncture feasibility study. Alzheimers Dement 12, 154–163.

29.

Lleó

, Alcolea

, Martínez-Lage

, Scheltens

, Parnetti

, Poirier

, Simonsen

, Verbeek

, Rosa-Neto

, Slot

RER

, Tainta

, Izaguirre

, Reijs

BLR

, Farotti

, Tsolaki

, Vandenbergue

, Freund-Levi

, Verhey

FRJ

, Clarimón

, Fortea

, Frolich

, Santana

, Molinuevo

, Lehmann

, Visser

, Teunissen

, Zetterberg

, Blennow

(2019) Longitudinal cerebrospinal fluid biomarker trajectories along the Alzheimer’s disease continuum in the BIOMARKAPD study. Alzheimers Dement 15, 742–753.

30.

Dias

IHK

, Milic

, Lip

GYH

, Devitt

, Polidori

, Griffiths

(2018) Simvastatin reduces circulating oxysterol levels in men with hypercholesterolaemia. Redox Biol 16, 139–145.

31.

Blazquez

, De

, Ruiz-Martinez

, Emparanza

, Saenz

, Otaegui

, Sistiaga

, Martinez-Lage

, Lamet

, Samaranch

, Buiza

, Etxeberria

, Arriola

, Cuadrado

, Urdaneta

, Yanguas

, Lopez

de MA

(2007) Genes related to iron metabolism and susceptibility to Alzheimer’s disease in Basque population. Neurobiol Aging 28, 1941–1943.

32.

Fazekas

, Chawluk

, Alavi

, Hurtig

, Zimmerman

(1987) MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 149, 351–356.

33.

Smith

, Nichols

(2009) Threshold-free cluster enhancement: Addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 44, 83–98.

34.

Mayo

, Garcia-Barrera

, Mazerolle

, Ritchie

, Fisk

, Gawryluk

(2019) Relationship between DTI metrics and cognitive function in Alzheimer’s disease. Front Aging Neurosci 10, 436.

35.

Borghesani

, Madhyastha

, Aylward

, Reiter

, Swarny

, Schaie

, Willis

(2013) The association between higher order abilities, processing speed, and age are variably mediated by white matter integrity during typical aging. Neuropsychologia 51, 1435–1444.

36.

Zhuang

, Sachdev

, Trollor

, Reppermund

, Kochan

, Brodaty

, Wen

(2013) Microstructural white matter changes, not hippocampal atrophy, detect early amnestic mild cognitive impairment. PLoS One 8, e58887.

37.

Wang

, Fratiglioni

, Laukka

, Lovden

, Kalpouzos

, Keller

, Graff

, Salami

, Backman

, Qiu

(2015) Effects of vascular risk factors and APOE 4 on white matter integrity and cognitive decline. Neurology 84, 1128–1135.

38.

Mateos

, Ismail

MAM

, Gil-Bea

, Leoni

, Winblad

, Björkhem

, Cedazo-Mínguez

(2011) Upregulation of brain renin angiotensin system by 27-hydroxycholesterol in Alzheimer’s disease. J Alzheimers Dis 24, 669–679.

39.

Popp

, Lewczuk

, Kölsch

, Meichsner

, Maier

, Kornhuber

, Jessen

, Lütjohann

(2012) Cholesterol metabolism is associated with soluble amyloid precursor protein production in Alzheimer’s disease. J Neurochem 123, 310–316.

40.

Testa

, Staurenghi

, Zerbinati

, Gargiulo

, Iuliano

, Giaccone

, Fantò

, Poli

, Leonarduzzi

, Gamba

(2016) Changes in brain oxysterols at different stages of Alzheimer’s disease: Their involvement in neuroinflammation. Redox Biol 10, 24–33.

41.

Brown

, Jessup

(2009) Oxysterols: Sources, cellular storage and metabolism, and new insights into their roles in cholesterol homeostasis. Mol Aspects Med 30, 111–122.

42.

Phan

HTT

, Vestergaard

, Baek

, Shimokawa

, Takagi

(2014) Localization of amyloid beta (Aβ1-42) protofibrils in membrane lateral compartments: Effect of cholesterol and 7-Ketocholesterol. FEBS Lett 588, 3483–3490.

43.

Kim

, Frangos

(2008) Effects of amyloid β-peptides on the lysis tension of lipid bilayer vesicles containing oxysterols. Biophys J 95, 620–628.

44.

Phan

HTT

, Shimokawa

, Sharma

, Takagi

, Vestergaard

(2018) Strikingly different effects of cholesterol and 7-ketocholesterol on lipid bilayer-mediated aggregation of amyloid beta (1-42). Biochem Biophys Rep 14, 98–103.

45.

Zhao

, Long

, Mu

, Chew

(2012) The toxicity of amyloid β oligomers. Int J Mol Sci 13, 7303–7327.

46.

Winklewski

, Sabisz

, Naumczyk

, Jodzio

, Szurowska

, Szarmach

(2018) Understanding the physiopathology behind axial and radial diffusivity changes-what do we know? Front Neurol 9, 92.

47.

Bartzokis

, Lu

, Heydari

, Couvrette

, Lee

, Kalashyan

, Freeman

, Grinstead

, Villablanca

, Finn

, Mintz

, Alger

, Altshuler

(2012) Multimodal magnetic resonance imaging assessment of white matter aging trajectories over the lifespan of healthy individuals. Biol Psychiatry 72, 1026–1034.

48.

Benitez

, Fieremans

, Jensen

, Falangola

, Tabesh

, Ferris

, Helpern

(2014) White matter tract integrity metrics reflect the vulnerability of late-myelinating tracts in Alzheimer’s disease. Neuroimage Clin 4, 64–71.

49.

Braak

, Braak

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

50.

, Feng

, Lin

, Huang

Z-G

, Liu

, Zhou

, Yao

, Zheng

, Li

, Wang

, Zhang

, Guo

, Wang

, An

, Zhu

, Zhang

, Wang

(2018) Cognition-related white matter integrity dysfunction in Alzheimer’s disease with diffusion tensor image. Brain Res Bull 143, 207–216.

51.

Ringman

, O’Neill

, Geschwind

, Medina

, Apostolova

, Rodriguez

, Schaffer

, Varpetian

, Tseng

, Ortiz

, Fitten

, Cummings

, Bartzokis

(2007) Diffusion tensor imaging in preclinical and presymptomatic carriers of familial Alzheimer’s disease mutations. Brain 130, 1767–76.

52.

Khalil

, Teunissen

, Otto

, Piehl

, Sormani

, Gattringer

, Barro

, Kappos

, Comabella

, Fazekas

, Petzold

, Blennow

, Zetterberg

, Kuhle

(2018) Neurofilaments as biomarkers in neurological disorders. Nat Rev Neurol 14, 577–589.

53.

Song

S-K

, Sun

S-W

, Ramsbottom

, Chang

, Russell

, Cross

(2002) Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage 17, 1429–1436.

54.

Song

S-K

, Yoshino

, Le

, Lin

S-J

, Sun

S-W

, Cross

, Armstrong

(2005) Demyelination increases radial diffusivity in corpus callosum of mouse brain. Neuroimage 26, 132–140.

55.

Wheeler-Kingshott

CAM

, Cercignani

(2009) About axial and radial diffusivities. Magn Reson Med 61, 1255–1260.

56.

Sato

, Ishihara

, Nagayama

, Saiki

, Tatsuno

(2017) 7-ketocholesterol induces apoptosis of MC3T3-E1 cells associated with reactive oxygen species generation, endoplasmic reticulum stress and caspase-3/7 dependent pathway. Mol Genet Metab Rep 10, 56–60.

57.

Jones

, Knösche

, Turner

(2013) White matter integrity, fiber count, and other fallacies: The do’s and don’ts of diffusion MRI. Neuroimage 73, 239–254.