Cerebrospinal Fluid Metals and the Association with Cerebral Small Vessel Disease

Abstract

Background:

Brain metal homeostasis is essential for brain health, and deregulation can result in oxidative stress on the brain parenchyma.

Objective:

Our objective in this study was to focus on two hemorrhagic MRI manifestations of small vessel disease [cerebral microbleeds (CMBs) and cortical superficial siderosis (cSS)] and associations with cerebrospinal fluid (CSF) iron levels. In addition, we aimed to analyze CSF biomarkers for dementia and associations with CSF metal levels.

Methods:

This is a cross-sectional study of 196 patients who underwent memory clinic investigation, including brain MRI. CSF was collected and analyzed for metals, amyloid-β (Aβ) 42, total tau (T-tau), and phosphorylated tau (P-tau), and CSF/serum albumin ratios. Statistical analyses were performed using generalized linear models.

Results:

No significant difference was found between CSF metal levels across diagnostic groups. Higher iron and copper levels were associated with higher CSF levels of Aβ₄₂, T-tau, P-tau, and CSF/serum albumin ratios (p < 0.05). Zinc was associated with higher CSF/serum albumin ratios. There was no significant association between CMBs or cSS and CSF iron levels. An increase in CSF iron with the number of CMBs was seen in APOE ɛ4 carriers.

Conclusion:

CSF iron levels are elevated with cerebral microbleeds in APOE ɛ4 carriers, with no other association seen with hemorrhagic markers of small vessel disease. The association of elevated CSF iron and copper with tau could represent findings of increased neurodegeneration in these patients.

Keywords

Alzheimer’s disease cerebrospinal fluid cognitive aging dementia magnetic resonance imaging

INTRODUCTION

Iron has an essential role in the brain—it is an integral component of DNA synthesis, neurotransmitters, and myelin, and serves as a cofactor in neuronal oxidative metabolism [1, 2]. With aging, neuronal iron metabolism is thought to be deregulated, with iron deposition in the brain [1, 2]. Iron deposition has been suggested to cause neurodegeneration through inflammation and release of oxidative factors, and be a causative factor for cognitive impairment and dementia [1 –3]. Zinc and copper have similarly been proposed to be involved in amyloid-β (Aβ) 42 plaque formation, oxidative mechanisms, and neurodegeneration [4, 5].

Cerebral microbleeds (CMBs) and cortical superficial siderosis (cSS) are putative markers of cerebral small vessel disease (SVD), which is commonly associated with cognitive impairment in the elderly. CMBs represent foci of hemosiderin deposits in the brain parenchyma. cSS represents linear hemosiderin deposition along the surface of the cortex, leptomeninges, and within the subarachnoid space [6]. cSS has shown to be a sensitive marker for probable cerebral amyloid angiopathy (CAA), whereas CMBs in lobar locations are likely to be associated with CAA. Both markers are seen with susceptibility weighted imaging (SWI) or T2*-weighted imaging on magnetic resonance imaging (MRI), and have been suggested to lead to neurodegeneration and cognitive decline.

We hypothesize that brain iron deposition as manifested by CMBs and cSS is reflected in cerebrospinal fluid (CSF) iron concentrations, with higher iron levels being associated with these markers. We also hypothesize that increased CSF metal concentrations are associated with CSF biomarkers for plaque and tangle pathology (Aβ₄₂ and phosphorylated tau (P-tau), respectively), and neurodegeneration (total tau (T-tau)).

MATERIAL AND METHODS

Patients

This is a cross-sectional study of a memory clinic population including 196 patients (Alzheimer’s disease = 85; mild cognitive impairment = 72; subjective cognitive impairment = 32; vascular dementia = 7). All patients underwent memory clinic investigation, including detailed neuropsychological testing, blood tests, lumbar puncture with CSF analysis, as well as brain MRI. Diagnosis was made according to ICD-10 in multidisciplinary teams, after consideration of all clinical data. Patients were enrolled from January 2006 to January 2012, and were included in the study if they had CSF samples saved in a biobank for study analysis. Our exclusion criterion included suboptimal quality of the MRI scan (e.g., due to patient motion); however, we did not exclude any patients in our final cohort. Informed consent was obtained from each patient, and study approval was obtained from the regional ethics review board at the Karolinska Institutet (protocol nr 2015/686-32).

MRI protocol and image analysis

All brain scans were performed on a 1.5T or 3T MRI scanner, with random assignment of scanner type based on clinical availability. Patients had a full MRI protocol, including hemosiderin sensitive MRI sequences such as T2* and/or SWI. For each patient routine imaging sequences such as T1, T2, FLAIR, and DWI were also obtained. MRI parameters have previously been described [7]. Image analysis was performed in accordance with the STRIVE criteria [8], by a neuroradiologist and an experienced MD, PhD as previously described [7, 9].

CSF analysis and apolipoprotein E

Initial CSF analysis was done as previously described [10] and included routine CSF biomarker analysis: Aβ₄₂, T-tau, P-tau, and CSF/serum albumin ratios. Patients included in this study had their CSF samples saved in a biobank, and the samples were subsequently sent for additional analysis, focusing on the measurement of iron-related markers and iron concentration in the CSF. Moreover, levels of an extended panel of metals including chromium, manganese, copper, cobalt, nickel and zinc were also obtained. CSF samples with blood contamination were excluded from analysis, with no CSF samples having to be excluded in this subset of patients. Apolipoprotein E (APOE) genotyping was done on a sub-cohort of patients (n = 102), on blood samples.

CSF metal concentrations

Metals in CSF were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) with an octopole reaction system (Agilent 7700x ICP-MS; Agilent Technologies, Santa Clara, Ca, USA). The reaction system was operated in the helium collision mode to eliminate interference from isobaric polyatomic species via kinetic energy discrimination. All samples were diluted 10 times with a basic diluent containing 1-butanol (2% w/v), ethylenediaminetetraacetic acid (EDTA) (0.05% w/v), TritonX-100 (0.05% w/v), and ammonium hydroxide (1% w/v). Germanium was used as internal. All samples were analyzed in batches with external quality control (QC) samples included. Two QC samples were used (Seronorm™ Trace Elements Urine L-1, Lot no. 1011644, and Seronorm™ Trace Elements Urine L-2, Lot No. 1011645; Sero AS, Billingstad, Norway). All element concentrations were within the stated acceptable limits. Cobalt was, however, below the lower limit of detection for all samples.

Statistical analysis

SPSS 24.0 was used for all statistical analyses. CSF metal concentrations and CSF biomarkers were log transformed when appropriate to obtain normality. Analysis of variance (ANOVA) was used to assess differences in CSF metal concentrations amongst groups. Simple linear regression models were built to assess associations between CSF metals and CSF biomarkers with results presented as the standardized regression coefficient β. Clinical parameters and CSF metal associations were tested in negative binomial regression models (results presented with the unstandardized regression coefficient B) as well as binary logistic regression models. Negative binomial regression models were built, adding an interaction between APOE ɛ4 and iron concentrations. Multivariable regressions were performed adjusting for age, gender, diagnosis, MRI field strength, and hemosiderin sequence. The threshold for statistical significance was set as p < 0.05. After accounting for multiple testing according to the Benjamini-Hochberg procedure, significance was set as p < 0.02[11].

Data availability statement

Anonymized data will be shared by request from any qualified investigator.

RESULTS

Demographic information from 196 patients included in our study is presented in Table 1. CSF concentrations of iron, manganese, nickel, copper, and zinc did not significantly differ between diagnostic groups (Table 1). CSF chromium levels were higher in Alzheimer’s disease (AD) than subjective cognitive impairment (SCI) (p = 0.003). Mild cognitive impairment (MCI) was associated with higher chromium levels than SCI (p < 0.001) and vascular dementia (VaD) (p = 0.012).

Table 1

Patient demographics and metal concentrations in CSF

	AD (n = 85)	MCI (n = 72)	VaD (n = 7)	SCI (n = 32)
Age (y) mean±SD	66±8	63±8	65±11	57±8
Gender, female/male %	55/45	33/67	57/43	53/47
MMSE (score 0–30), mean±SD	23±5	27±4	24±4	28±2
CMB, n (%)	28 (33)	27 (38)	4 (57)	14 (44)
Lobar CMBs, n (%)	25 (29)	16 (22)	0	16 (50)
Deep CMBs, n(%)	8 (9)	4 (6)	1 (14)	2 (6)
cSS, n (%)	8 (9)	3 (4)	0	0
CSF iron (μg/L), median (IQR)	16 (13–19)	16 (13–20)	17 (13–17)	16 (14–19)
CSF chromium (μg/L), median (IQR)	0.9 (0.8–1.1)^a	0.9 (0.7–1.1)^b	0.5 (0.5–0.5)^b	0.6 (0.5–0.7)^a,b
CSF manganese (μg/L), median (IQR)	0.4 (0.4–0.5)	0.4 (0.3–0.5)	0.3 (0.3–0.4)	0.4 (0.4–0.5)
CSF nickel (μg/L), median (IQR)	0.2 (0.1–0.3)	0.2 (0.2–0.5)	–	–
CSF copper (μg/L), median (IQR)	10.7 (9.2–12.4)	10.9 (9.6–12.4)	10.6 (8.4–15.3)	12.0 (10.9–12.6)
CSF zinc (μg/L), median (IQR)	12.6 (11.0–16.1)	13.5 (11.3–17.2)	11.8 (9.2–13.8)	13.3 (11.7–15.4)
Aβ₄₂ (ng/L), median (IQR)	482 (414–586)	732 (500–976)	994 (618–1170)	1115 (974–1288)
T–tau (ng/L), median (IQR)	508 (385–749)	262 (188–382)	216 (154–252)	228 (178–321)
P-tau (ng/L), median (IQR)	83 (65–99)	56 (38–77)	35 (26–53)	51 (41–62)
CSF/Serum albumin ratio, median (IQR)	244 (193–328)	281 (210–322)	310 (190–430)	275 (214–338)
APOEɛ2 carrier, n (%)	1 (2)	4 (13)	0	1 (4)
APOEɛ4 carrier, n (%)	26 (62)	21 (66)	0	1 (11)

^ap < 0.05 between AD and SCI. ^bp < 0.05 between MCI and SCI, and MCI and VaD. Three patients in total had disseminated cSS. Values for Nickel for VaD and SCI were below the lower limit for quantification, why no results are displayed. AD, Alzheimer’s disease; CMB, cerebral microbleed; CSF, cerebrospinal fluid; cSS, cortical superficial siderosis; IQR, interquartile range; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; SCI, subjective cognitive impairment; VaD, vascular dementia.

Exploring the relationship of CSF metals with CSF biomarkers for AD pathology showed an association between Aβ₄₂ and higher iron and copper levels in CSF (Table 2; Fig. 1). Higher levels of iron and copper were associated with higher levels of T-tau. Higher levels of CSF iron and copper were associated with higher levels of P-tau. A disrupted integrity of the blood-brain barrier, as reflected by the CSF/serum albumin ratio, was associated with higher levels of iron, copper and zinc deposition in the brain. No associations were seen between CSF metals and Mini-Mental State Examination score or hypertension. Chromium was the only metal found to be associated with increasing age. CSF AD biomarkers and iron, copper, and zinc were further analyzed in detail in the diagnostic groups (Table 3). Increased Aβ₄₂ levels were associated with increased level of CSF iron and copper in AD and MCI, and only with zinc in AD. T-tau increased with increasing CSF iron and copper concentrations in MCI and SCI. Similarly, P-tau increased with increasing iron concentrations in MCI and SCI. P-tau also increased with increasing copper concentrations in AD and MCI. Increased CSF/serum albumin ratios were associated with increased CSF iron and copper concentrations in AD, MCI, and SCI, as well as zinc in AD.

Table 2

CSF metals and associations with clinical parameters

Clinical Parameters	Chromium	Manganese	Iron	Nickel	Copper	Zinc
Aβ₄₂	–0.08, p = 0.27	–0.09, p = 0.23	0.21, p = 0.004	0.05, p = 0.78	0.23, p = 0.001	0.10, p = 0.15
T-Tau	0.13, p = 0.06	0.13, p = 0.08	0.27 p < 0.001	0.07, p = 0.68	0.23, p = 0.001	0.08, p = 0.3
P-Tau	0.11, p = 0.13	0.13, p = 0.07	0.30, p < 0.001	0.03, p = 0.86	0.26, p < 0.001	0.08, p = 0.28
T-Tau/Aβ₄₂ ratio	0.15, p = 0.06	0.33, p = 0.32	0.411, p = 0.06	0.01, p = 0.944	–0.001, p = 0.922
CSF/serum albumin ratio	0.03, p = 0.65	–0.12, p = 0.09	0.52, p < 0.001	0.12, p = 0.49	0.65, p < 0.001	0.17, p = 0.02
MMSE	0.012, p = 0.06	0.005, p = 0.74	–0.001, p = 0.63	0.014, p = 0.12	–0.001, p = 0.16	–0.001, p = 0.21
Age	1.661, p < 0.001	0.001, p = 0.985	0.006, p = 0.673	0.041, p = 0.151	0.070, p = 0.761	0.002, p = 0.887
Hypertension	0.01, p = 0.05	–0.007, p = 0.55	0.009, p = 0.75	0.024, p = 0.11	0.003, p = 0.70	0.005, p = 0.85

Values are given as the standardized regression coefficient β for Aβ₄₂, T-tau, P-tau and CSF/serum albumin ratio and the non-standardized regression coefficient for MMSE (Mini-Mental State Examination), Age, and Hypertension. Cobalt was below the lower limit for quantification, and thus no results are displayed.

Fig.1

CSF iron levels and associations with CSF biomarkers.

A subset of patients (n = 102) had APOE genotyping done (Table 4A, B). Findings showed an association (p < 0.05) between high copper levels and Aβ₄₂, high T-tau and iron, high P-tau and iron and copper. CSF/serum albumin ratio was decreased with manganese and increased with iron and copper.

Table 3

CSF iron, copper, and zinc levels within the separate diagnostic groups

Iron	AD	MCI	SCI
Aβ₄₂	0.31, p = 0.005	0.27, p = 0.02	0.16, p = 0.39
T-tau	0.15, p = 0.17	0.46, p < 0.001	0.57, p = 0.001
P-tau	0.21, p = 0.060	0.45, p < 0.001	0.36, p = 0.045
CSF/serum albumin ratio	0.58, p < 0.001	0.46, p < 0.001	0.38, p = 0.03
Copper
Aβ₄₂	0.26, p = 0.02	0.29, p = 0.01	0.14, p = 0.45
T-tau	0.18, p = 0.10	0.42, p < 0.001	0.44, p = 0.01
P-tau	0.27, p = 0.01	0.37, p = 0.001	0.17, p = 0.36
CSF/serum albumin ratio	0.63, p < 0.001	0.64, p < 0.001	0.65, p < 0.001
Zinc
Aβ₄₂	0.40, p < 0.001	–0.01, p = 1.0	0.19, p = 0.30
T-tau	–0.02, p = 0.83	0.19, p = 0.11	0.37, p = 0.04
P-tau	0.04, p = 0.74	0.15, p = 0.21	–0.001, p = 1.0
CSF/serum albumin ratio	0.31, p = 0.005	0.009, p = 0.9	0.30, p = 0.09

Values are given as the standardized regression coefficient β. AD, Alzheimer’s disease; MCI, mild cognitive impairment; SCI, subjective cognitive impairment.

Table 4A

CSF metal levels and biomarkers in patients with the APOE allele genotyping test performed (n = 102)

Clinical Parameters	Chromium	Manganese	Iron	Nickel	Copper	Zinc
Aβ₄₂	–0.03, p = 0.51	–0.08, p = 0.63	0.005, p = 0.245	–0.06, p = 0.89	0.14, p = 0.04	0.005, p = 0.12
T-Tau	0.01, p = 0.14	0.25, p = 0.26	0.01 p = 0.04	0.04, p = 0.31	0.015, p = 0.12	0.04, p = 0.3
P-Tau	0.03, p = 0.452	0.20, p = 0.16	0.01, p = 0.009	0.03, p = 0.22	0.01, p = 0.03	0.003, p = 0.24
T-Tau/Aβ₄₂ ratio	0.13, p = 0.19	0.33, p = 0.32	0.007, p = 0.434	0.05, p = 0.52	0.001, p = 0.944	–0.001, p = 0.922
CSF/serum albumin ratio	–0.03, p = 0.44	–0.36, p = 0.003	0.02, p < 0.001	0.03, p = 0.37	0.04, p < 0.001	0.004, p = 0.140

Values are given as the standardized regression coefficient β for Aβ₄₂, T-tau, P-tau, and CSF/serum albumin ratios.

Table 4B

Concentration of CSF metal levels in a subset of patients with the APOE allele genotyping test performed

CSF Metals
CSF Iron (μg/L), median (IQR)	16 (13–19)
CSF Chromium (μg/L), median (IQR)	0.8 (0.6–1.1)
CSF Manganese (μg/L), median (IQR)	0.4 (0.4–0.5)
CSF Nickel (μg/L), median (IQR)	0.3 (0.2–0.6)
CSF Copper (μg/L), median (IQR)	10.9 (9.4–12.4)
CSF Zinc (μg/L), median (IQR)	12.8 (11.0–16.0)

IQR, interquartile range.

CMBs did not show any association with CSF iron levels (Table 5). However, when looking at interactions with APOE ɛ4, carriers had a higher CSF iron concentration associated with increasing number of CMBs (Table 5). cSS showed no association with CSF iron levels in the whole cohort or in the diagnostic groups even when adding interactions with APOE ɛ4. There was no difference in CSF iron concentrations between patients with and without probable CAA, classified according to CMB location.

Table 5

Cerebral microbleeds and associations with iron concentration

	Whole cohort	AD	MCI	SCI
Presence of CMBs	0.14 p = 0.09	0.14, p = 0.21	0.20, p = 0.09	–0.06, p = 0.76
Number of CMBs	0.12 p = 0.09	0.15, p = 0.16	0.12, p = 0.30	–0.04, p = 0.82
Number of CMBs if APOEɛ4 carrier	0.12, p = 0.01	0.06, P = 0.14	0.32, p < 0.001	–0.09, p = 0.54

Values are given as the regression coefficient B. AD, Alzheimer’s disease; MCI, mild cognitive impairment; SCI, subjective cognitive impairment.

For all analyses results did not change when adjusting for age, gender, diagnosis, field strength, and MRI sequence.

DISCUSSION

Our results show higher iron and copper levels with CSF markers of neurodegeneration (T-tau), neurofibrillary tangles (P-Tau), and an association with a disrupted blood-brain barrier. Metals were associated with less amyloid deposition in the brain, as reflected by higher CSF Aβ₄₂ levels. This suggests the impact, and possible role of metals in neurodegenerative disorders. Further, increased iron levels were associated with increased CMBs in APOE ɛ4 carriers.

Metals hold an important role in several important biological processes and regulated metal homeostasis is of the essence. In the brain, iron contributes to cell growth, enzymatic activities, immune responses, and oxygenation [12], and other metals have equally essential roles.

Metal dyshomeostasis has been hypothesized to cause damage, among others, through oxidative stress, interaction with DNA, proteins, and other macromolecules [13]. Oxidative stress has been proposed to be one of the earliest mechanisms on the path to MCI and AD [14, 15]. The creation of reactive oxygen species may lead to interaction with cell signaling, DNA disruption as well as cell apoptosis, and neuronal damage [13]. Iron and copper are also highly involved in Aβ₄₂ aggregation in AD and may thus further contribute to the disease process [16]. Copper has equally shown to cause reactive oxygen species, and is thought to interact with the amyloid-β protein precursor, promoting the production of Aβ₄₂ plaques and neurofibrillary tangles in the brain [14].

Prior studies on CSF metals and neurodegeneration are scarce, and especially studies investigating the inter-relation with SVD was non-existent to date. SVD has been hypothesized to cause dementia, and the fragile, leaking microvasculature leading to CMBs representing an additional source of hemosiderin in the brain parenchyma. CMBs have shown to be associated with amyloid brain deposition [17], supporting the intriguing theory that the CMB in itself may trigger Aβ₄₂ plaque formation.

Iron has been widely studied in neurodegeneration. Cortical iron deposition has been linked with AD-type pathology [18], which has been shown not to be linked to presence of microbleeds [19]. In a case report, superficial CNS siderosis with iron toxicity was associated with increased CSF T-tau and P-tau concentrations, which may reflect Alzheimer-like neurodegeneration, while CSF Aβ₄₂ was normal (i.e., high in the CSF), suggesting no cerebral β-amyloidosis [20]. These results are similar to ours, and additional analysis of Aβ₄₀ may provide further insight in the underlying pathophysiology. MCI and AD have been shown to have a higher load of iron in the brain parenchyma on histopathology [14]. Similar to our study high levels of tau has been seen with iron deposition in the brain on PET/MRI [21]. Hippocampal atrophy rate has shown to be higher in patients with higher CSF ferritin levels [22]. CSF ferritin levels have also shown to be associated with cognition, as well as prediction of conversion from MCI to AD [22]. APOE ɛ4 has shown to be associated with higher CSF ferritin levels, indicating a possible pathway of pathology of the allele [22]. In our study, the association between iron and dementia was seen in the strong association between high CSF iron levels and increased T-tau and P-tau, as well as a disrupted blood-brain barrier, reflected by the CSF/serum albumin ratio. The association between iron and high CSF Aβ₄₂, reflecting less amyloid deposition in the brain is elusive. Findings may reflect the fact that oxidation is not associated with amyloid pathology, but rather with neurodegeneration and neurofibrillary tangles, as reflected by higher T-tau and P-tau levels in the CSF. Aβ₄₀ levels would be helpful in further analyzing this association, with presumably Aβ₄₀ being lower with increased CSF iron. CSF iron levels increased with number of CMBs in APOE ɛ4 carriers.

Free copper has also shown to be higher in patients with AD, and to be able to predict conversion from MCI to AD [23]. CSF copper levels have shown to be higher in patients with AD, and especially with late onset AD when compared with controls [5]. The importance of copper is supported in our study with the associations to biomarkers for neurofibrillary tangles, neurodegeneration, and blood-brain barrier disruption. However, similar to iron, copper was associated with higher Aβ₄₂ levels in the CSF.

Aβ₄₂ plaques has shown to be enriched with zinc and copper [4 , 24]. The Aβ₄₂ forming process is thought to deplete neurons from their intracellular zinc storage, leading to microtubule disarray as well as formation of neurofibrillary tangles [4, 14]. CSF zinc has shown to be higher with late onset AD, albeit in a cohort with only 14 participants [5]. A higher level of CSF zinc was associated with blood-brain barrier disruption as well as higher Aβ₄₂ in AD in our study. The association with Aβ₄₂ may once again indicate that the pathomechanistic effects of zinc take place in an early stage, and are not specific to AD. Associations between high CSF copper and zinc levels with low levels of Aβ₄₂ has previously been demonstrated [24], the opposite of our results. Zinc dysregulation is thought to occur in AD leading to Aβ₄₂ plaque buildup.

Other metals including chromium, manganese, cobalt, and nickel are scarcely studied in neurodegeneration. In a similar study to ours on CSF metals in AD, cobalt showed to be the only metal that was higher in AD than in healthy controls, and this held true when only looking at a sub-cohort with severe AD [25]. Although we were not able to take CSF cobalt concentrations into account in our study, no significant associations between the other metals mentioned above and biomarkers of dementia were seen in our study.

Limitations of our study include the fact that CSF metal levels may not accurately depict that of the brain parenchyma [26]; however, our analysis may reflect the condition of the brain parenchyma, and postmortem analysis may not always reflect that of the living brain. Quantitative susceptibility mapping provides an additional way of in vivo iron quantification, and will be used in our future studies. The lack of a strict control group also poses as a limitation and would have helped to reinforce the findings in our patient population. The subjective cognitive impairment group can be argued to function as a control group, however a strict control group would have been better for comparison of CSF levels between diagnoses. Additional limitations include that of varying field strength (1.5 and 3.0T) and hemosiderin sequences (SWI and T2*); however, findings did not change when controlling for these variables. Our analyses and results are also only applicable in a cohort with cognitive impairment, as it has not been explored whether findings would be different in subjects without cognitive impairment in the very early stages of disease. Strengths of our study include the use of a wide array of metal analysis in a large cohort, and neuroimaging analysis.

Conclusions

CSF iron levels are elevated with cerebral microbleeds in Apolipoprotein E ɛ4 carriers, with no other association seen with hemorrhagic markers of small vessel disease. The association of elevated CSF iron and copper with tau could represent findings of increased neurodegeneration in these patients.

Footnotes

ACKNOWLEDGMENTS

Funding for this work was provided by Stockholm County Council, Karolinska Institutet, and the Swedish Dementia Association.

Authors’ disclosures available online ().

References

Acosta-Cabronero

, Betts

, Cardenas-Blanco

, Yang

, Nestor

(2016) In vivo MRI mapping of brain iron deposition across the adult lifespan. J Neurosci 36, 364–374.

Daugherty

, Haacke

, Raz

(2015) Striatal iron content predicts its shrinkage and changes in verbal working memory after two years in healthy adults. J Neurosci 35, 6731–6743.

Penke

, Valdés Hernandéz

, Maniega

, Gow

, Murray

, Starr

, Bastin

, Deary

, Wardlaw

(2012) Brain iron deposits are associated with general cognitive ability and cognitive aging. Neurobiol Aging 33, 510–517.e2.

Craddock

TJA

, Tuszynski

, Chopra

, Casey

, Goldstein

, Hameroff

, Tanzi

(2012) The zinc dyshomeostasis hypothesis of Alzheimer’s disease. PLoS One 7, e33552.

Hozumi

, Hasegawa

, Honda

, Ozawa

, Hayashi

, Hashimoto

, Yamada

, Koumura

, Sakurai

, Kimura

, Tanaka

, Satoh

, Inuzuka

(2011) Patterns of levels of biological metals in CSF differ among neurodegenerative diseases. J Neurol Sci 303, 95–99.

Charidimou

, Linn

, Vernooij

, Opherk

, Akoudad

, Baron

J-C

, Greenberg

, Jäger

, Werring

(2015) Cortical superficial siderosis: Detection and clinical significance in cerebral amyloid angiopathy and related conditions. Brain 138, 2126–2139.

Shams

, Martola

, Granberg

, Li

, Shams

, Fereshtehnejad

, Cavallin

, Aspelin

, Kristoffersen-Wiberg

, Wahlund

(2015) Cerebral microbleeds: Different prevalence, topography, and risk factors depending on dementia diagnosis—The Karolinska Imaging Dementia Study. Am J Neuroradiol 36, 661–666.

Wardlaw

, Smith

, Biessels

, Cordonnier

, Fazekas

, Frayne

, Lindley

, O’Brien

, Barkhof

, Benavente

, Black

, Brayne

, Breteler

, Chabriat

, Decarli

, de Leeuw

F-E

, Doubal

, Duering

, Fox

, Greenberg

, Hachinski

, Kilimann

, Mok

, Oostenbrugge

R van

, Pantoni

, Speck

, Stephan

BCM

, Teipel

, Viswanathan

, Werring

, Chen

, Smith

, van Buchem

, Norrving

, Gorelick

, Dichgans

, STandards for ReportIng Vascular changes on nEuroimaging (STRIVE v1) (2013) Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822–838.

Shams

, Martola

, Charidimou

, Cavallin

, Granberg

, Shams

, Forslin

, Aspelin

, Kristoffersen-Wiberg

, Wahlund

L-O

(2016) Cortical superficial siderosis: Prevalence and biomarker profile in a memory clinic population. Neurology 87, 1110–1117.

10.

Shams

, Granberg

, Martola

, Li

, Shams

, Fereshtehnejad

S-M

, Cavallin

, Aspelin

, Kristoffersen-Wiberg

, Wahlund

L-O

(2016) Cerebrospinal fluid profiles with increasing number of cerebral microbleeds in a continuum of cognitive impairment. J Cereb Blood Flow Metab 36, 621–628.

11.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 57, 289–300.

12.

Jomova

, Valko

(2011) Advances in metal-induced oxidative stress and human disease. Toxicology 283, 65–87.

13.

Cicero

, Mostile

, Vasta

, Rapisarda

, Signorelli

, Ferrante

, Zappia

, Nicoletti

(2017) Metals and neurodegenerative diseases. A systematic review. Environ Res 159, 82–94.

14.

Smith

, Zhu

, Tabaton

, Liu

, McKeel

Jr , Cohen

, Wang

, Siedlak

, Dwyer

, Hayashi

, Nakamura

, Nunomura

, Perry

(2010) Increased iron and free radical generation in preclinical Alzheimer disease and mild cognitive impairment. J Alzheimers Dis 19, 363–372.

15.

Smith

(2006) Oxidative stress and iron imbalance in Alzheimer disease: How rust became the fuss! J Alzheimers Dis 9, 305–308.

16.

Mattson

(2004) Pathways towards and away from Alzheimer’s disease. Nature 430, 631–639.

17.

Park

J-H

, Seo

, Kim

, Noh

, Kim

, Kwak

K-C

, Yoon

, Lee

, Shin

, Kim

, Noh

, Cho

, Kim

, Yoon

, Oh

, Kim

, Choe

, Lee

K-H

, Lee

J-H

, Ewers

, Weiner

, Werring

, Na

(2013) Pathogenesis of cerebral microbleeds: In vivo imaging of amyloid and subcortical ischemic small vessel disease in 226 individuals with cognitive impairment. Ann Neurol 73, 584–593.

18.

Ayton

, Wang

, Diouf

, Schneider

, Brockman

, Morris

, Bush

(2020) Brain iron is associated with accelerated cognitive decline in people with Alzheimer pathology. Mol Psychiatry 25, 2932–2941.

19.

Fazlollahi

, Raniga

, Bourgeat

, Yates

, Bush

, Salvado

, Ayton

(2020) Restricted effect of cerebral microbleeds on regional magnetic susceptibility. J Alzheimers Dis 76, 571–577.

20.

Kondziella

, Zetterberg

(2008) Hyperphosphorylation of tau protein in superficial CNS siderosis. J Neurol Sci 273, 130–132.

21.

Spotorno

, Acosta-Cabronero

, Stomrud

, Lampinen

, Strandberg

, van Westen

, Hansson

(2020) Relationship between cortical iron and tau aggregation in Alzheimer’s disease. Brain 143, 1341–1349.

22.

Ayton

, Faux

, Bush

(2015) Ferritin levels in the cerebrospinal fluid predict Alzheimer’s disease outcomes and are regulated by APOE. Nat Commun 6, 6760.

23.

Squitti

, Ghidoni

, Siotto

, Ventriglia

, Benussi

, Paterlini

, Magri

, Binetti

, Cassetta

, Caprara

, Vernieri

, Rossini

, Pasqualetti

(2014) Value of serum nonceruloplasmin copper for prediction of mild cognitive impairment conversion to Alzheimer disease. Ann Neurol 75, 574–580.

24.

Strozyk

, Launer

, Adlard

, Cherny

, Tsatsanis

, Volitakis

, Blennow

, Petrovitch

, White

, Bush

(2009) Zinc and copper modulate Alzheimer Aβ levels in human cerebrospinal fluid. Neurobiol Aging 30, 1069–1077.

25.

Gerhardsson

, Lundh

, Londos

, Minthon

(2011) Cerebrospinal fluid/plasma quotients of essential and non-essential metals in patients with Alzheimer’s disease. J Neural Transm 118, 957–962.

26.

Szabo

, Harry

, Hayden

, Szabo

, Birnbaum

(2016) Comparison of metal levels between postmortem brain and ventricular fluid in Alzheimer’s disease and nondemented elderly controls. Toxicol Sci 150, 292–300.