Directed Functional Brain Connectivity is Altered in Sub-threshold Amyloid-β Accumulation in Cognitively Normal Individuals

Abstract

Several studies have shown that amyloid-β (Aβ) deposition below the clinically relevant cut-off levels is associated with subtle changes in cognitive function and increases the risk of developing future Alzheimer’s disease (AD). Although functional MRI is sensitive to early alterations occurring during AD, sub-threshold changes in Aβ levels have not been linked to functional connectivity measures. This study aimed to apply directed functional connectivity to identify early changes in network function in cognitively unimpaired participants who, at baseline, exhibit Aβ accumulation below the clinically relevant threshold. To this end, we analyzed baseline functional MRI data from 113 cognitively unimpaired participants of the Alzheimer’s Disease Neuroimaging Initiative cohort who underwent at least one ¹⁸F-florbetapir-PET after the baseline scan. Using the longitudinal PET data, we classified these participants as Aβ negative (Aβ−) non-accumulators (n = 46) and Aβ− accumulators (n = 31). We also included 36 individuals who were amyloid-positive (Aβ+) at baseline and continued to accumulate Aβ (Aβ+ accumulators). For each participant, we calculated whole-brain directed functional connectivity networks using our own anti-symmetric correlation method and evaluated their global and nodal properties using measures of network segregation (clustering coefficient) and integration (global efficiency). When compared to Aβ− non-accumulators, the Aβ− accumulators showed lower global clustering coefficient. Moreover, the Aβ+ accumulator group exhibited reduced global efficiency and clustering coefficient, which at the nodal level mainly affected the superior frontal gyrus, anterior cingulate cortex, and caudate nucleus. In Aβ− accumulators, global measures were associated with lower baseline regional PET uptake values, as well as higher scores on the Modified Preclinical Alzheimer Cognitive Composite. Our findings indicate that directed connectivity network properties are sensitive to subtle changes occurring in individuals who have not yet reached the threshold for Aβ positivity, which makes them a potentially viable marker to detect negative downstream effects of very early Aβ pathology.

Keywords

Functional MRI directed functional connectivity anti-symmetric correlations amyloid-β positron emission tomography preclinical AD longitudinal amyloid-β accumulation

Significance Statement

Directed functional connectivity measures are sensitive to very early changes in amyloid-β deposition.

Amyloid-β burden is associated with decreases in global and nodal-directed measures of network integration and segregation.

Measures of directed network integration and segregation are associated with subclinical cognitive changes in Aβ-negative accumulators.

Introduction

Alzheimer’s disease (AD) is a neurodegenerative disorder that results in a progressive loss of memory and other cognitive functions.¹ However, the pathological processes in AD begin many years before any symptoms of dementia and other cognitive impairments become apparent.² In particular, Aβ accumulation, which can be measured by an increased burden in amyloid positron emission tomography (PET), is an early sign of preclinical AD stages, being one of the first events in the sequence of pathological processes that ultimately results in the development of dementia.^3,4

Although Aβ plaques are a hallmark of AD, studies using amyloid PET imaging found they are also present in the brains of otherwise healthy older individuals who show no indications of cognitive impairment.^5,6 The deposition of Aβ in such individuals occurs over a very long time period⁷ and results in changes in the functional connectivity patterns.⁸ However, widespread Aβ changes may not be necessary for observing connectivity changes.⁹ Instead, such effects can be evident also in the presence of emergent Aβ pathology,⁹ the process of amyloid accumulation in cognitively healthy people who have not yet reached the threshold for clinical Aβ positivity. Such emergent pathology has been associated with higher tau accumulation and atrophy rates, as well as memory impairment and alterations in functional connectivity in previous studies.^10-14

Here, we assessed whether early Aβ accumulation in otherwise healthy, cognitively normal individuals that are Aβ-negative at baseline has an effect on the organization of whole-brain directed functional connectivity networks. The directed functional networks were calculated using the method of anti-symmetric delayed correlations that we developed earlier,¹⁵ since this method is more sensitive to detect subtle changes in functional connectivity when compared to conventional methods. Our results show that early amyloid deposition is associated with a decrease of segregation in functional networks. These changes were also observed, although to a greater extent, in individuals that were already Aβ+ at baseline, who additionally showed changes in functional integration. At the regional level, the most significant changes in connectivity patterns were observed in the superior frontal gyrus, caudate nucleus, and anterior cingulate cortex. Finally, the global network properties were associated with the extent of amyloid burden and preclinical cognitive composite scores. These findings suggest that, even at sub-threshold Aβ deposition levels, the decrease of directional flow in the functional network could be an indication of neuronal changes and subtle changes in cognition in individuals at the preclinical stages of AD.

Materials and Methods

Participants

Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), PET, other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early AD. ADNI is conducted in accordance with the ethical standards of the Institutional Research Committees and with the 1975 Helsinki declaration and its later amendments. Written informed consent, obtained from all subjects and/or authorized representatives and study partners, and ethical permits have been obtained at each participating site of ADNI and we have signed the data user agreements to analyze the data.

For the purposes of this study, we included only cognitively normal participants from ADNI3 with a baseline functional MRI scan, performed within a year of a PET scan for the majority of the participants, and at least two ¹⁸F-Florbetapir PET scans. The inclusion/exclusion criteria for ADNI are described in detail at http://www.adni-info.org/. In short, participants were aged between 55 and 90 years, were fluent in Spanish or English while having also completed at least 6 years of education, had Clinical Dementia Rating scores of 0 and no other significant neurological disorder. We used the Preclinical Alzheimer Cognitive Composite¹⁶ (PACC, with trials test) to assess the cognitive status of the individuals. This composite calculates a composite score based on tests that assess global cognition, episodic memory and executive function and has been shown to be sensitive to the very initial signs of cognitive decline, before the appearance of clinical symptoms of mild cognitive impairment stage.

Image acquisition and preprocessing

Subjects in the ADNI3 study underwent standardized MRI scanning protocols at the acquisition sites using 3T MRI scanners from different vendors, and ¹⁸F-florbetapir PET scans. The full list of MRI/PET scanners and detailed imaging protocols can be found at the ADNI study website (https://adni.loni.usc.edu/). From the full suite of imaging data, we used the sagittal 3D accelerated MPRAGE sequence (full head coverage, voxel size = 1 mm × 1 mm × 1 mm, field of view = 208 mm × 240 mm × 256 mm, repetition time = 2300 ms, inversion time = 900 ms) and the axial EPI-BOLD functional MRI sequence (voxel size = 3.4 mm × 3.4 mm × 3.4 mm, field of view = 220 mm × 220 mm × 163 mm, repetition time = 3000 ms, echo time = 30 ms, flip angle = 90°). Functional and structural MRI scans were pre-processed using a standardized pipeline implemented in fMRIPrep¹⁷ (v20.2.4, https://fmriprep.org/en/stable/). The resulting functional images additionally underwent motion correction using the Friston et al¹⁸ -24 head motion model and nuisance regression for signal from the white matter and cerebrospinal fluid. ¹⁸F-florbetapir PET scans were acquired in 4 minutes × 5 minutes frames, 50 to 70 minutes after the injection of 10 mCi dose on average.

Group classification

We divided the participants into 3 groups based on Aβ levels measured by ¹⁸F-Florbetapir PET. Using their PET scan closest to the functional MRI scan as a reference, we classified the participants as being Aβ-negative or Aβ-positive using the standardized uptake value ratios (SUVRs) for a global composite region comprising the caudal anterior cingulate, frontal, lateral parietal and lateral temporal gyri on ¹⁸F-Florbetapir PET, normalized by the whole cerebellum, that has been earlier used for cross-sectional analysis.¹⁹ A previously defined cut-off was used to determine amyloid-positivity.²⁰ Since normalizing the SUVR values by cerebellar regions could result in noise in the longitudinal measurements,²¹ we calculated the rate of change in SUVR using the longitudinal SUVR composite values, which were normalized by a composite reference region composed of the whole cerebellum, brainstem/pons and eroded subcortical white matter. Based on these values, we further subdivided the Aβ-negative individuals into two groups that showed either positive (accumulators) annual rate of change in their Aβ levels (ie, the change in the SUVR values per year) or remained largely stable with a small negative rate of change in Aβ levels (non-accumulators). While the negative slopes could represent measurement noise, some studies have suggested that there is a possibility that they are due to Aβ clearance.¹⁴ This procedure resulted in 3 groups of participants: 31 Aβ-negative non-accumulators (Aβ−/NAc); 46 Aβ-negative accumulators (Aβ−/Ac); and 36 Aβ-positive (Aβ+). The rate of change in Aβ was calculated using the 4 PET scans that the participants had on average, with the Aβ− non-accumulators group having 4.04 scans ±0.76, the Aβ− accumulators group having 3.81 scans ±1.01 and the Aβ+ accumulators group having 3.64 scans ±0.96. The scans were conducted 2.5 years apart from each other, with the Aβ− non-accumulators group having an average gap of 2.35 years ±0.57, the Aβ− accumulators group an average gap of 2.36 years ±0.69 and the Aβ+ accumulators an average gap of 2.53 years ±1.01. The characteristics of each group are summarized in Table 1, and they were compared between all groups using the Kruskal–Wallis rank sum test.

Table 1.

Demographic and clinical data of the participants.

	Aβ− non-accumulators (n = 46)	Aβ− accumulators (n = 31)	Aβ+ accumulators (n = 36)	P-values
Age	75.57 (6.24)	76.26 (6.35)	78.27 (6.32)	.112
Sex (M/F)	23/23	10/21	19/17	.19
Education	16.78 (2.30)	16.71 (2.51)	16.17 (2.59)	.482
APOE e4 carriers (%)	15.22	30	45.71	.007
Aβ PET summary SUVR (normalized by whole cerebellum)	0.98 (0.06)	1.02 (0.05)	1.30 (0.16)	<.001
Aβ PET baseline summary SUVR (normalized by composite reference region)	0.70 (0.03)	0.72 (0.04)	0.81 (0.09)	.047
Average rate of change in Aβ PET composite SUVR	−3 × 10⁻³ (1.9 × 10⁻³)	4.4 × 10⁻³ (4 × 10⁻³)	8.8 × 10⁻³ (5.7 × 10⁻³)	<.001
Percentage change in Aβ PET composite SUVR (%)	3.9 (2.2)	5.1 (4.7)	10.6 (5.6)	<.001
MMSE	28.93 (1.39)	29.42 (0.89)	28.67 (1.64)	.053
PACCtrailsB	0.42 (1.89)	0.17 (2.41)	−0.73 (2.89)	.219

Abbreviations: Aβ, amyloid-β; Aβ−, Aβ-negative; Aβ+, Aβ-positive; M/F, Male/Female; APOE ε4, apolipoprotein E ε4 allele; MMSE, Mini Mental State Examination; PACCtrailsB, modified Preclinical Alzheimer Cognitive Composite with trails test; PET SUVR, positron emission tomography standardized uptake value ratios.

Means are followed by standard deviations. The SUVR values normalized by whole cerebellum are presented for the PET scan nearest in time to the fMRI scan.

Anti-symmetric functional networks capture directed functional connectivity between brain areas

Temporal delays can arise in the interaction between brain areas because of the dispersed nature of brain regions and the limited speed of information transfer between them.²² Therefore, to achieve a coherent characterization of the functional connectivity, it is crucial to capture the information contained in this complicated temporal lag structure.^23,24 To this purpose, we have developed an anti-symmetric correlation-based approach for determining the directed connections between pairs of brain areas, allowing for the computation of whole-brain directed functional connectivity networks.¹⁵ In this approach, if one brain area’s activation time series has similar properties to the time-shifted version of the activation pattern in another brain region, then the first region is considered to have a directed interaction with the second region. The level of directed interaction is quantified by the inverse of the time lag between the time series of those regions (which captures the fact that directly connected brain regions are expected to activate with a much shorter delay than indirectly connected regions). The direction of the interaction is determined by the order of precedence in time (ie, the early region is the source, and the late region is the end of the connection).

Figure 1A illustrates the calculation of the anti-symmetric connectivity networks for a set of 5 brain regions and their corresponding time series. The lagged correlation matrix can be evaluated by calculating the time-lagged Pearson’s correlation coefficient between all pairs of regions at a given temporal lag. This matrix captures the directed connection between the regions in both directions, that is, the matrix entries (i, j) and (j, i) provide an estimate of the directed connection from region i to j and vice versa. As with any other square matrix, the lagged correlation adjacency matrix can be uniquely expressed as the sum of a symmetric and anti-symmetric matrix. We use the anti-symmetric matrix to approximate the whole-brain directed functional connectivity; this method summarizes the directed connection between regions i and j with a single entry that captures its direction and magnitude. More details about the method are presented elsewhere.¹⁵

Figure 1.

Sample differences between different groups in global and nodal directed functional network topology. (A) Overview of the calculation procedure for the anti-symmetric correlation. For simplicity, we illustrate the procedure using the time activation series for 5 nodes as an example. The lagged Pearson’s correlation coefficient between these time series at varying temporal lags allows the estimation of lagged correlation functional networks (in this example, at lag of 1). The lagged adjacency matrix can be written as a sum of symmetric and anti-symmetric and matrices. Here, we use the anti-symmetric matrices to estimate the functional connectivity of each individual. Differences between the different groups at lag 3 (calculated as group 2-group 1 in the figure) in the (B) clustering coefficient and (C) global efficiency. The differences in the corresponding network measures between groups are plotted in orange circles as a function of network density and the upper and lower bounds of the 95% confidence interval are plotted in blue. The differences are considered statistically significant if they fall outside the CIs. (D) Differences between Aβ−/NAc and Aβ+ groups in nodal network measures. The plotted regions remained significant after correcting for multiple comparisons using FDR at q < 0.05.

Network construction and analysis

We calculated a 200 × 200 weighted connectivity matrix for each participant, with the edges generated using anti-symmetric correlation and the nodes corresponding to the 200 Craddock et al atlas brain regions.²⁵ We calculated a binary matrix for each adjacency matrix in which the correlation coefficient was assigned a value of 1 if it was greater than a certain threshold and 0 otherwise. We performed this binarization procedure throughout the complete range of network densities (D) available to the anti-symmetric correlation (D_range = 1%-50%) and compared the global topology of the binary networks across that range. In the case of nodal measures, we estimated measure-specific area under the curve (AUC) by numerically integrating the nodal measure values over the whole density range; the curves depict the evolution of the appropriate nodal network measure as a function of the network density for a specific brain area. These AUC values were used to assess the differences in the nodaldirected connectivity patterns between the different groups. We assessed the global and nodal topology by calculating the directed clustering coefficient and global efficiency using the BRAPH software package.²⁶

Statistical analysis

The statistical significance of differences in network measures across groups was determined using nonparametric permutation testing with 10 000 permutations, with P < .05 considered significant for a two-tailed test of the null hypothesis. The nodal measures were corrected for multiple comparisons by using the Benjamini-Hochberg procedure²⁷ to apply false discovery rate (FDR) corrections at q < 0.05 to adjust for the number of brain regions. All analyses were carried out using age, sex, and education as covariates. The correlation analyses were performed at each density for the global measures and for brain regions that showed between-group significant differences. All network measures were adjusted for age, sex, and education before carrying out the correlation analyses. In the case of global measures, the correlations were considered consistent if they remained significant at an extended range of densities after the application of FDR correction to control for the number of densities.

Results

Amyloid deposition results in changes in functional segregation and integration properties of the network

We have previously shown that the changes in directed functional connectivity due to neurodegenerative processes can be captured by small temporal lags, up to the lag of 7.¹⁵ As the current study employs functional MRI scans that have been acquired with longer repetition times, we evaluate the between-group differences at small delays up to a lag of 5, which corresponds to the same temporal scale used in Mijalkov et al.¹⁵ The between-group differences in the clustering coefficient and global efficiency at a lag of 3 are shown in Figure 1B and Figure 1C respectively, whereas the corresponding results for lags 2 and 4 are plotted in Supplemental Figure S1. The anti-symmetric correlation methods showed widespread significant decreases in the clustering coefficient in the Aβ-negative accumulators when compared to Aβ-negative non-accumulators group. Similar differences were observed in the Aβ+ group; in addition, the Aβ+ group showed also decreased global efficiency when compared to Aβ-negative non-accumulators (Figure 1B and Figure 1C). The differences between Aβ+ individuals and Aβ− non-accumulators were most pronounced in the superior frontal gyrus, caudate nucleus, and anterior cingulate cortex at different temporal lags (Figure 1D).

Directed functional connectivity is associated with global amyloid deposition level and cognitive scores in Aβ− accumulators

In the Aβ− accumulators group, the clustering coefficient was significantly associated with baseline global SUVR values at lag 4 at high-density values (33%-50%). The global efficiency showed significant associations with the SUVR values at lags 2 (5%-9% and 13%-39%), 3 (3%-40%), and 4 (2%-39%). Regarding the cognitive tests, the PACC with trails test scores showed a significant association with the global efficiency at lag 2 (18%-34%). A representative correlation for each association at a single density is shown in Figure 2.

Figure 2.

Correlation between Aβ deposition levels and cognitive scores and global measures for Aβ−/Ac group. The plots show the best line fit for Aβ−/Ac group between baseline Aβ SUVR deposition levels and (A) clustering coefficient at lag 4, and global efficiency at lags (B) 2, (C) 3, and (D) 4. The correlation between PACCtrailsB test scores and global efficiency at lag 2 is shown in (F). The global measure values were regressed with age, sex, and education as covariates; the residuals of the regression were used to calculate the best line fit. A representative fit at density (A) 39% and (B-F) 25% is shown.

Discussion

Anti-symmetric correlations can be used for quantifying directed connectivity between brain regions by harvesting the information contained in the temporal lags between the activation time series of different regions.¹⁵ In this study, we used this approach to demonstrate the impact of Aβ deposition on directed functional connectivity in a group of healthy, cognitively normal adults. We observed that groups with larger Aβ load exhibited reduced levels of functional segregation and integration in their functional connectomes, both at global network levels as well as in the superior frontal gyrus, caudate nucleus, and anterior cingulate cortex. The global measures in the Aβ negative group with positive rates of amyloid accumulation were associated with global Aβ load and cognitive tests, indicating that they are sensitive to pathological and clinical alterations that may occur during preclinical AD. Overall, our results suggest that the directional flow in brain activity includes unique information that might serve as a novel biomarker to study functional abnormalities in preclinical AD.

In the preclinical stage of AD, accumulating protein aggregates already affect neurons and induce synaptic dysfunction that can be detected with fMRI even in the absence of express neurodegeneration.²⁸ In particular, before the presence of detectable amyloid plaques in the brain, soluble Aβ oligomers already appear in the intracellular space and disrupt neurotransmission on both pre- and postsynaptic levels,²⁹ while exerting a specific effect on glutamatergic transmission by interfering with the mechanisms of long-term depression and potentiation that contribute to early memory deficits during the course of the disease.³⁰ The earliest known sites of Aβ-accumulation are the precuneus, posterior cingulate, and medial prefrontal cortices,³¹ which are all part of the default mode network.³⁰ Studies have demonstrated that this initial accumulation of Aβ, even in individuals who have not yet reached clinically relevant threshold levels, leads to functional reorganization in the default mode as well as the salience network.³² Furthermore, a significant decrease in functional connectivity within the default mode and other resting state networks has also been observed among cognitively normal older individuals with elevated levels of Aβ,^31,33-35 which correlated with markers of pre- and postsynaptic activity.³⁶ Such individuals also exhibited changes in the functional connectivity between the different resting-state networks, suggesting that regions connected to multiple networks may be particularly vulnerable to Aβ induced alterations.^37,38 Overall, this suggests that Aβ has a widespread effect on functional connectivity, impacting the connectivity of multiple networks, both within and between networks, as well as being associated with individual cognitive capabilities. By demonstrating changes in both localized (ie, segregation) and global (ie, integration) functional connectivity patterns that are also associated with cognitive deficits, our findings align with these reports. Furthermore, they also highlight the utility of directed functional connectivity as a sensitive marker to detect these changes in individuals showing subthreshold amyloid accumulation.

Anti-symmetric correlation networks had an abnormal global topology in the individuals with higher Aβ deposition, characterized by decreases in global efficiency and clustering coefficient. When compared to Aβ− non-accumulators, these differences were more pronounced in the Aβ positive group than in Aβ negative accumulators, suggesting a link between the topology and amount of Aβ in the brain. The clustering coefficient provides an estimation of the functional segregation of the connectivity network and the ability to perform specialized tasks between neighboring regions, while the global efficiency estimates the functional integration and efficiency of the information transfer between distant brain regions.^26,39 The existence of locally clustered connections and high functional integration are indicative of a small-world network, an organization characteristic of a well-functioning brain network.^39,40 Therefore, the reductions in the clustering coefficient and global efficiency can be interpreted as a loss of the small-world property of the functional networks, a finding consistently observed in AD.^8,41,42 Given the importance of functional connectivity for cognitive performance,⁴³ disruptions in the integrity of the functional networks can have a significant role in the development of cognitive deficits in AD. This is also supported by our findings, which demonstrate that impaired cognitive function is linked to disrupted functional integration in preclinical AD.

Regarding alterations at the nodal level, we found reduced global efficiency and clustering coefficient in the superior frontal gyrus, the anterior cingulate cortex, and the caudate nucleus in Aβ positive individuals. The superior frontal gyrus is part of the frontoparietal executive and control networks important in executive function.⁴⁴ The disruption of network measures in the superior frontal gyrus is in line with previous reports showing that executive function is affected early on in the disease course.⁴⁵ The anterior cingulate cortex is a prominent area of the salience network that is also impaired in AD.³² Functional changes of the caudate nucleus are not widely reported in AD, however, a loss of volume in the caudate nucleus has been reported as an early sign of the disease.⁴⁶ Additionally, these areas participate in the function of the default mode network, one of the earliest affected brain networks in AD.³⁸ To summarize, alterations of nodal network characteristics reinforce earlier reports of executive dysfunction as one of the earliest affected cognitive domains during the course of AD, while also affecting the default mode network.

One limitation of this study is the potential impact of test-retest variability on the calculated rate of change in Aβ values. Test-retest variability refers to the degree of consistency in measurements taken at different points in time, which was found to be 1.5% ± 0.84 for the Florbetapir PET scans of controls,⁴⁷ suggesting that smaller percentage changes in SUVR values may not accurately reflect changes in Aβ levels. Despite this, the group-average percentage changes in SUVR values in this study were higher than the test-retest variability rate (Table 1), indicating that the calculated slopes accurately capture changes in Aβ in these individuals. However, as there is a within-group variability among the individuals that highlights the persisting degree of uncertainty when assessing changes in AD,^47,48 the results of this study should be interpreted with this limitation in mind.

In conclusion, we demonstrate that the temporal activation delays can be used to evaluate the directed functional connectivity between the different brain regions. Our findings indicate that the global topology of such whole-brain directed functional networks, as well as the directed connectivity patterns of several brain regions previously implicated in AD, are altered in sub-threshold Aβ− accumulators. Furthermore, the degree of the observed alterations was associated with the levels of amyloid-beta deposition and cognitive test scores reflecting executive functions, suggesting that the method of anti-symmetric correlations could potentially be used as a marker for identifying individuals at early stages of amyloid-related pathology.

Supplemental Material

sj-docx-1-exn-10.1177_26331055231161625 – Supplemental material for Directed Functional Brain Connectivity is Altered in Sub-threshold Amyloid-β Accumulation in Cognitively Normal Individuals

Supplemental material, sj-docx-1-exn-10.1177_26331055231161625 for Directed Functional Brain Connectivity is Altered in Sub-threshold Amyloid-β Accumulation in Cognitively Normal Individuals by Mite Mijalkov, Dániel Veréb, Anna Canal-Garcia, Giovanni Volpe and Joana B Pereira in Neuroscience Insights

Footnotes

Acknowledgements

Data collection and sharing of ADNI was funded by the National Institutes of Health Grant U01 AG024904 and Department of Defense award number W81XWH-12-2-0012. ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; BioClinica, Inc.; Biogen Idec Inc.; Bristol-Myers Squibb Company; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; F. Hoffmann-LaRoche Ltd and its affiliated company Genentech, Inc.; GEHealthcare; Innogenetics, N.V.; IXICO Ltd; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Medpace, Inc.; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Synarc Inc.; and Takeda Pharmaceutical Company. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Disease Cooperative Study at the University of California, San Diego. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of California, Los Angeles.

Author Contributions

MM: Methodology, formal analysis, preparation of original draft, review and editing, funding acquisition. DV: data preprocessing, preparation of original draft, review and editing. ACG: data preprocessing, review and editing. GV: methodology, review and editing. JBP: supervision, methodology, preparation of original draft, review and editing, funding acquisition.

Funding:

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Swedish Research Council; Swedish Alzheimer Foundation; Swedish Brain Foundation; Strategic Research Area Neuroscience (StratNeuro); Center for Medical Innovation (CIMED); Foundation for Geriatric Diseases at Karolinska Institutet; Gamla Tjänarinnor; Stohne’s Foundation; Lars Hierta Memorial Foundation.

Declaration of Conflicting Interests:

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability

Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu), which is available to researchers after application for access. The BRAPH software can be freely downloaded from: http://braph.org/. The investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in the analysis or writing of this report. A complete listing of ADNI investigators can be found at:

ORCID iDs

Mite Mijalkov

Dániel Veréb

Supplemental Material

Supplemental material for this article is available online.

References

Winblad

Amouyel

Andrieu

, et al. Defeating Alzheimer’s disease and other dementias: a priority for European science and society. Lancet Neurol. 2016;15:455-532.

Dubois

Hampel

Feldman

, et al. Preclinical Alzheimer’s disease: definition, natural history, and diagnostic criteria. Alzheimers Dement. 2016;12:292-323.

Jack

Jr Knopman

Jagust

, et al. Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. 2013;12:207-216.

Hardy

Selkoe

DJ.

The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science. 2002;297:353-356.

Mormino

Papp

KV.

Amyloid accumulation and cognitive decline in clinically normal older individuals: Implications for aging and early Alzheimer’s disease. J Alzheimers Dis. 2018;64:S633-S646.

Aizenstein

Nebes

Saxton

, et al. Frequent amyloid deposition without significant cognitive impairment among the elderly. Arch Neurol. 2008;65:1509-1517.

Villemagne

Burnham

Bourgeat

, et al. Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study. Lancet Neurol. 2013;12:357-367.

Pievani

de Haan

Seeley

Frisoni

GB.

Functional network disruption in the degenerative dementias. Lancet Neurol. 2011;10:829-843.

Insel

Ossenkoppele

Gessert

, et al. Time to amyloid positivity and preclinical changes in brain metabolism, atrophy, and cognition: evidence for emerging amyloid pathology in Alzheimer’s disease. Front Neurosci. 2017;11:281.

10.

Leal

Lockhart

Maass

Bell

Jagust

WJ.

Subthreshold amyloid predicts tau deposition in aging. Neurosci J. 2018;38:4482-4489.

11.

Koychev

Vaci

Bilgel

, et al. Prediction of rapid amyloid and phosphorylated-tau accumulation in cognitively healthy individuals. Alzheimer’s Dement. 2020;12:e12019.

12.

Mattsson

Insel

Nosheny

, et al. Emerging β-amyloid pathology and accelerated cortical atrophy. JAMA Neurol. 2014;71:725-734.

13.

Hedden

Van Dijk

Becker

, et al. Disruption of functional connectivity in clinically normal older adults harboring amyloid burden. J Neurosci. 2009;29:12686-12694.

14.

Landau

Horng

Jagust

. Alzheimer’s Disease Neuroimaging Initiative. Memory decline accompanies subthreshold amyloid accumulation. Neurology. 2018;90:e1452-e1460.

15.

Mijalkov

Volpe

Pereira

JB.

Directed brain connectivity identifies widespread functional network abnormalities in Parkinson’s disease. Cereb Cortex. 2022;32:593-607.

16.

Donohue

Sperling

Salmon

, et al. The preclinical Alzheimer cognitive composite: measuring amyloid-related decline. JAMA Neurol. 2014;71:961-970.

17.

Esteban

Markiewicz

Blair

, et al. fMRIPrep: a robust preprocessing pipeline for functional MRI. Nat Methods. 2019;16:111-116.

18.

Friston

Williams

Howard

Frackowiak

Turner

Movement-related effects in fMRI time-series. Magn Reson Med. 1996;35:346-355.

19.

Landau

Mintun

Joshi

, et al. Amyloid deposition, hypometabolism, and longitudinal cognitive decline. Ann Neurol. 2012;72:578-586.

20.

Landau

Breault

Joshi

, et al. Amyloid-β imaging with Pittsburgh compound B and florbetapir: comparing radiotracers and quantification methods. J Nucl Med. 2013;54:70-77.

21.

Landau

Fero

Baker

, et al. Measurement of longitudinal β-amyloid change with 18F-florbetapir PET and standardized uptake value ratios. J Nucl Med. 2015;56:567-574.

22.

Deco

Jirsa

McIntosh

Sporns

Kötter

Key role of coupling, delay, and noise in resting brain fluctuations. Proc Natl Acad Sci USA. 2009;106:10302-10307.

23.

Lahaye

Poline

Flandin

Dodel

Garnero

Functional connectivity: studying nonlinear, delayed interactions between BOLD signals. Neuroimage. 2003;20:962-974.

24.

Mijalkov

Pereira

Volpe

Delayed correlations improve the reconstruction of the brain connectome. PLoS One. 2020;15:e0228334.

25.

Craddock

James

Holtzheimer

3rd Hu

Mayberg

HS.

A whole brain fMRI atlas generated via spatially constrained spectral clustering. Hum Brain Mapp. 2012;33:1914-1928.

26.

Mijalkov

Kakaei

Pereira

, et al. BRAPH: a graph theory software for the analysis of brain connectivity. PLoS One. 2017;12:e0178798.

27.

Benjamini

Hochberg

Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B. 1995;57:289-300.

28.

Sperling

Aisen

Beckett

, et al. Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s Dement. 2011;7:280-292.

29.

Lacor

Buniel

Chang

, et al. Synaptic targeting by Alzheimer’s-related amyloid β oligomers. Neurosci J. 2004;24:10191-10200.

30.

Mucke

Selkoe

DJ.

Neurotoxicity of amyloid β-protein: synaptic and network dysfunction. Cold Spring Harb Perspect Med. 2012;2:a006338.

31.

Palmqvist

Schöll

Strandberg

, et al. Earliest accumulation of β-amyloid occurs within the default-mode network and concurrently affects brain connectivity. Nat Commun. 2017;8:1214-1313.

32.

Butt

Meeker

Wisch

, et al. Network dysfunction in cognitively normal APOE ε4 carriers is related to subclinical tau. Alzheimer’s Dement. 2022;18:116-126.

33.

Nuttall

Pasquini

Scherr

Sorg

. Alzheimer’s Disease Neuroimaging Initiative. Degradation in intrinsic connectivity networks across the Alzheimer’s disease spectrum. Alzheimer’s Dement. 2016;5:35-42.

34.

Pasquini

Benson

Grothe

, et al. Individual correspondence of amyloid-β and intrinsic connectivity in the posterior default mode network across stages of Alzheimer’s disease. J Alzheimers Dis. 2017;58:763-773.

35.

Nakamura

Cuesta

Kato

, et al. Early functional network alterations in asymptomatic elders at risk for Alzheimer’s disease. Sci Rep. 2017;7:6517-6611.

36.

Pereira

Janelidze

Ossenkoppele

, et al. Untangling the association of amyloid-β and tau with synaptic and axonal loss in Alzheimer’s disease. Brain. 2021;144:310-324.

37.

Elman

Madison

Baker

, et al. Effects of beta-amyloid on resting state functional connectivity within and between networks reflect known patterns of regional vulnerability. Cereb Cortex. 2014;26:695-707.

38.

Jones

Knopman

Gunter

, et al. Cascading network failure across the Alzheimer’s disease spectrum. Brain. 2016;139:547-562.

39.

Fornito

Zalesky

Breakspear

Graph analysis of the human connectome: promise, progress, and pitfalls. Neuroimage. 2013;80:426-444.

40.

Watts

Strogatz

SH.

Collective dynamics of ‘small-world’ networks. Nature. 1998;393:440-442.

41.

Supekar

Menon

Rubin

Musen

Greicius

MD.

Network analysis of intrinsic functional brain connectivity in Alzheimer’s disease. PLoS Comput Biol. 2008;4:e1000100.

42.

Sanz-Arigita

Schoonheim

Damoiseaux

, et al. Loss of ‘small-world’ networks in Alzheimer’s disease: graph analysis of FMRI resting-state functional connectivity. PLoS One. 2010;5:e13788.

43.

van den Heuvel

Hulshoff Pol

HE.

Exploring the brain network: a review on resting-state fMRI functional connectivity. Eur Neuropsychopharmacol. 2010;20:519-534.

44.

Menon

Large-scale brain networks and psychopathology: a unifying triple network model. Trends Cogn Sci. 2011;15:483-506.

45.

Tideman

Stomrud

Leuzy

Mattsson-Carlgren

Palmqvist

Hansson

Association of β-amyloid accumulation with executive function in adults with unimpaired cognition. Neurology. 2022;98:e1525-e1533.

46.

Jiji

Smitha

Gupta

Pillai

Jayasree

RS.

Segmentation and volumetric analysis of the caudate nucleus in Alzheimer’s disease. Eur J Radiol. 2013;82:1525-1530.

47.

Joshi

Pontecorvo

Clark

, et al. Performance characteristics of amyloid PET with florbetapir F 18 in patients with Alzheimer’s disease and cognitively normal subjects. J Nucl Med. 2012;53:378-384.

48.

Dubois

Feldman

Jacova

, et al. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol. 2007;6:734-746.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB