Neurobiology of Aging 96 (2020) 1e11

Contents lists available at ScienceDirect

Neurobiology of Aging
journal homepage: www.elsevier.com/locate/neuaging

Profound regional spectral, connectivity, and network changes

reflect visual deficits in posterior cortical atrophy: an EEG study
Casper T. Brielsa, b, *, Jakoba J. Eertinkb, Cornelis J. Stamb, Wiesje M. van der Fliera, c,
Philip Scheltensa, Alida A. Gouwb
a
Alzheimer Center Amsterdam, Department of Neurology, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam,
the Netherlands
b
Department of Clinical Neurophysiology, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, the Netherlands
c
Department of Epidemiology and Biostatistics, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam UMC, Amsterdam, the
Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Patients with posterior cortical atrophy (PCA-AD) show more severe visuospatial and perceptual deficits
Received 20 March 2020 than those with typical AD (tAD). The aim of this study was to investigate whether functional alterations
Received in revised form 20 July 2020 measured by electroencephalography can help understand the mechanisms that explain this clinical
Accepted 29 July 2020
heterogeneity. 21-channel electroencephalography recordings of 29 patients with PCA-AD were
Available online 5 August 2020
compared with 29 patients with tAD and 29 controls matched for age, gender, and disease severity.
Patients with PCA-AD and tAD both showed a global decrease in fast and increase in slow oscillatory
Keywords:
activity compared with controls. This pattern was, however, more profound in patients with PCA-AD
Posterior cortical atrophy (PCA)
Alzheimer's disease (AD)
which was driven by more extensive slowing of the posterior regions. Alpha band functional connec-
Electroencephalography (EEG) tivity showed a similar decrease in PCA-AD and tAD. Compared with controls, a less integrated network
Oscillatory activity topology was observed in PCA-AD, with a decrease of posterior and an increase of frontal hubness. In
Functional connectivity PCA-AD, decreased right parietal peak frequency correlated with worse performance on visual tasks.
Functional networks Regional vulnerability of the posterior network might explain the atypical pattern of neurodegeneration
in PCA-AD.
Ó 2020 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction
Posterior cortical atrophy (PCA) is a clinico-radiological neuro-
degenerative syndrome that is clinically characterized by an initially
isolated progressive impairment of higher order visual functions
with relative preservation of memory, insight, and judgment until
late in the clinical course of the disease (Beh et al., 2015; Crutch et al.,
2013, 2017; Migliaccio elt al., 2012). The most common underlying
pathology of PCA is Alzheimer's disease (PCA-AD), and the onset is
typically between 50 and 65 years of age (Crutch et al., 2012, 2017).
The clinical presentation of PCA-AD is in contrast with ‘typical’ Alz-
heimer's disease (tAD), where memory loss is the clinically pre-
dominant symptom next to other potential deficits in language,
executive functioning, behavior, and visual function.
Radiological findings in PCA-AD are typically represented by
predominant atrophy and hypometabolism of the parieto-occipital
* Corresponding author at: Alzheimer Center Amsterdam, Department of
Neurology, Amsterdam Neuroscience, Vrije Universiteit Amsterdam, Amsterdam
UMC, De Boelelaan 1117, 1081 HV Amsterdam, the Netherlands. Tel.: þ31 20
4440685; fax: þ31 20 4448529.
E-mail address: c.briels@amsterdamumc.nl (C.T. Briels).
0197-4580/Ó 2020 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
https://doi.org/10.1016/j.neurobiolaging.2020.07.029
cortex (Alves et al., 2013; Crutch et al., 2017; Guerrier et al., 2019).
Different patterns in pathological cerebral amyloid and tau distribu-
tion can be found in PCA-AD. The distribution of amyloid pathology
does not differentiate patients with PCA-AD from patients with tAD
(Ossenkoppele et al., 2016; Putcha et al., 2019), but the deposition of
tau pathology in PCA-AD is in accordance with the atrophy of the
parietal, occipital, and temporal brain regions (Lehmann et al., 2013a;
Rabinovici et al., 2008; Rosenbloom et al., 2011). This pattern of tau
deposition is different from tAD where the medio-temporal lobes and
lateral temporoparietal cortex are mostly affected (Crutch et al., 2012;
Ossenkoppele et al., 2016). Furthermore, in PCA-AD, the distinct
patterns of atrophy, tau deposition, and hypometabolism (measured
by fluorodeoxyglucose positron-emission tomography (FDG-PET))
correlate with the level of visual dysfunction (Chen et al., 2019). Even
though neuroimaging has given understanding of the development
of the disease, the pathological mechanisms leading to tAD are not
yet fully understood. Neither is it understood why most patients
develop the typical symptoms of tAD and some the ‘atypical’ symp-
toms of PCA-AD. One explanation of this heterogeneity could be a
difference in vulnerability of distinct functional networks of the brain
network (Cho et al., 2016; Lehmann et al., 2013b; Stam, 2014; Zhou
2 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11
et al., 2012), leading to changes of brain activity in these specific
networks and therefore to distinct clinical symptoms. Changes in
functional brain networks can be captured by electroencephalog-
raphy (EEG), which measures synchronized postsynaptic activity of
the cortical pyramidal neurons (Lopes da Silva, 2013).
EEG and magnetoencephalography (MEG) have been thoroughly
used to characterize AD. A diffuse slowing of the posterior dominant
rhythm is typically seen in patients with AD (Babiloni et al., 2004; de
Waal et al., 2012; Jeong, 2004). In addition, EEG offers a high tem-
poral resolution which enables investigating functional connectivity
and functional networks. Compared with controls, patients with AD
show a general decrease in functional connectivity (Dauwels et al.,
2010; Jeong, 2004) and a loss of communication between func-
tional networks (de Haan et al., 2012b) with changing hub locations
(Engels et al., 2015). EEG research covering PCA-AD specifically is
sparse and mainly consists of a few case studies and 2 retrospective
studies which did not investigate the quantitative differences be-
tween PCA-AD and tAD (Goldstein et al., 2018; Ranasinghe et al.,
2014; Rogelet et al., 1996; Tom et al., 1998; Victoroff et al., 1994).
These studies indicate that abnormalities in EEG recordings in pa-
tients with PCA-AD are likely, but it is unknown whether these
findings are different from what is known in tAD. In this study, we
aim to investigate the EEG oscillatory, functional connectivity, and
network signature of PCA-AD in comparison with tAD and controls.
Because the EEG signal is a representation of synaptic functioning,
we hypothesize a predominant posterior oscillatory slowing in
accordance with the findings of previous research with synaptic
markers such as FDG-PET (Ossenkoppele et al., 2015). Furthermore,
we hypothesize that a selective deterioration of the posterior well-
connected regions in the network, the so called ‘hubs’, in PCA-AD
underlies the oscillatory slowing. Hence, we would expect a rela-
tive sparing in the rest of the functional network.
2. Methods
2.1. Population
Based on availability, we included 29 patients with PCA-AD and
matched 29 patients with tAD and 29 controls with subjective
cognitive decline (SCD) from the Amsterdam Dementia Cohort (van
der Flier and Scheltens, 2018). Patients were matched on the age,
sex, and educational level of the PCA-AD group. Patients for the tAD
group were additionally matched on Clinical Dementia Rating
(CDR) with the PCA-AD group, and only patients with SCD with a
CDR score of 0 were included in the control group. All patients had
been referred to the Alzheimer Center Amsterdam, Amsterdam
UMC. The diagnostic procedure consisted of a standard battery of
investigations including a patient- and informant-based medical
history, physical and neurological examination, magnetic reso-
nance imaging (MRI), cerebrospinal fluid (CSF) analysis, and EEG.
Diagnoses were established by a multidisciplinary team. Patients
with PCA-AD and tAD both fulfilled diagnostic National Institute on
AgingeAlzheimer's Association criteria for probable AD (McKhann
et al., 2011), with biomarker evidence for AD pathology (Ab1-42 <
813 ng/L, t-tau >375 ng/L and p-tau >52 ng/L or positive amyloid
PET imaging) (Tijms et al., 2018; Zwan et al., 2014). Patients with
PCA-AD additionally fulfilled the criteria of Mendez et al. and Tang-
Wai et al. (Mendez et al., 2002; Tang-Wai et al., 2004), and thus
fulfilled the PCA-plus and PCA-AD criteria as defined by the
consensus classification of Crutch et al. (Crutch et al., 2017). Patients
with tAD all presented with initial memory complaints and had
pronounced dysfunction in the memory domain. As a control group,
we included patients who presented with SCD at our memory
clinic, but performed normal on all clinical investigations (i.e.,
criteria for MCI, dementia or psychiatric disorder not fulfilled).
Exclusion criteria for all groups were other significant neurological
diseases and major psychiatric disorders.
2.2. Standard protocol approvals, registrations, and patient
consents
The ethical review board of the VUmc approved this protocol. All
patients gave informed consent for the use of their clinical data for
research purposes.
2.3. Cognitive tasks
We used a standard cognitive test battery, covering the major
cognitive domains (van der Flier and Scheltens, 2018). The visuo-
spatial ability domain was assessed using the dot count task and the
fragmented-letters task. For the dot counting task, patients have to
count the number of unsorted dots presented on a 500 x 300 card
(maximum score ¼ 10). In the fragmented-letters task, patients
identify letters, which increase in level of degradation (maximum
score ¼ 20). Executive functioning and attention were assessed
using the trail-making test (TMT) A and B and the Dutch version of
controlled oral word association test (COWAT). Memory was
assessed by the Dutch version of the Rey auditory verbal learning
test (RAVLT), in which a list of 15 words was repeated 5 times and
the immediate recall was tested after each repetition. The total sum
of correct answers on the immediate recall on each of the 5 trials
was used for the statistical analyses. Higher scores indicate better
cognitive performance on the fragmented-letters task, dot count
task, COWAT, and RAVLT. Lower scores indicate better cognitive
performance on the TMT-A and TMT-B tests.
2.4. EEG recordings
An eyes-closed task-free EEG of 20 minutes was recorded in all
patients with OSG digital equipment (Brainlab and BrainRT; OSG b.v.
Belgium). Twenty-one electrodes were placed in accordance with
the 10e20 international system (channels 1 to 21 respectively
represent the following: Fp2, Fp1, F8, F7, F4, F3, A2, A1 T4, T3, C4, C3,
T6, T5, P4, P3, O2, O1, Fz, Pz, and Cz). Electrode impedance was below
5 kU. Filter settings were as follows: time constant 1 second and low
pass filter 70 Hz. Sample frequency was 500 Hz, and analogedigital
precision was 16e20 bit. During recording, a common reference
electrode was used (average of all electrodes or Fz). Patients were
seated in a slightly reclined chair in a normally lit, quiet room. Pa-
tients sat with their eyes closed and were alerted with acoustic
stimuli when slow horizontal eye movements or slowing of the
posterior alpha rhythm appeared. For each patient, 5 epochs of
10 seconds were visually selected by an experienced neurophysiol-
ogist based on signal quality (artifact-free: no eye-blinks, slow eye
movements, excess muscle activity, or electrocardiogram artifacts)
and patient status (eyes closed, awake). Epochs were converted to
ASCII format. The first 4096 samples of each epoch were used for
further analysis. EEGs were digitally re-referenced offline to the
source reference derivation (Hjorth, 1975). BrainWave software,
written by C.J. Stam (Stam, 2019), was used to calculate spectral
measures, functional connectivity, and the network measures.
2.5. EEG spectral analysis
Relative power was calculated for each EEG channel using the
fast Fourier transform in the following frequency bands: delta
(0.5e4 Hz), theta (4e8 Hz), alpha (8e13 Hz), beta (13e30 Hz). Peak
frequency, defined as the frequency with the highest amplitude in
the 4e13 Hz range, was calculated for each EEG channel. For all
analyses, all 5 epochs of each patient were averaged to obtain a
 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11 3

single value for each spectral measure per patient (van Diessen
et al., 2015). The values of all EEG channels were averaged to
obtain global spectral measures.
2.6. EEG functional connectivity and network measures
Functional connectivity was measured in the alpha band by the
amplitude envelope correlation with correction for volume con-
duction (AEC-c) (Hipp et al., 2012). This measure was chosen
because of its validity (Hipp et al., 2012; Tewarie et al., 2019a), test-
retest reliability (Colclough et al., 2016), and reproducibility in the
alpha frequency band in AD (Briels et al., under review). The AEC-c
is a functional connectivity measure which correlates envelopes of
the amplitude of the signal between channels. The AEC-c was
calculated using the following steps. The raw signal was first band-
pass filtered from 8 to 13 Hz. To correct for volume conduction, this
filtered signal was pairwise orthogonalized in both directions, X to
Y and Y to X. The orthogonalized data were used to calculate the
amplitude envelope correlation (AEC) (Bruns et al., 2000), hence
creating the AEC-c. An adjacency matrix was created with the AEC-c
for each possible pair of channels of each epoch. Global AEC-c was
calculated by averaging all channels, and regional AEC-c was
calculated by averaging per channel. The average of the global and
regional values over 5 epochs was used for each patient.
The network properties of the adjacency matrices of the alpha
band AEC-c were estimated with the minimum spanning tree
(MST). The MST was based on the algorithm of Kruskal (Jackson and
Read, 2010; Kruskal, 1956). In short, the algorithm used creates a
network between all nodes (EEG channels) with the minimum
possible total weight. The algorithm does this by creating the first
edge at the smallest weight (i.e., strongest connection) and repet-
itively adding edges in an order from a small to a large weight,
without creating loops, until all nodes are connected. The resulting
graph is unweighted and undirected. To compare MSTs between
groups, the diameter (longest shortest path of the network), degree
(number of edges to a node), leaf fraction (fraction of nodes with a
degree of 1), and betweenness centrality (fraction of shortest paths
that run through a certain node) of the graph were calculated
(Barabási, 2016; Stam et al., 2014; Tewarie et al., 2015). These results
were averaged over 5 epochs for each patient. The MST diameter
and leaf fraction are indicators of the efficiency of the whole
network. In general, high diameter and low leaf fraction indicate a
less efficient network. MST betweenness centrality and degree are
indicators of how many connections run through a node. In other
words, they show the hubness of a node. Higher degree and higher
betweenness centrality indicate a higher hubness of a node.
2.7. Statistical analysis
Statistical analyses were performed with SPSS (IBM SPSS Statistics
for Windows, version 26.0.0.1) and RStudio ((Allaire, 2012) version
1.1.463) software. Visual representations of the data were made by
using RStudio, BrainNet Viewer ((Xia et al., 2013) version 1.63),
EEGLAB ((Delorme and Makeig, 2004) version 14.1.1), and MATLAB
(The MathWorks, Inc., version R2019b). Differences in group de-
mographics were analyzed with independent t-tests, c2 tests, or
pairwise permutation testing where appropriate. Pairwise permuta-
tion testing with 100,000 iterations was used to test differences be-
tween spectral measures, functional connectivity, and network
measures between the 3 diagnostic groups. Spectral differences be-
tween groups at an electrode level were tested with pairwise per-
mutation testing with additional false discovery rate (FDR) correction
for multiple testing (Benjamini and Hochberg, 1995). Within the PCA-
AD group, Pearson correlations were calculated between regional
peak frequency and cognitive tasks for visuospatial functioning (the
dot counting and fragmented-letters tasks). The cognitive tasks were
chosen because they test performance in the visual domain which is
typically affected in PCA-AD and (relatively) spared in tAD and con-
trols. Peak frequency was chosen because it represents the dominant
oscillatory activity and is a marker of general slowing in AD (Babiloni
et al., 2020; Jeong, 2004). The RAVLT was additionally compared with
the peak frequency to function as a control task. The other cognitive
tasks were regarded less relevant and therefore not analyzed. Hereby
we aimed to limit the chance of false positive results. The threshold
for significance was set at p < 0.05.
3. Results
3.1. Demographics
Table 1 shows the demographics of PCA-AD, tAD, and controls.
Groups were well matched for age and sex. On average, patients
were 63.4 year old, and 59% was female. Both the PCA-AD and tAD
groups had lower CDR and mini-mental state examination scores

Table 1
Demographic and clinical characteristics of the diagnostic groups

Characteristics Control (n ¼ 29) tAD (n ¼ 29) PCA-AD (n ¼ 29)

Age 63.5  5.4 63.3  5.2 63.4  5.4
Sex ratio (f/m) 16/13 18/11 17/12
CDR 0 (0e0) 1 (1e1)a 1 (1e2)b
MMSE 29 (28e30) 18 (16e22)a 19 (16e23)b
Disease duration 2 (1e3) 3 (2e4)a 4 (2e5)b
Educational level 5 (4e6) 5 (4e6) 5 (4e6)
Fragmented-letters task 20 (19e20) (n ¼ 22) 17.5 (12.5e19.0) (n ¼ 16)a 7.0 (2.5e12.5) (n ¼ 25)b,c
Dot counting task 10 (10e10) (n ¼ 22) 10 (8e10) (n ¼ 18)a 7 (4e7) (n ¼ 25)b,c
RAVLT 43 (34e49) (n ¼ 28) 16 (9e19) (n ¼ 24)a 16 (14e23) (n ¼ 27)b
COWAT 39 (30e44) (n ¼ 20) 27 (16e30) (n ¼ 19)a 28 (21e35) (n ¼ 21)b
TMT-A 35 (28e48) (n ¼ 27) 76 (49e130) (¼27)a 200 (147e251) (n ¼ 19)b,c
TMT-B 86 (69e120) (n ¼ 27) 195 (170e247) (n ¼ 11)a 324 (266e453) (n ¼ 6)b,c
Amyloid-beta 42 1082 (996e1210) (n ¼ 19) 657 (592e714) (n ¼ 27)a 642 (561e730) (n ¼ 27)b
Total tau 267 (206e333) (n ¼ 19) 598 (540e962) (n ¼ 27)a 627 (496e1001) (n ¼ 27)b
Phosphorylated tau 45 (35e53) (n ¼ 19) 93 (68e106) (n ¼ 27)a 86 (68e110) (n ¼ 27)b

Data are represented as mean SD or median (IQR). Disease duration is measured as years since onset of complaints. Level of education was rated according to Verhage
(Verhage and Van Der Werff, 1964). TMT-A and TMT-B scores are presented as time needed to complete the task, higher scores mean worse performance. RALVT is presented as
the sum score of 5 trials.
Differences between groups were tested using permutation tests, c2 tests, and independent t-tests, where appropriate. a ¼p < 0.05 between tAD and controls, b ¼ p < 0.05
between PCA-AD and controls, c ¼ p < 0.05 between PCA-AD and tAD.
Key: CDR, clinical dementia rating; COWAT, controlled word association test; IQR, interquartile range; MMSE, mini-mental state examination; RALVT, Ray's auditory verbal
learning test; TMT-A, trail-making test part A; TMT-B, trail-making test part B.
4 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11

Fig. 1. Global relative power and peak frequency measures per diagnosis. Boxplots of global peak frequency and global relative power per frequency band (delta, theta, alpha, and
beta) per diagnosis (SCD, PCA, and AD). The left y-axis represents the amount of relative power in each of the frequency bands. The right y-axis represents the height of the
frequency in Hertz (Hz) for the peak frequency (peak freq). Boxes represent the first to third interquartile range including the median value.

and performed worse on the cognitive test than controls (all p <
0.05). Furthermore, patients with PCA-AD performed worse on the
fragmented-letter, dot count, TMT-A, and TMT-B tasks than patients
with tAD (all p < 0.05). CSF amyloid and tau biomarkers confirmed
the Alzheimer profile in patients with tAD and PCA-AD.
3.2. Spectral analysis
Fig. 1 shows the distribution of the values of the global relative
power and global peak frequency for each group, and Table 2 shows
the results of the pairwise comparisons between the groups.
Overall, controls had spectral characteristics which reflected fastest
oscillatory activity, whereas PCA-AD showed most outspoken
changes from controls indicating slowest oscillatory activity. Pa-
tients with tAD showed spectral characteristics intermediate to
controls and PCA-AD. More specifically, patients with PCA-AD
showed significant differences in all spectral power measures
compared with controls, reflected by lower global peak frequency
(difference of medians (Dm) ¼ 2.18, p < 0.0001), decreases of
faster oscillatory activity (relative alpha power (Dm ¼ 0.16, p <
0.0001), relative beta power (Dm ¼ 0.05, p ¼ 0.02)), and increases
of slower activity (relative theta (Dm ¼ 0.14, p < 0.0001) and delta
power (Dm ¼ 0.07, p < 0.01)). Moreover, patients with PCA-AD
showed more severe oscillatory slowing than patients with tAD,
as indicated by more slow activity (relative delta power (Dm ¼ 0.05,
p < 0.05)) and less fast activity (relative alpha power (Dm ¼ 0.02,
p < 0.05)). In addition, patients with tAD had oscillatory power
results intermediate to controls and PCA-AD, where we found
significantly lower global peak frequency (Dm ¼ 1.84 Hz, p <
0.001), relative alpha power (Dm ¼ 0.14, p < 0.0001), and higher
relative theta (Dm ¼ 0.10, p < 0.0001) (but not delta) power
compared with controls.
Electrode level spectral characteristics gave additional insight on
the global changes. Overall, patients with PCA-AD showed the most

Table 2
Global relative power and peak frequency measures per diagnosis

Measure (median, IQR) Control (n ¼ 29) tAD (n ¼ 29) PCA-AD (n ¼ 29)

Global PF (Hz) 9.41 (8.86e9.81) 7.57 (6.51e8.63)a 7.23 (6.61e7.77)b
Global delta power 0.27 (0.21e0.34) 0.28 (0.24e0.34) 0.34 (0.28e0.41)b,c
Global theta power 0.09 (0.07e0.11) 0.19 (0.14e0.25)a 0.23 (0.18e0.28)b
Global alpha power 0.32 (0.28e0.45) 0.18 (0.14e0.27)a 0.16 (0.11e0.21)b,c
Global beta power 0.21 (0.17e0.25) 0.17 (0.13e0.25)a 0.16 (0.12e0.23)b

For each group, the median and interquartile range (IQR) are shown. Significant differences (p < 0.05) between groups are indicated by a (control versus tAD), b (PCA-AD versus
control), and c (PCA-AD versus tAD).
Key: IQR, interquartile range; PF, peak frequency.
 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11 5

Fig. 2. Peak frequency at electrode level per diagnosis. Topological representation of median peak frequency (Hz) at each electrode plotted per diagnosis. Red indicates a relatively
high peak frequency, whereas blue indicates a relatively low peak frequency. In both tAD and PCA-AD, there is a global slowing compared with controls. Compared with tAD, the
PCA-AD group shows a more severe slowing of the parietal, posterior temporal, and occipital channels. (For interpretation of the references to color in this figure legend, the reader
is referred to the Web version of this article.)

profound slowing of the posterior channels, whereas the other re- regional relative theta and beta power were found between tAD
gions had values equal to tAD. Fig. 2 shows peak frequency at the and PCA-AD.
electrode level for each group, and electrode level relative band
power is shown in supplementary Fig. 1. The controls showed a
posterior dominant peak frequency higher than 9 Hz. When
3.3. Functional connectivity
comparing tAD with controls, the patients with tAD showed a
significantly lower peak frequency in both frontal (around 7 Hz) and
Functional connectivity was estimated by the AEC-c in the alpha
posterior channels (around 8 Hz) (decreased peak frequency in all
frequency band on a global and regional level. Fig. 3 shows the
21 channels, pcorr < 0.01). The patients with PCA-AD also showed a
functional connectivity matrices of the 3 groups. The differences
significant decrease in peak frequency in all channels compared
between these matrices and the p-values of the permutation tests
with controls (all 21 channels, pcorr < 0.01), and importantly,
corresponding to these matrices are shown in supplementary Fig. 2.
compared with tAD, there was an even lower peak frequency in the
Both the tAD and PCA-AD groups showed a decrease in global alpha
posterior channels (below 7 Hz) with similar values in the frontal
band AEC-c compared with the control group (respectively
channels (around 7 Hz) (respectively O1 Dm ¼ 1.98 Hz, pcorr ¼ 0.01;
Dm ¼ 0.023, p ¼ 0.002; Dm ¼ 0.031, p < 0.001). However, there
O2 Dm ¼ 1.95 Hz, pcorr ¼ 0.01; P4 Dm ¼ 2.39 Hz, pcorr ¼ 0.02).
was no significant difference in global functional connectivity be-
Furthermore, regional analyses of relative band power
tween the tAD and PCA-AD groups (Dm ¼ 0.008, p ¼ 0.49). Regional
(supplementary Fig. 1) supported the observation that the more
analyses of functional connectivity showed similar results. When
severe global slowing in patients with PCA-AD compared with
comparing controls with tAD, all regional functional connectivity
patients with tAD was driven by more severe changes in the pos-
values, except channel F3, T4, and A2, were significantly reduced (p
terior regions. The PCA-AD group showed, after FDR correction,
< 0.05 after FDR correction). Similarly, all regional functional con-
significantly higher regional relative delta (channels T6, P3, O1, and
nectivity values were significantly reduced in PCA-AD compared
O2), lower alpha (channels T5, T6, O1, and O2) compared with tAD.
with controls. No regional differences were found between PCA and
In congruence with the analyses on a global level, no differences in
tAD.

Fig. 3. Functional connectivity matrices per subject group. Pairwise alpha band AEC-c values between channels. Each channel is represented on the x-axis (1e21) and y-axis (21e1).
Dark red represents relative high and dark blue relative low AEC-c values between the corresponding channels. Both the tAD and PCA-AD groups show a global decrease in
functional connectivity compared with controls. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
6 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11
Fig. 4. Minimum spanning tree per diagnosis. Minimum spanning trees based on the alpha band AEC-c adjacency matrix for each group. Each surface electrode represents a node of
the network. For each diagnosis, left, right, and top-down views are shown. The MST of the control group shows posterior hubs connecting to the frontal nodes where the tAD and
PCA-AD groups show more frontal hubs and less connected posterior nodes.
3.4. Network measures
The functional networks were estimated by the MST based on
the adjacency matrix of the alpha band AEC-c. The average MST of
each group is shown in Fig. 4. Differences in global MST charac-
teristics (diameter and leaf fraction) are shown in Table 3. The PCA-
AD group MST had a longer diameter (Dm ¼ 0.05, p ¼ 0.008) and
smaller leaf fraction (Dm ¼ 0.06, p < 0.001) than the control
group, indicating a less efficient network topology. The tAD group
showed a significant smaller leaf fraction than the control group
(Dm ¼ 0.04, p ¼ 0.02), but the diameter was not significantly
longer (Dm ¼ 0.03, p ¼ 0.20). No significant differences were found
between the tAD and PCA-AD groups. tAD values were, on average,
intermediates between controls and PCA-AD.
Differences in regional MST characteristics (degree and
betweenness centrality) are shown in Fig. 5. Both the PCA-AD and
tAD groups showed a lower maximum degree than controls
(respectively Dm ¼ 0.03, p < 0.001), indicating that the most
connected regions, the hubs, have weakened. In PCA-AD, compared
with controls, the degree showed a significant (p < 0.05) increase in
frontal regions (Fp1, F3, F4, and F7) and a decrease in parietal re-
gions (P3, Cz, and Pz). After FDR correction, however, none of the
values reached statistical significance. The betweenness centrality
was increased in the frontal regions (F3, F4, and F7) in PCA-AD
compared with controls. After FDR correction, only region F3
reached statistical significance. The regional MST characteristics of
the tAD group showed intermediate values between controls and
PCA-AD. Although globally the tAD group showed a lower
maximum degree and betweenness centrality than controls
(respectively Dm ¼ 0.02, p < 0.01; Dm ¼ 0.02, p < 0.01), regional
differences were not significantly different from the control or PCA-
AD group, which was potentially caused by a lack of power.
3.5. Correlations with cognitive tasks
Correlations between peak frequency of the individual EEG
channels and the scores of the dot counting task, fragmented-
Table 3
Minimum spanning tree characteristics per diagnosis
MST measure (median, IQR) Control (n ¼ 29)
Diameter 0.41 (0.40e0.44)
Leaf fraction 0.56 (0.54e0.58)
Maximum degree 0.28 (0.26e0.30)
Maximum BC 0.71 (0.70e0.74)
For each group, the median and interquartile range (IQR) are shown.
Significant differences (p < 0.05) between groups are indicated by a (control versus tAD), b (PCA-AD versus control), and c (PCA-AD versus tAD).
Key: BC, betweenness centrality; IQR, interquartile range.
letters task, and RAVLT were assessed within the PCA-AD group.
Regional plots and the individual data points of the correlations are
shown in Fig. 6. The fragmented-letters task was significantly
correlated with right parietal peak frequency (channel P4, r ¼ 0.51,
p ¼ 0.008), in which a higher peak frequency was associated with a
higher score on the task. The dot counting task scores significantly
correlated with the peak frequency of the right temporal-parietal
region (significant correlations with channel P4 and T4 individu-
ally, correlation of combined channels: r ¼ 0.50, p ¼ 0.012). Again, a
higher regional peak frequency of these channels correlated with a
higher score on the task. No significant correlations were found
between the scores of the RAVLT and the peak frequency of chan-
nels P4 and T4 (channel P4 and T4 combined r ¼ 0.10, p ¼ 0.614).
Neither did the peak frequency of any of the other channels
correlate with the RAVLT.
4. Discussion
The results of this study indicate that patients with PCA-AD have
severely affected oscillatory EEG activity together with disrupted
functional connectivity and less efficient network properties. Pa-
tients with PCA-AD were more severely affected than typical age,
sex, educational level, and disease severity matched patients with
tAD with respect to brain oscillatory activity. Spectral changes were
characterized by a pronounced decrease in posterior alpha power
and an increase in posterior delta power. In addition, right (tem-
poral-) parietal peak frequency correlated with performance on
visual tasks. Functional connectivity was decreased compared with
controls but did not differ compared with patients with tAD.
Network properties indicate a less integrated network in patients
with PCA-AD, weakened hubs, and a relative shift in hubness from
posterior to frontal brain areas compared with controls, indicating a
selective deterioration of the posterior hubs with relative sparing of
the frontal hubs.
The comparison of EEG spectral measures between patients
with tAD and controls confirmed earlier studies of EEG slowing in
tAD. This was represented by a higher global relative theta power
tAD (n ¼ 29) PCA-AD (n ¼ 29)
0.44 (0.42e0.46) 0.46 (0.42e0.47)b
0.52 (0.50e0.56)a 0.50 (0.49e0.54)b
0.26 (0.23e0.28)a 0.25 (0.23e0.26)b
0.69 (0.68e0.72)a 0.69 (0.67e0.71)b
 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11 7

Fig. 5. MST betweenness centrality and degree at electrode level per patient group. Topological representation of MST betweenness centrality (top row) and degree (bottom row)
per subject group (columns). Red indicates relatively high values, and blue indicates relatively low values. In controls, the posterior channels show the most connections with a
relative high degree and betweenness centrality. In PCA-AD, the posterior distribution has disappeared and the frontal channels have a relatively higher betweenness centrality and
degree. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

and lower relative alpha and beta power in the patients with tAD
(de Waal et al., 2012; Jeong, 2004). We extend on this by showing
that patients with PCA-AD had more severe slowing of the oscil-
latory brain rhythms than patients with tAD. More specifically,
patients with PCA-AD showed similar changes as patients with tAD
in global relative theta and beta power, but in addition showed
higher global relative delta power and lower alpha power. This
difference was largely driven by a pronounced decrease in the
dominant frequency of the posterior regions. Compared with con-
trols, both patients with PCA-AD and tAD showed a global slowing
of the oscillatory activity, but the slowing of the parietal, posterior
temporal, and occipital channels was more severe in PCA-AD. These
findings support our hypothesis that slowing of oscillatory brain
rhythms in patients with PCA-AD was most pronounced in the
posterior part of the brain. This indicates that there is a profound
loss of synapses in these areas (Lopes da Silva, 2013). Although the
spatial resolution of EEG does not allow for detailed conclusions at
the level of functional subnetworks, the regions reported in this
study are largely congruent with the visuospatial network, involved
in PCA-AD. This predominant deterioration of the posterior regions
in PCA-AD is also reported by studies that show posterior glucose
hypometabolism, posterior gray matter atrophy, and higher pos-
terior tau deposition in patients with PCA-AD (Ossenkoppele et al.,
2016; Whitwell et al., 2017). The causational relation between
these markers remains unclear. Perhaps the deterioration of func-
tional outcome measures like EEG oscillatory activity and FDG up-
take are driven by regional tau-mediated neurodegeneration. On
the other hand, regional tau accumulation could also be driven by
regional network vulnerability (de Haan et al., 2012a). Therefore,
we looked more closely at the underlying functional network of
each group.
The analyses of the functional connectivity showed that both
patients with tAD and PCA-AD had decreased global alpha band
functional connectivity compared with controls. Although patients
with PCA-AD have more severe visual dysfunction, we could not
find a difference in functional connectivity compared with the pa-
tients with tAD. Characteristics of functional network topology gave
more information into the difference between the 2 diagnoses. The
networks were compared using the diameter, leaf fraction, regional
degree, and regional betweenness centrality of the MST. Less inte-
grated (or less efficient) networks generally show a longer diameter
and a loss of hubs (Stam et al., 2014). Hubs are locally highly con-
nected nodes, allowing short path lengths from one node to the
other, which makes the network more efficient, or ‘compact’. In our
study, patients with PCA-AD showed the highest MST diameter and
lowest MST leaf fraction and therefore the least efficient network
topology, compared with the other groups. Moreover, an indication
of weakened parietal hub region (lower degree) and stronger
frontal hubness (increase in frontal betweenness centrality and
degree) was observed in PCA-AD. In summary, our results indicated
that the functional network in PCA-AD becomes less integrated
because of a global loss of connectivity together with a selective loss
of function of the posterior hubs with a relative increase of frontal
hubness. These results were in line with our hypothesis, and
perhaps the selective vulnerability of the posterior network could
explain the clinical and pathological characteristics of PCA-AD.
Unfortunately, no EEG or MEG functional network studies have
been performed in patients with PCA-AD to compare our functional
network results with. Studies using functional MRI (fMRI) have,
however, shown similar patterns in network changes in PCA-AD. One
study has shown increased connectivity in the dorsal inferior
network in correlation with the amount of parietal atrophy in
8 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11

Fig. 6. Correlation between regional peak frequency and cognitive tasks. Correlation coefficients, within the PCA-AD group, between individual EEG channel peak frequency and the
dot count task (column A), fragmented-letters task (column B), and RAVLT (sum of 5 trials, column C) score are shown in headplots. The strength of the correlation is represented by
the area color. Black indicates no correlation where white indicates a perfect correlation. The individual data points of the significant correlations (T4 and P4 in A, P4 in B) are plotted
on the bottom row. No channels significantly correlated in C and the data points represent the relevant channels of A and B (T4 and P4). The 2 channels in A and C were averaged to
create a single plot for both channels. A decrease in right posterior peak frequency correlated with worse performance on dot counting and fragmented-letters tasks. No correlation
was found with performance on the memory task. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

patients with PCA-AD (Migliaccio et al., 2016). Another fMRI study
has shown an increased connectivity in the anterior default mode
network and a decrease in the visual network in patients with PCA-
AD (Lehmann et al., 2015). And finally, a third study has found a
similar pattern in PCA-AD with an increase in the frontal salience
network and default mode network accompanied by a decreased
visuospatial network connectivity (Fredericks et al., 2019). This last
study hypothesized the medial and lateral pulvinar of the thalamus
to coordinate compensatory shifts across the functional networks.
Where the lateral pulvinar showed decreased connectivity with the
visual network, the medial pulvinar showed increased connectivity
with the salience network. The results of these fMRI studies are in
line with the increase in frontal and decrease in posterior hubness
we have observed. Drawing relations or conclusions between (f)MRI
networks and EEG networks should, however, be carried out with
caution. Where fMRI has a high and EEG a low spatial resolution, EEG
has a high and fMRI a low temporal resolution. With this in mind, our
results combined with previous fMRI research, indicate a deterio-
ration of the posterior visual network with a, potentially compen-
satory or relative, shift of hubness toward the frontal network in
PCA-AD. Frontal hubs may partially take over the function as pos-
terior hubs get disconnected. However, a relative loss of posterior
connectivity and relative survival of frontal connectivity could have
driven the MST estimation. This might have created a change of hubs
in the network because it was based on the selection of the strongest
connections. Our results confirmed that there is a selective deterio-
ration of the posterior hubs in PCA-AD. A selective vulnerability of
the visual network could explain the clinical presentation of PCA-AD
(Lehmann et al., 2013b). To further investigate the hypothesis of
network vulnerability, a longitudinal study would be needed to not
only study the relation between network properties and clinical
progression but also try to predict the clinical progression with
network properties. Another interesting design would be to combine
longitudinal EEG and tau-PET data to observe the relation between
tau accumulation with network deterioration. This could answer
the question whether there is network deterioration due to
tau-mediated neurodegeneration or whether tau accumulation is
driven by network vulnerability.
Within the PCA-AD group, a decrease in right posterior peak
frequency was correlated with worse visual performance, assessed
with the fragmented-letters and dot counting task, indicating that
these effects are likely to be specific for PCA-AD. These correlations
are in line with a previous study. For example, in a study by Guerrier
et al. (Guerrier et al., 2019), who found similar correlations of visual
performance, tested by various visual tasks and including the dot
counting task, with right posterior [18F]FDG-PET metabolism in
patients with PCA. Both our study and the study by Guerrier et al.
have found significant correlations of the visual performance with
the right posterior cortex but not the left posterior cortex. Another
study, using FDG-PET, found this right predominant asymmetry of
the posterior cortex in PCA as well (Nestor et al., 2003). In conclu-
sion, our results indicate that the performance of PCA-AD on visual
tasks is strongly dependent on the functionality of the right pos-
terior cortex.
In literature, 2 retrospective studies using EEG or MEG to char-
acterize patients with PCA-AD have been reported. The first MEG
study addressed cognitive functioning across different AD subtypes,
including 7 patients with PCA, with functional connectivity esti-
mated by the alpha band imaginary coherence (Ranasinghe et al.,
2014). This study did, however, not compare cognitive functioning
 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11
with functional connectivity within the patients with PCA-AD, nor
did it look at differences between different AD subtypes at spectral,
functional connectivity, or network level. The second study is a
retrospective EEG study (Goldstein et al., 2018) on a subset of data
from a larger set of patients with PCA. EEG data were available for
23 patients. In this study, the EEG characteristics were only assessed
by visual rating and not quantitatively. Of 23 patients with PCA, 17
showed an abnormal EEG with either generalized or focal slowing
of the background rhythm. In addition, 2 patients showed gener-
alized or temporal discharges. Our results confirm that patients
with PCA-AD have profound slowing of the background rhythm,
and we further defined the regional characteristics in comparison
with tAD. We suspect few MEG or EEG studies have been carried out
because of the relatively rareness of the clinical syndrome of PCA-
AD. But nonetheless, EEG is a minimal invasive, cost-efficient, and
widely available method, and these results highlight its potential in
further fundamental or trial-based research.
In this study, patients with tAD and PCA-AD were matched on
age, sex, educational level, and CDR as a marker of disease severity.
The main difference in performance on cognitive tasks between the
patients with tAD and the patients with PCA-AD was their perfor-
mance on visual tasks. This was reflected by the worse performance
of patients with PCA-AD on the dot counting and fragmented-
letters tasks compared with patients with tAD. Patients with PCA-
AD and patients with tAD performed similar on the mini-mental
state examination, RAVLT, and COWAT tasks. It is perhaps not
contradictory that patients with PCA-AD performed worse, than
patients with tAD, on the TMT-A and TMT-B tasks, but no difference
was seen on the COWAT task. Although the TMT-A and TMT-B are
executive tasks, they also rely on visual functioning (SÁNchez-
Cubillo et al., 2009). The COWAT is also a task of executive func-
tioning but focusses on verbal fluency without the need of visual
functions (Ross et al., 2007). When taking the differences and
similarities in cognitive performance into account, the main
contributor of the reported EEG differences is likely to be the
pathological heterogeneity of these AD subtypes rather than an
effect of disease severity.
Strengths of our study include well-defined clinical subgroups
with availability of biomarkers and cognitive test results. Further-
more, we have investigated the EEG recordings in 3 modalities:
oscillatory activity, functional connectivity, and network properties.
We have used validated outcome measures to estimate these mo-
dalities. A limitation of our study is the use of patients with SCD as
controls. CSF results show, however, that these individuals were
amyloid and tau negative which rules out influence of Alzheimer's
disease on the results of the control group. Additional limitations
can be found in the materials and methods of recording and
analyzing the EEGs in this study. Several factors can influence the
resolution and validity of the functional network estimations. The
use of an EEG system with 21 electrodes has limited the resolution
of our network estimations. Future studies could consider the use of
high density EEG or MEG which provides more potential in terms of
source reconstruction and defining more detailed regions of inter-
est. In this article, we have chosen the AEC-c to estimate functional
connectivity. This measure is an amplitude-based pairwise measure
of functional connectivity. As this article was the first to describe
EEG functional networks in PCA-AD, this choice was made based on
the reported reliability, validity, and reproducibility of this measure
of functional connectivity (Briels et al., 2020; Colclough et al., 2016;
Hipp et al., 2012). However, previous research has indicated that
amplitude coupling might only capture one aspect of functional
connectivity (Engel et al., 2013). We encourage future studies to use
other methods including phase-based measures, phase-amplitude
coupling, dynamic connectivity, and multivariate approaches.
These measures could give more detailed information on other
9
modalities of functional connectivity, the direction of flow and the
influence of common drive effects (Kus et al., 2004; Reid et al., 2019;
Seymour et al., 2017; Stam et al., 2007; Tewarie et al., 2019b). A
different limitation is the relatively small sample size of this study.
PCA-AD is a rare clinical syndrome and even inclusion in a large
clinical center is low. Our study was the first to investigate quan-
titative EEG characteristics of patients with PCA-AD, and further
research is needed to confirm these findings. The profound poste-
rior changes found with our EEG recordings potentially give ground
to use EEG as a syndrome classifier, but larger groups are needed to
perform such investigations.
5. Conclusion
This study found that patients with PCA-AD were more severely
affected than typical age- and gender-matched patients with AD
with respect to brain oscillatory activity, driven by slowing in the
posterior regions, and network properties. Visual performance
correlated with the oscillatory activity of the right (temporal-) pa-
rietal cortex. This reflects the selective deterioration of the parietal-
occipital regions in PCA-AD, which is potentially caused by local
vulnerability of the underlying network. In addition, these results
indicate that EEG is a potential sensitive modality for future
fundamental or trial-based research in PCA-AD.
Disclosure statement
First author C.T. Briels declares no conflict of interest. J.J. Eertink
and C.J. Stam declare no conflict of interest. The chair of W. van der
Flier is supported by the Pasman stichting. P. Scheltens has received
consultancy/speaker fees (paid to the institution) from Biogen,
People Bio, Roche (Diagnostics), Novartis Cardiology. He is PI of
studies with Vivoryon, EIP Pharma, IONIS, CogRx, AC Immune and
Toyama. A.A. Gouw has received past (2014e2018) research support
from Probiodrug and Boehringer Ingelheim via the VUmc Alz-
heimer Center.
CRediT authorship contribution statement
Casper T. Briels: Conceptualization, Methodology, Formal anal-
ysis, Writing - original draft, Visualization. Jakoba J. Eertink:
Conceptualization, Methodology, Validation, Writing - original
draft. Cornelis J. Stam: Resources, Conceptualization, Writing -
review & editing. Wiesje M. van der Flier: Resources, Writing -
review & editing. Philip Scheltens: Resources, Writing - review &
editing. Alida A. Gouw: Conceptualization, Writing - review &
editing, Supervision.
Acknowledgements
Research of Alzheimer center Amsterdam is part of the neuro-
degeneration research program of Amsterdam Neuroscience. Alz-
heimer Center Amsterdam is supported by Stichting Alzheimer
Nederland and Stichting VUmc fonds. The clinical database struc-
ture was developed with funding from Stichting Dioraphte. W. van
der Flier is a recipient of a ZonMW Top grant (project #91217043).
A.A. Gouw is a recipient of a ZonMW Memorabel grant (#
733050812) and a recipient of a ZonMW Top grant (#91218018).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.neurobiolaging.2020.07.029.
10 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11

References Guerrier, L., Cransac, C., Pages, B., Saint-Aubert, L., Payoux, P., Péran, P., Pariente, J.,
2019. Posterior cortical atrophy: does complaint match the impairment? A
neuropsychological and FDG-PET study. Front Neurol. 10, 1010.
Allaire, J., 2012. RStudio, 537. integrated development environment for R, Boston,
Hipp, J.F., Hawellek, D.J., Corbetta, M., Siegel, M., Engel, A.K., 2012. Large-scale
MA, p. 538.
cortical correlation structure of spontaneous oscillatory activity. Nat. Neurosci.
Alves, J., Soares, J.M., Sampaio, A., Gonçalves, Ó.F., 2013. Posterior cortical atrophy
15, 884e890.
and Alzheimer’s disease: a meta-analytic review of neuropsychological and
Hjorth, B., 1975. An on-line transformation of EEG scalp potentials into orthogonal
brain morphometry studies. Brain Imaging Behav. 7, 353e361.
source derivations. Electroencephalogr. Clin. Neurophysiol. 39, 526e530.
Babiloni, C., Barry, R.J., Basar, E., Blinowska, K.J., Cichocki, A., Drinkenburg, W.,
Jackson, T.S., Read, N., 2010. Theory of minimum spanning trees. I. Mean-field
Klimesch, W., Knight, R.T., Lopes da Silva, F., Nunez, P., Oostenveld, R., Jeong, J.,
theory and strongly disordered spin-glass model. Phys. Rev. E 81, 021130.
Pascual-Marqui, R., Valdes-Sosa, P., Hallett, M., 2020. International Federation of
Jeong, J., 2004. EEG dynamics in patients with Alzheimer's disease. Clin. Neuro-
Clinical Neurophysiology (IFCN) - EEG research workgroup: recommendations
physiol. 115, 1490e1505.
on frequency and topographic analysis of resting state EEG rhythms. Part 1:
Kruskal, J.B., 1956. On the shortest spanning subtree of a graph and the traveling
applications in clinical research studies. Clin. Neurophysiol. 131, 285e307.
salesman problem. Proc. Am. Math. Soc. 7, 48e50.
Babiloni, C., Binetti, G., Cassetta, E., Cerboneschi, D., Dal Forno, G., Del Percio, C.,
Kus, R., Kaminski, M., Blinowska, K.J., 2004. Determination of EEG activity propa-
Ferreri, F., Ferri, R., Lanuzza, B., Miniussi, C., Moretti, D.V., Nobili, F., Pascual-
gation: pair-wise versus multichannel estimate. IEEE Trans. Biomed. Eng. 51,
Marqui, R.D., Rodriguez, G., Romani, G.L., Salinari, S., Tecchio, F., Vitali, P.,
1501e1510.
Zanetti, O., Zappasodi, F., Rossini, P.M., 2004. Mapping distributed sources of
Lehmann, M., Ghosh, P.M., Madison, C., Laforce Jr., R., Corbetta-Rastelli, C.,
cortical rhythms in mild Alzheimer's disease. A multicentric EEG study. Neu-
Weiner, M.W., Greicius, M.D., Seeley, W.W., Gorno-Tempini, M.L., Rosen, H.J.,
roimage 22, 57e67.
Miller, B.L., Jagust, W.J., Rabinovici, G.D., 2013a. Diverging patterns of amyloid
Barabási, A.-L., 2016. Network Science. Cambridge university press. http://
deposition and hypometabolism in clinical variants of probable Alzheimer's
networksciencebook.com/last. Accessed 1 July 2020.
disease. Brain 136 (Pt 3), 844e858.
Beh, S.C., Muthusamy, B., Calabresi, P., Hart, J., Zee, D., Patel, V., Frohman, E., 2015.
Lehmann, M., Madison, C., Ghosh, P.M., Miller, Z.A., Greicius, M.D., Kramer, J.H.,
Hiding in plain sight: a closer look at posterior cortical atrophy. Pract. Neurol. 15,
Coppola, G., Miller, B.L., Jagust, W.J., Gorno-Tempini, M.L., Seeley, W.W.,
5e13.
Rabinovici, G.D., 2015. Loss of functional connectivity is greater outside the
Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate - a practical
default mode network in nonfamilial early-onset Alzheimer's disease variants.
and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Stat. Methodol.
Neurobiol. Aging 36, 2678e2686.
57, 289e300.
Lehmann, M., Madison, C.M., Ghosh, P.M., Seeley, W.W., Mormino, E., Greicius, M.D.,
Briels, C.T., Schoonhoven, D.N., Stam, C.J., de Waal, H., Scheltens, P., Gouw, A.A., 2020.
Gorno-Tempini, M.L., Kramer, J.H., Miller, B.L., Jagust, W.J., Rabinovici, G.D.,
Reproducibility of EEG functional connectivity in Alzheimer's disease. Alz-
2013b. Intrinsic connectivity networks in healthy subjects explain clinical
heimers Res. Ther. 12, 68.
variability in Alzheimer's disease. Proc. Natl. Acad. Sci. U. S. A. 110, 11606e11611.
Bruns, A., Eckhorn, R., Jokeit, H., Ebner, A., 2000. Amplitude envelope correlation
Lopes da Silva, F., 2013. EEG and MEG: relevance to neuroscience. Neuron 80,
detects coupling among incoherent brain signals. Neuroreport 11, 1509e1514.
1112e1128.
Chen, Y., Liu, P., Wang, Y., Peng, G., 2019. Neural mechanisms of visual dysfunction in
McKhann, G.M., Knopman, D.S., Chertkow, H., Hyman, B.T., Jack Jr., C.R.,
posterior cortical atrophy. Front Neurol. 10.
Kawas, C.H., Klunk, W.E., Koroshetz, W.J., Manly, J.J., Mayeux, R., Mohs, R.C.,
Cho, H., Choi, J.Y., Hwang, M.S., Kim, Y.J., Lee, H.M., Lee, H.S., Lee, J.H., Ryu, Y.H.,
Morris, J.C., Rossor, M.N., Scheltens, P., Carrillo, M.C., Thies, B., Weintraub, S.,
Lee, M.S., Lyoo, C.H., 2016. In vivo cortical spreading pattern of tau and amyloid
Phelps, C.H., 2011. The diagnosis of dementia due to Alzheimer's disease:
in the Alzheimer disease spectrum. Ann. Neurol. 80, 247e258.
recommendations from the National Institute on Aging-Alzheimer's Associa-
Colclough, G.L., Woolrich, M.W., Tewarie, P.K., Brookes, M.J., Quinn, A.J., Smith, S.M.,
tion workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers
2016. How reliable are MEG resting-state connectivity metrics? Neuroimage
Dement. 7, 263e269.
138, 284e293.
Mendez, M.F., Ghajarania, M., Perryman, K.M., 2002. Posterior cortical atrophy:
Crutch, S.J., Lehmann, M., Schott, J.M., Rabinovici, G.D., Rossor, M.N., Fox, N.C., 2012.
clinical characteristics and differences compared to Alzheimer's disease.
Posterior cortical atrophy. Lancet Neurol. 11, 170e178.
Dement Geriatr. Cogn. Disord. 14, 33e40.
Crutch, S.J., Schott, J.M., Rabinovici, G.D., Boeve, B.F., Cappa, S.F., Dickerson, B.C.,
Migliaccio, R., Agosta, F., Toba, M.N., Samri, D., Corlier, F., de Souza, L.C., Chupin, M.,
Dubois, B., Graff-Radford, N.R., Krolak-Salmon, P., Lehmann, M., Mendez, M.F.,
Sharman, M., Gorno-Tempini, M.L., Dubois, B., Filippi, M., Bartolomeo, P., 2012.
Pijnenburg, Y., Ryan, N.S., Scheltens, P., Shakespeare, T., Tang-Wai, D.F., van der
Brain networks in posterior cortical atrophy: a single case tractography study
Flier, W.M., Bain, L., Carrillo, M.C., Fox, N.C., 2013. Shining a light on posterior
and literature review. Cortex 48, 1298e1309.
cortical atrophy. Alzheimers Dement. 9, 463e465.
Migliaccio, R., Gallea, C., Kas, A., Perlbarg, V., Samri, D., Trotta, L., Michon, A.,
Crutch, S.J., Schott, J.M., Rabinovici, G.D., Murray, M., Snowden, J.S., van der
Lacomblez, L., Dubois, B., Lehericy, S., Bartolomeo, P., 2016. Functional connec-
Flier, W.M., Dickerson, B.C., Vandenberghe, R., Ahmed, S., Bak, T.H., Boeve, B.F.,
tivity of ventral and dorsal visual streams in posterior cortical atrophy.
Butler, C., Cappa, S.F., Ceccaldi, M., de Souza, L.C., Dubois, B., Felician, O.,
J. Alzheimers Dis. 51, 1119e1130.
Galasko, D., Graff-Radford, J., Graff-Radford, N.R., Hof, P.R., Krolak-Salmon, P.,
Nestor, P.J., Caine, D., Fryer, T.D., Clarke, J., Hodges, J.R., 2003. The topography of
Lehmann, M., Magnin, E., Mendez, M.F., Nestor, P.J., Onyike, C.U., Pelak, V.S.,
metabolic deficits in posterior cortical atrophy (the visual variant of Alzheimer's
Pijnenburg, Y., Primativo, S., Rossor, M.N., Ryan, N.S., Scheltens, P.,
disease) with FDG-PET. J. Neurol. Neurosurg. Psychiatry 74, 1521e1529.
Shakespeare, T.J., Suarez Gonzalez, A., Tang-Wai, D.F., Yong, K.X.X., Carrillo, M.,
Ossenkoppele, R., Schonhaut, D.R., Baker, S.L., O'Neil, J.P., Janabi, M., Ghosh, P.M.,
Fox, N.C., Alzheimer's Association, I.A.A.s.D., Associated Syndromes Professional
Santos, M., Miller, Z.A., Bettcher, B.M., Gorno-Tempini, M.L., Miller, B.L.,
Interest, A., 2017. Consensus classification of posterior cortical atrophy. Alz-
Jagust, W.J., Rabinovici, G.D., 2015. Tau, amyloid, and hypometabolism in a pa-
heimers Dement. 13, 870e884.
tient with posterior cortical atrophy. Ann. Neurol. 77, 338e342.
Dauwels, J., Vialatte, F., Cichocki, A., 2010. Diagnosis of Alzheimers disease from EEG
Ossenkoppele, R., Schonhaut, D.R., Scholl, M., Lockhart, S.N., Ayakta, N., Baker, S.L.,
signals: where are we standing? Curr. Alzheimer Res. 7, 487e505.
O'Neil, J.P., Janabi, M., Lazaris, A., Cantwell, A., Vogel, J., Santos, M., Miller, Z.A.,
de Haan, W., Mott, K., van Straaten, E.C., Scheltens, P., Stam, C.J., 2012a. Activity
Bettcher, B.M., Vossel, K.A., Kramer, J.H., Gorno-Tempini, M.L., Miller, B.L.,
dependent degeneration explains hub vulnerability in Alzheimer's disease. PLoS
Jagust, W.J., Rabinovici, G.D., 2016. Tau PET patterns mirror clinical and neuro-
Comput. Biol. 8, e1002582.
anatomical variability in Alzheimer's disease. Brain 139 (Pt 5), 1551e1567.
de Haan, W., van der Flier, W.M., Koene, T., Smits, L.L., Scheltens, P., Stam, C.J., 2012b.
Putcha, D., Brickhouse, M., Touroutoglou, A., Collins, J.A., Quimby, M., Wong, B.,
Disrupted modular brain dynamics reflect cognitive dysfunction in Alzheimer's
Eldaief, M., Schultz, A., El Fakhri, G., Johnson, K., Dickerson, B.C., McGinnis, S.M.,
disease. Neuroimage 59, 3085e3093.
2019. Visual cognition in non-amnestic Alzheimer's disease: relations to tau,
de Waal, H., Stam, C.J., de Haan, W., van Straaten, E.C., Scheltens, P., van der
amyloid, and cortical atrophy. Neuroimage Clin. 23, 101889.
Flier, W.M., 2012. Young Alzheimer patients show distinct regional changes of
Rabinovici, G.D., Jagust, W.J., Furst, A.J., Ogar, J.M., Racine, C.A., Mormino, E.C.,
oscillatory brain dynamics. Neurobiol. Aging 33, 1008.e1025-1031.
O'Neil, J.P., Lal, R.A., Dronkers, N.F., Miller, B.L., Gorno-Tempini, M.L., 2008. Abeta
Delorme, A., Makeig, S., 2004. EEGLAB: an open source toolbox for analysis of
amyloid and glucose metabolism in three variants of primary progressive
single-trial EEG dynamics including independent component analysis.
aphasia. Ann. Neurol. 64, 388e401.
J. Neurosci. Methods 134, 9e21.
Ranasinghe, K.G., Hinkley, L.B., Beagle, A.J., Mizuiri, D., Dowling, A.F., Honma, S.M.,
Engel, A.K., Gerloff, C., Hilgetag, C.C., Nolte, G., 2013. Intrinsic coupling modes:
Finucane, M.M., Scherling, C., Miller, B.L., Nagarajan, S.S., Vossel, K.A., 2014.
multiscale interactions in ongoing brain activity. Neuron 80, 867e886.
Regional functional connectivity predicts distinct cognitive impairments in
Engels, M.M., Stam, C.J., van der Flier, W.M., Scheltens, P., de Waal, H., van
Alzheimer's disease spectrum. Neuroimage Clin. 5, 385e395.
Straaten, E.C., 2015. Declining functional connectivity and changing hub loca-
Reid, A.T., Headley, D.B., Mill, R.D., Sanchez-Romero, R., Uddin, L.Q., Marinazzo, D.,
tions in Alzheimer's disease: an EEG study. BMC Neurol. 15, 145.
Lurie, D.J., Valdés-Sosa, P.A., Hanson, S.J., Biswal, B.B., Calhoun, V., Poldrack, R.A.,
Fredericks, C.A., Brown, J.A., Deng, J., Kramer, A., Ossenkoppele, R., Rankin, K.,
Cole, M.W., 2019. Advancing functional connectivity research from association
Kramer, J.H., Miller, B.L., Rabinovici, G.D., Seeley, W.W., 2019. Intrinsic connec-
to causation. Nat. Neurosci. 22, 1751e1760.
tivity networks in posterior cortical atrophy: a role for the pulvinar? Neuro-
Rogelet, P., Delafosse, A., Destee, A., 1996. Posterior cortical atrophy: unusual feature
image Clin. 21, 101628-101628.
of alzheimer's disease. Neurocase 2, 495e501.
Goldstein, E.D., Ertekin-Taner, N., Stephens, A., Carrasquillo, M.M., Boeve, B.,
Rosenbloom, M.H., Alkalay, A., Agarwal, N., Baker, S.L., O'Neil, J.P., Janabi, M., Yen, I.V.,
Tatum, W.O., Feyissa, A.M., 2018. Electroencephalogram findings in patients
Growdon, M., Jang, J., Madison, C., Mormino, E.C., Rosen, H.J., Gorno-
with posterior cortical atrophy. Neurol. Neurochir. Pol. 52, 690e694.
Tempini, M.L., Weiner, M.W., Miller, B.L., Jagust, W.J., Rabinovici, G.D., 2011.
 C.T. Briels et al. / Neurobiology of Aging 96 (2020) 1e11
Distinct clinical and metabolic deficits in PCA and AD are not related to amyloid
distribution. Neurology 76, 1789e1796.
Ross, T.P., Calhoun, E., Cox, T., Wenner, C., Kono, W., Pleasant, M., 2007. The reliability
and validity of qualitative scores for the controlled oral word association test.
Arch. Clin. Neuropsychol. 22, 475e488.
SÁNchez-Cubillo, I., PeriÁÑEz, J.A., Adrover-Roig, D., RodrÍGuez-SÁNchez, J.M., RÍOs-
Lago, M., Tirapu, J., BarcelÓ, F., 2009. Construct validity of the Trail Making Test:
role of task-switching, working memory, inhibition/interference control, and
visuomotor abilities. J. Int. Neuropsychol. Soc. 15, 438e450.
Seymour, R.A., Rippon, G., Kessler, K., 2017. The detection of phase Amplitude
coupling during sensory processing. Front. Neurosci. 11, 487.
Stam, C.J., 2014. Modern network science of neurological disorders. Nat. Rev. Neu-
rosci. 15, 683e695.
Stam, C.J., 2019. ‘’Brainwave’’ software version 0.9.152.12.26. Accessed 7-2019 2019.
http://home.kpn.nl/stam7883/brainwave.html. Accessed 1 July 2020.
Stam, C.J., Nolte, G., Daffertshofer, A., 2007. Phase lag index: assessment of func-
tional connectivity from multi channel EEG and MEG with diminished bias from
common sources. Hum. Brain Mapp. 28, 1178e1193.
Stam, C.J., Tewarie, P., Van Dellen, E., van Straaten, E.C., Hillebrand, A., Van
Mieghem, P., 2014. The trees and the forest: characterization of complex brain
networks with minimum spanning trees. Int. J. Psychophysiol. 92, 129e138.
Tang-Wai, D.F., Graff-Radford, N.R., Boeve, B.F., Dickson, D.W., Parisi, J.E., Crook, R.,
Caselli, R.J., Knopman, D.S., Petersen, R.C., 2004. Clinical, genetic, and neuro-
pathologic characteristics of posterior cortical atrophy. Neurology 63, 1168e1174.
Tewarie, P., Hunt, B.A.E., O'Neill, G.C., Byrne, A., Aquino, K., Bauer, M., Mullinger, K.J.,
Coombes, S., Brookes, M.J., 2019a. Relationships between neuronal oscillatory
amplitude and dynamic functional connectivity. Cereb. Cortex 29, 2668e2681.
Tewarie, P., Liuzzi, L., O'Neill, G.C., Quinn, A.J., Griffa, A., Woolrich, M.W., Stam, C.J.,
Hillebrand, A., Brookes, M.J., 2019b. Tracking dynamic brain networks using
high temporal resolution MEG measures of functional connectivity. Neuroimage
200, 38e50.
11
Tewarie, P., van Dellen, E., Hillebrand, A., Stam, C.J., 2015. The minimum spanning tree:
an unbiased method for brain network analysis. Neuroimage 104, 177e188.
Tijms, B.M., Willemse, E.A.J., Zwan, M.D., Mulder, S.D., Visser, P.J., van Berckel, B.N.M.,
van der Flier, W.M., Scheltens, P., Teunissen, C.E., 2018. Unbiased approach to
counteract upward drift in cerebrospinal fluid amyloid-beta 1-42 analysis re-
sults. Clin. Chem. 64, 576e585.
Tom, T., Cummings, J.L., Pollak, J., 1998. Posterior cortical atrophy: unique features.
Neurocase 4, 15e20.
van der Flier, W.M., Scheltens, P., 2018. Amsterdam dementia cohort: performing
research to optimize care. J. Alzheimers Dis. 62, 1091e1111.
van Diessen, E., Numan, T., van Dellen, E., van der Kooi, A.W., Boersma, M.,
Hofman, D., van Lutterveld, R., van Dijk, B.W., van Straaten, E.C., Hillebrand, A.,
Stam, C.J., 2015. Opportunities and methodological challenges in EEG and MEG
resting state functional brain network research. Clin. Neurophysiol. 126,
1468e1481.
Verhage, F., Van Der Werff, J.J., 1964. [AN analysis OF variance based on the gro-
ninger intelligence test scores]. Ned. Tijdschr. Psychol. 19, 497e509.
Victoroff, J., Ross, G.W., Benson, D.F., Verity, M.A., Vinters, H.V., 1994. Posterior
cortical atrophy: neuropathologic correlations. Arch. Neurol. 51, 269e274.
Whitwell, J.L., Graff-Radford, J., Singh, T.D., Drubach, D.A., Senjem, M.L., Spychalla, A.J.,
Tosakulwong, N., Lowe, V.J., Josephs, K.A., 2017. (18)F-FDG PET in posterior cortical
atrophy and dementia with lewy bodies. J. Nucl. Med. 58, 632e638.
Xia, M., Wang, J., He, Y., 2013. BrainNet Viewer: a network visualization tool for
human brain connectomics. PLoS One 8.
Zhou, J., Gennatas, E.D., Kramer, J.H., Miller, B.L., Seeley, W.W., 2012. Predicting
regional neurodegeneration from the healthy brain functional connectome.
Neuron 73, 1216e1227.
Zwan, M., van Harten, A., Ossenkoppele, R., Bouwman, F., Teunissen, C.,
Adriaanse, S., Lammertsma, A., Scheltens, P., van Berckel, B., van der Flier, W.,
2014. Concordance between cerebrospinal fluid biomarkers and [11C]PIB PET in
a memory clinic cohort. J. Alzheimers Dis. 41, 801e807.