Cited by 132Functional Brain Connectivity StudiesDementia and Cognitive Impairment ResearchAlzheimer's disease research and treatments

The DOI matches its Crossref record; 57 of the 57 references it cites were found in a registry or an index.

Within-patient correspondence of amyloid-β and intrinsic network connectivity in Alzheimer’s disease

Open access

Abstract

There is striking overlap between the spatial distribution of amyloid-β pathology in patients with Alzheimer's disease and the spatial distribution of high intrinsic functional connectivity in healthy persons. This overlap suggests a mechanistic link between amyloid-β and intrinsic connectivity, and indeed there is evidence in patients for the detrimental effects of amyloid-β plaque accumulation on intrinsic connectivity in areas of high connectivity in heteromodal hubs, and particularly in the default mode network. However, the observed spatial extent of amyloid-β exceeds these tightly circumscribed areas, suggesting that previous studies may have underestimated the negative impact of amyloid-β on intrinsic connectivity. We hypothesized that the known positive baseline correlation between patterns of amyloid-β and intrinsic connectivity may mask the larger extent of the negative effects of amyloid-β on connectivity. Crucially, a test of this hypothesis requires the within-patient comparison of intrinsic connectivity and amyloid-β distributions. Here we compared spatial patterns of amyloid-β-plaques (measured by Pittsburgh compound B positron emission tomography) and intrinsic functional connectivity (measured by resting-state functional magnetic resonance imaging) in patients with prodromal Alzheimer's disease via spatial correlations in intrinsic networks covering fronto-parietal heteromodal cortices. At the global network level, we found that amyloid-β and intrinsic connectivity patterns were positively correlated in the default mode and several fronto-parietal attention networks, confirming that amyloid-β aggregates in areas of high intrinsic connectivity on a within-network basis. Further, we saw an internetwork gradient of the magnitude of correlation that depended on network plaque-load. After accounting for this globally positive correlation, local amyloid-β-plaque concentration in regions of high connectivity co-varied negatively with intrinsic connectivity, indicating that amyloid-β pathology adversely reduces connectivity anywhere in an affected network as a function of local amyloid-β-plaque concentration. The local negative association between amyloid-β and intrinsic connectivity was much more pronounced than conventional group comparisons of intrinsic connectivity would suggest. Our findings indicate that the negative impact of amyloid-β on intrinsic connectivity in heteromodal networks is underestimated by conventional analyses. Moreover, our results provide first within-patient evidence for correspondent patterns of amyloid-β and intrinsic connectivity, with the distribution of amyloid-β pathology following functional connectivity gradients within and across intrinsic networks.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/brain/awu103

Cites

57 of 57 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Overview of analysis approach. For each individual, voxel-wise PiB-uptake was determined as a measure of amyloid plaque density. For each intrinsic network j (as determined by resting-state functional MRI and independent component analysis independent component analysis) in each individual k, we identified the voxels belonging to that network (top right) and extracted the intrinsic functional connectivity (iFC) values of those (∼10 000) voxels. We also extracted that individual’s PiB-uptake values for the same voxels. To determine average plaque load in each network, we then calculated the median PiB-uptake (shown in Fig. 2A and B). We then calculated global and local correlations between both modalities across the entire network and confined to neighbouring voxels. To determine the global spatial correlation between intrinsic functional connectivity and PiB-uptake, we calculated the Pearson correlation coefficient across all selected voxels in the network (rGLOBAL, bottom left and Fig. 2C). After correcting for rGLOBAL (via orthogonalization, see ‘Materials and methods’ section), we then used a searchlight approach to calculate local spatial correlation in the neighborhood surrounding each voxel in the network (rLOCAL, bottom right and Fig. 3). Aβ = amyloid-β.
Figure 1 Overview of analysis approach. For each individual, voxel-wise PiB-uptake was determined as a measure of amyloid plaque density. For each intrinsic network j (as determined by resting-state functional MRI and independent component analysis independent component analysis) in each individual k, we identified the voxels belonging to that network (top right) and extracted the intrinsic functional connectivity (iFC) values of those (∼10 000) voxels. We also extracted that individual’s PiB-uptake values for the same voxels. To determine average plaque load in each network, we then calculated the median PiB-uptake (shown in Fig. 2A and B). We then calculated global and local correlations between both modalities across the entire network and confined to neighbouring voxels. To determine the global spatial correlation between intrinsic functional connectivity and PiB-uptake, we calculated the Pearson correlation coefficient across all selected voxels in the network (rGLOBAL, bottom left and Fig. 2C). After correcting for rGLOBAL (via orthogonalization, see ‘Materials and methods’ section), we then used a searchlight approach to calculate local spatial correlation in the neighborhood surrounding each voxel in the network (rLOCAL, bottom right and Fig. 3). Aβ = amyloid-β.CC BY · PubMed Central
Figure 2 Amyloid-β deposition is increased in a number of heteromodal networks and correlates with intrinsic functional connectivity. (A) Each image shows a group-level intrinsic functional connectivity (iFC) network map (cold colours) at a FWE-corrected threshold of P < 0.05, superimposed on the group-level difference map in amyloid-β deposition (warm colors), thresholded at P < 0.0001, minimum cluster extent of 100 voxels. The pattern of amyloid-β (Aβ) deposition overlaps noticeably with a number of fronto-parietal networks (lATN = left lateral frontoparietal ATN; rATN = right frontoparietal ATN; dATN = posterior parietal dorsal ATN; SN = salience network) besides the DMN [split into a posterior (pDMN) and anterior (aDMN) component], and even with regions of primary sensory cortex, such as the primary auditory network (pAN). Notably, in some networks amyloid-β plaque deposition appears limited to network hubs (as in the left ATN and dorsal ATN, where frontal network regions lack significant amyloid-β burden at the group level). (B) Patients showed significantly increased PiB-uptake in all tested networks [main effect of group across seven networks, F(1,33) = 22.11, P < 10−4]. There was also a main effect of network that interacted with group [main effect of network for patient group, F(6,132) = 5.37, P < 10−4], indicating that there is a gradient of amyloid-β deposition across networks in patients. (C) A number of networks in the patient group exhibited a positive network-wide correlation between voxel-wise intrinsic functional connectivity and voxel-wise PiB-uptake (i.e. rGLOBAL), indicating that voxels in the network core (i.e. voxels with high intrinsic functional connectivity) were burdened with higher amyloid-β deposition than the network periphery (voxels with low intrinsic functional connectivity). This correlation was modest in magnitude but consistently positive [main effect of group, F(1,33) = 47.19, P < 10−7, t-tests of Fisher-transformed Pearson correlation coefficients against control group, all t(33) > 3.61, Bonferroni-corrected P < 0.01, except for the auditory network, t(33) = 1.87, non-significant]. Correlations in the control group were negligible. There was a gradient of correlation strength across networks as the six default and attention networks showed a stronger correlation than the posterior auditory network. Also, the amyloid-β–intrinsic functional connectivity correlation was stronger in the posterior DMN than in any of the other networks except the anterior DMN. Error bars indicate standard error of the mean. ***P < 0.001, **P < 0.01, *P < 0.05, all Bonferroni-corrected, ns = not significant. (D) Median PiB-uptake is positively correlated with rGLOBAL in patients (left), but not in controls (right). Each point in the scatter plot represents one network in one individual, with network identity denoted by colour. The regression lines are derived from a linear fit across all individuals and networks in each group. Nevertheless, correlations were also performed separately for each network.
Figure 2 Amyloid-β deposition is increased in a number of heteromodal networks and correlates with intrinsic functional connectivity. (A) Each image shows a group-level intrinsic functional connectivity (iFC) network map (cold colours) at a FWE-corrected threshold of P < 0.05, superimposed on the group-level difference map in amyloid-β deposition (warm colors), thresholded at P < 0.0001, minimum cluster extent of 100 voxels. The pattern of amyloid-β (Aβ) deposition overlaps noticeably with a number of fronto-parietal networks (lATN = left lateral frontoparietal ATN; rATN = right frontoparietal ATN; dATN = posterior parietal dorsal ATN; SN = salience network) besides the DMN [split into a posterior (pDMN) and anterior (aDMN) component], and even with regions of primary sensory cortex, such as the primary auditory network (pAN). Notably, in some networks amyloid-β plaque deposition appears limited to network hubs (as in the left ATN and dorsal ATN, where frontal network regions lack significant amyloid-β burden at the group level). (B) Patients showed significantly increased PiB-uptake in all tested networks [main effect of group across seven networks, F(1,33) = 22.11, P < 10−4]. There was also a main effect of network that interacted with group [main effect of network for patient group, F(6,132) = 5.37, P < 10−4], indicating that there is a gradient of amyloid-β deposition across networks in patients. (C) A number of networks in the patient group exhibited a positive network-wide correlation between voxel-wise intrinsic functional connectivity and voxel-wise PiB-uptake (i.e. rGLOBAL), indicating that voxels in the network core (i.e. voxels with high intrinsic functional connectivity) were burdened with higher amyloid-β deposition than the network periphery (voxels with low intrinsic functional connectivity). This correlation was modest in magnitude but consistently positive [main effect of group, F(1,33) = 47.19, P < 10−7, t-tests of Fisher-transformed Pearson correlation coefficients against control group, all t(33) > 3.61, Bonferroni-corrected P < 0.01, except for the auditory network, t(33) = 1.87, non-significant]. Correlations in the control group were negligible. There was a gradient of correlation strength across networks as the six default and attention networks showed a stronger correlation than the posterior auditory network. Also, the amyloid-β–intrinsic functional connectivity correlation was stronger in the posterior DMN than in any of the other networks except the anterior DMN. Error bars indicate standard error of the mean. ***P < 0.001, **P < 0.01, *P < 0.05, all Bonferroni-corrected, ns = not significant. (D) Median PiB-uptake is positively correlated with rGLOBAL in patients (left), but not in controls (right). Each point in the scatter plot represents one network in one individual, with network identity denoted by colour. The regression lines are derived from a linear fit across all individuals and networks in each group. Nevertheless, correlations were also performed separately for each network.CC BY · PubMed Central
Figure 3 Local negative impact of amyloid-β on intrinsic connectivity. (A) Searchlight analysis demonstrated negative correlations between amyloid-β plaques (Aβ) and intrinsic functional connectivity (iFC) in small neighbourhoods with generally homogeneous connectivity strength. After accounting for the impact of intrinsic functional connectivity on PiB-uptake across an entire network (i.e. rGLOBAL described in Figs 1B and 2) through orthogonalization, we calculated at each network voxel the local amyloid-β–intrinsic functional connectivity (Aβ,iFC) correlation rLOCAL across surrounding voxels in a 6-mm radius. For each group, we submitted the subject-level maps of correlation coefficients to t-tests, using FWE to correct for multiple comparisons. In the prodromal AD group (top), all networks except for the auditory network (pAN) showed significant negative rLOCAL across large parts of the networks, focusing on network hubs. There were no such correlations in the control group (bottom). Colour scale indicates t-statistics at the group level. A t-value of 3.5 corresponds to P = 0.001, uncorrected. (B) In patients, median rLOCAL values were significantly below 0 for all networks except the auditory network. The control group showed no significant correlations. Hence, direct comparisons between patients and controls revealed a strong main effect of group [F(1,33) = 43.25, P < 10−6, all t(33) <−3.17, corrected P < 0.025], with the exception of the auditory network, which showed no difference in median rLOCAL [t(33) = 0.54, ns]. Error bars indicate standard error of the mean. a/pDMN = anterior/posterior DMN; l/r/dATN = left/right/dorsal ATN; SN = salience network.
Figure 3 Local negative impact of amyloid-β on intrinsic connectivity. (A) Searchlight analysis demonstrated negative correlations between amyloid-β plaques (Aβ) and intrinsic functional connectivity (iFC) in small neighbourhoods with generally homogeneous connectivity strength. After accounting for the impact of intrinsic functional connectivity on PiB-uptake across an entire network (i.e. rGLOBAL described in Figs 1B and 2) through orthogonalization, we calculated at each network voxel the local amyloid-β–intrinsic functional connectivity (Aβ,iFC) correlation rLOCAL across surrounding voxels in a 6-mm radius. For each group, we submitted the subject-level maps of correlation coefficients to t-tests, using FWE to correct for multiple comparisons. In the prodromal AD group (top), all networks except for the auditory network (pAN) showed significant negative rLOCAL across large parts of the networks, focusing on network hubs. There were no such correlations in the control group (bottom). Colour scale indicates t-statistics at the group level. A t-value of 3.5 corresponds to P = 0.001, uncorrected. (B) In patients, median rLOCAL values were significantly below 0 for all networks except the auditory network. The control group showed no significant correlations. Hence, direct comparisons between patients and controls revealed a strong main effect of group [F(1,33) = 43.25, P < 10−6, all t(33) <−3.17, corrected P < 0.025], with the exception of the auditory network, which showed no difference in median rLOCAL [t(33) = 0.54, ns]. Error bars indicate standard error of the mean. a/pDMN = anterior/posterior DMN; l/r/dATN = left/right/dorsal ATN; SN = salience network.CC BY · PubMed Central
Figure 4 Proposed model of graded network degeneration. Graded network degeneration hypothesis of amyloid-β propagation in Alzheimer’s disease (AD). Left: In spite of a predominant focus on amyoid-β pathology accumulation and spread along intrinsic connectivity in the DMN [top bar, with colour indicating the (putatively amyloid-related) connectivity decrease over time], amyloid-β plaques are also found in other cortical regions that typically have lower resting-state metabolism and connectivity than the DMN (e.g. left and right ATNs). In an average person with prodromal Alzheimer’s disease, amyloid-β pathology deposition and connectivity-based spread will start in the DMN, followed in close succession by a number of other heteromodal networks, leading to a cross-network gradient (i.e. along the vertical grey lines cutting across networks). We propose that the same principle is at work ‘within’ networks (right), where the network core will accumulate amyloid-β and show signs of disruption sooner than areas in the periphery.
Figure 4 Proposed model of graded network degeneration. Graded network degeneration hypothesis of amyloid-β propagation in Alzheimer’s disease (AD). Left: In spite of a predominant focus on amyoid-β pathology accumulation and spread along intrinsic connectivity in the DMN [top bar, with colour indicating the (putatively amyloid-related) connectivity decrease over time], amyloid-β plaques are also found in other cortical regions that typically have lower resting-state metabolism and connectivity than the DMN (e.g. left and right ATNs). In an average person with prodromal Alzheimer’s disease, amyloid-β pathology deposition and connectivity-based spread will start in the DMN, followed in close succession by a number of other heteromodal networks, leading to a cross-network gradient (i.e. along the vertical grey lines cutting across networks). We propose that the same principle is at work ‘within’ networks (right), where the network core will accumulate amyloid-β and show signs of disruption sooner than areas in the periphery.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Cited by

Sorted by citations

See all 132

Metadata

Authors
Myers, N., Pasquini, L., Göttler, J., Grimmer, T., Koch, K., Ortner, M., Neitzel, J., Mühlau, M., Förster, S., Kurz, A., Förstl, H., Zimmer, C., Wohlschläger, A. M., Riedl, V., Drzezga, A., & Sorg, C.
Affiliations
  • TUM Klinikum (DE)
  • Technical University of Munich (DE)
  • University of Cologne (DE)
Corresponding author
Myers, Nicholas
Venue
Brain
Published
2014-04-26
Volume
137
Issue
7
Pages
2052-2064
ISSN
  • 0006-8950 (ISSN-L)
  • 1460-2156
OpenAlex
W2031921419
Cited by
132
References
57
Citations by year
2026: 3 · 2025: 3 · 2024: 9 · 2023: 13 · 2022: 9 · 2021: 15 · 2020: 5 · 2019: 10 · 2018: 13 · 2017: 20 · 2016: 11 · 2015: 14 · 2014: 7
FWCI
9.67
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Brain (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • eScholarship (California Digital Library) (submittedVersion, cc-by)
  • mediaTUM (Technical University of Munich) (submittedVersion)
Funders
  • Wellcome Trust
  • Bundesministerium für Bildung und Forschung (01ER0803, 01EV0710)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Functional Brain Connectivity Studies
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Alzheimer's disease research and treatments
Keywords
amyloid plaques, intrinsic functional connectivity, Alzheimer's disease, prodromal Alzheimer's disease, default mode network, frontoparietal attention network, within-patient comparison, Pittsburgh compound B PET, resting-state fMRI
MeSH terms
Aged, Aged, 80 and over, Alzheimer Disease, Analysis of Variance, Aniline Compounds, Brain, Brain Mapping, Female, Humans, Magnetic Resonance Imaging, Male, Middle Aged, Neural Pathways, Neuropsychological Tests, Thiazoles, Amyloid beta-Peptides, Positron-Emission Tomography, Plaque, Amyloid

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review