Network-Based Substrate of Cognitive Reserve in Alzheimer’s Disease

Abstract

Cognitive reserve (CR) is known to modulate the clinical features of Alzheimer’s disease (AD). This concept may be critical for the development of non-pharmacological interventions able to slow down patients’ cognitive decline in the absence of disease-modifying treatments. We aimed at identifying the neurobiological substrates of CR (i.e., neural reserve) over the transition between normal aging and AD, by assessing the underlying brain networks and their topological properties. A cohort of 154 participants (n = 68 with AD, n = 61 with amnestic mild cognitive impairment (aMCI), and 25 healthy subjects) underwent resting-state functional MRI and neuropsychological testing. Within each group, participants were classified as having high or low CR, and functional connectivity measures were compared, within group, between high and low CR individuals. Network-based statistics and topological network properties derived from graph theory were explored. Connectivity differences between high and low CR were evident only for aMCI patients, with participants with high CR showing a significant increase of connectivity in a network involving mainly fronto-parietal nodes. Conversely, they showed significantly decreased connectivity in a network involving fronto-temporo-cerebellar nodes. Consistently, changes to topological measures were observed in either direction, and were associated with measures of global cognitive function. These findings support the hypothesis that CR impacts on neurodegenerative process in the early phase of AD only. In addition, they fit with the existence of a “neural reserve”, characterized by specific neural networks and their efficiency. It remains to be demonstrated whether interventions later in life can modulate this “neural reserve”.

Keywords

Alzheimer’s disease amnestic mild cognitive impairment brain connectivity cognitive reserve compensatory network neural reserve

INTRODUCTION

Alzheimer’s disease (AD) is the most common form of cognitive decline in the elderly [1]. It was observed that, in the sporadic form, AD pathology impacts differently according to individual lifestyles [2]. For this reason, the concept of cognitive reserve (CR) was introduced [2], and its role as disease modulator was extensively demonstrated in neurodegeneration [3 –6], opening encouraging perspectives in terms of non-pharmacological interventions [7].

The CR has been quantified by several measures [2], including various aspects of the individual lifestyle [2]. Among all possible measures of CR, education is considered as one of the most relevant factors impacting on brain resilience [2]. Education is relatively easy to determine and due to its early occurrence in life, it is less affected by comorbidities and habits of adulthood. Additionally, education can be considered as a socializing process, promoting learning strategies, encouraging the development of divergent thinking, and enabling the individual to perform more competently on cognitive demand [8]. Education appears to be an important environmental experience that may enhance neural connectivity as well as the propensity to engage in mentally stimulating activities throughout life [9]. Finally, education may improve the intellectual approach to life events [10], which can lead to lifelong mental stimulation and an enhanced activation of the brain regions involved in cognitive processing [10].

A neurobiological substrate of CR has been previously demonstrated in AD in terms of metabolic [11 –13] or structural [5] brain variability, indicating that individuals with higher as compared to those with lower CR are more resilient to brain damage. This resilience might be potentially modulated by tailored treatments in patients with cognitive decline at pre-dementia stage.

The cognitive reserve hypothesis [2] has been recently improved by introducing the new concept of “neural reserve” [14], which reduces the distance between cognitive processes of reserve and their neurobiological correlates. According to this view, CR operates on brain networks’ efficiency rather than modifying structure or function in individual brain areas [14]. Brain networks, which are the closest representations we can obtain of neural reserve in vivo, can be assessed using connectomics [15, 16], a novel approach with the ability to assess the topological properties of brain networks.

The specific aims of study were, using approaches based on connectomics, to: 1) identify the neurobiological substrates of neural reserve by assessing the underlying networks and topological properties; 2) assess the differential impact of CR over the transition between normal aging and AD passing through the pre-dementia stage of mild cognitive impairment (MCI).

METHODS

Participants

A cohort of 154 participants, 68 with a diagnosis of probable AD (M/F = 25/43; mean age = 71.6, SD = 6.7 years), 61 with a diagnosis of amnestic MCI (aMCI) (M/F = 30/31; mean age = 70.4, SD = 9.0 years), and 25 healthy elderly subjects (HS) (M/F = 8/17; mean age = 68.5, SD = 6.7 years) were enrolled. The three groups were not significantly different for age or gender distribution.

The diagnosis of probable AD was according to the clinical criteria of the National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) [17, 18]. The diagnosis of aMCI was performed according to current criteria [19]. Patients could be either single- (n = 29) or multiple- (n = 32) domain MCI, and had not to respond to the diagnostic criteria for major cognitive disorder [20], showing a Clinical Dementia Rating [21] score not exceeding 0.5. As detailed below, medial temporal lobe atrophy was assessed in all subjects to confirm that they had an intermediate likelihood of underlying AD neuropathology, and to control for patients homogeneity across high and low CR subgroups. Cognitively normal subjects showing the presence of significant medial temporal lobe atrophy were excluded. All recruited subjects with a Hachinski score [22] higher than 4 were excluded. Major systemic, psychiatric, and other neurological illnesses were also carefully investigated and excluded in all participants. Finally, subjects had to be right-handed, as assessed by the Edinburgh Handedness Inventory [23].

The study was approved by the Ethical Committee of Santa Lucia Foundation, and written informed consent was obtained from all participants before study initiation. All procedures performed in this study were in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Neuropsychological assessment

All participants underwent an extensive neuropsychological battery (See Supplementary Material). For the purposes of the current study, focused on the CR, neuropsychological scores were not adjusted for age and education, as previously reported [5]. Seventeen one-way ANOVAs were used to assess between group differences in neuropsychological performances. To avoid the type-I error, Bonferroni’s correction was applied (p value threshold α= 0.05/17 = 0.003).

Classification criteria to define the level of cognitive reserve

As previously reported [5], we divided participants on the basis of their level of formal education. Within each group, the years of formal education were transformed in z scores, and individuals reporting a z score ≤0 were considered having low CR (AD-L_CR; n = 37; aMCI-L_CR = 33, and HS-L_CR = 13). Conversely, individuals with a z score >0 were considered having high CR (AD-H_CR; n = 31; aMCI-L_CR = 28, and HS-L_CR = 12). Individuals with high and low CR were equally distributed across groups. Table 1 summarizes the principal characteristics of all subjects.

MRI acquisition

MRI included a volumetric scan (TR = 1338 ms, TE = 2.4 ms, Matrix = 256×224, n. slices = 176, thickness = 1 mm), and resting state fMRI (RS-fMRI) (TR = 2080 ms, TE = 30 ms, 32 axial slices, matrix = 64×64, pixel size = 3×3 mm², slice thickness = 2.5 mm, 220 volumes).

Medial temporal lobe atrophy

The Medial Temporal lobe Atrophy scale (MTA) [24] was employed on volumetric images to assess the severity of atrophy in each subject. This scale provides a rating score from 0 to 4, with scores ≥1.5 [25] indicating significant atrophy. For each subject we averaged the scores obtained in the right and left hemispheres to obtain a single measure of medial-temporal lobe atrophy. One-way ANOVA was employed to control for within group differences (AD-L_CR versus AD-H_CR; aMCI-L_CR versus aMCI-H_CR; and HS-L_CR versus HS-H_CR, respectively).

Image analysis for RS-fMRI

Images were pre-processed for resting-state fMRI using SPM8 (http://www.fil.ion.ucl.ac.uk/spm/), and in-house Matlab scripts as previously described [4].

Construction of connectivity matrices

For each participant, the whole brain was parcellated into 116 regions of interests (ROIs) based on the automated anatomical labeling (AAL) atlas. Each ROI corresponds to a node of the network, and its mean time course was calculated as the average of the fMRI time series from all voxels within the region. Correlation matrices were then obtained calculating the correlation between all pairs of ROI mean signals. The matrices were analyzed using two complementary approaches: 1) Network-based statistics (NBS) [26], which compares the strength of connectivity for each pair of nodes in the network between groups; 2) Graph Theory [15, 16], which characterizes the network using special indices describing its shape and properties, including efficiency and resilience to attack.

Network-based analysis

To detect between-group differences in the inter-nodal functional connections, two-sample t-tests were performed in NBS (10000 permutation-tests and a p-value = 0.005 were set). To separately investigate the effects of CR at each stage of the disease, and to avoid the potential bias due to the different severity of grey matter atrophy between patients with AD, with aMCI and HS, we compared subjects with low and high CR within their own diagnostic group (AD-L_CR versus AD-H_CR; aMCI-L_CR versus aMCI-H_CR; and HS-L_CR versus HS-H_CR, respectively).

Graph-Theory analysis

We used Brain Connectivity Toolbox [16] and Matlab custom scripts to explore global and local topological properties of each participant’s brain. Undirected binary connectivity matrices were built as previously described [27], obtaining several connectivity matrices for each subject in a specific density range, where the density characterizes the number of connections compared to the maximum possible number of connections for that graph [28]. Details of graph theory and its application to brain networks can be found elsewhere [15, 16], together with a full description of all the indices that can be derived from it. For the purposes of the current investigation, we focused on the topological measures with a more direct clinical interpretation [29, 30]. As global metrics, mean clustering coefficient (a measure of segregation), characteristic path length (a measure of integration), assortativity (an index of how the high-degree nodes are linked together), modularity (measuring the tendency of the network to form distinct sub-networks), and small-worldness were calculated [16]. With respect to local metrics, we considered betweenness centrality, nodal degree, and nodal efficiency. Betweenness centrality is defined as the fraction of all the shortest paths passing through a given node; nodal degree expresses the number of connections for each node, nodal efficiency is inversely related to each node’s paths length, and identifies the less efficient nodes along certainroutes.

For each global and local measure, and at each density value, two-sample t-tests were performed in order to evaluate significant differences between subjects with low and high CR within their own diagnostic group (AD-L_CR versus AD-H_CR; aMCI-L_CR versus aMCI-H_CR; and HS-L_CR versus HS-H_CR, respectively). In addition, for every considered global measure we extracted, subject-by-subject, the values associated with each different density and, then, we averaged these values in order to obtain an individual mean value (Imv-). These values’ distributions were checked for normality using the Shapiro-Wilk test (W) and then were used to assess potential associations between global connectivity properties and clinical-cognitive characteristics. Therefore, in patients only (AD-L_CR, AD-H_CR, aMCI-L_CR, and aMCI-H_CR), the Imv-indices were correlated with the Mini-Mental State Examination (MMSE)[31, 32].

RESULTS

Demographic and clinical characteristics

There were no significant differences within and between groups in age (F_5,148 = 0.74, p = N.S.). As expected, general cognitive efficiency was significantly different across diagnostic groups (AD, aMCI, and HS) (F_5,148 = 50.3, p < 0.001), with AD patients reporting worse MMSE scores than aMCI patients and HS. When dividing patients according to the level of CR within groups, the subsamples did not show any significant difference in MMSE scores. All these data are summarized in Table 1.

Neuropsychological assessment

Table 2 shows between-group differences in each administered tests. As expected, AD and aMCI patients (irrespective of their CR group belonging) showed significantly worse performances than HS in tests for both verbal and visuo-spatial episodic memory. When dividing each group by CR level, AD patients with both high and low CR, but only low CR aMCI patients performed worse than HS in test assessing verbal and short-term memory. AD-H_CR performed worse than aMCI-H_CR in tests assessing constructional praxis and logical reasoning. In addition, AD patients compared to healthy controls and to aMCI-H_CR showed worse performances in test evaluating executive functions.

Interestingly, aMCI-L_CR obtained lower scores than aMCI-H_CR in the phonological verbal fluency test.

Medial temporal lobe atrophy

None of the HS reported MTA scores higher than 1. All patients (MCI and AD) reported MTA scores higher than 1, indicating intermediate likelihood of having AD pathology [18, 19]. Table 1 summarizes the MTA scores in all considered groups. The ANOVA revealed the presence of significant MTA in AD and MCI patients compared to HS (F_5,148 = 34.2, p < 0.001), while no significant within group differences were found when dividing subjects into high or low CR.

Network-based analysis

A-MCI-H_CR showed both significant increases and decreases of connectivity compared to aMCI-L_CR. In particular, aMCI-H_CR showed a significant increase of connectivity in a network involving mainly fronto-parietal nodes (see Fig. 1A, left side, and Supplementary Table 1A). Conversely, they showed significantly decreased connectivity in a network involving fronto-temporo-cerebellar nodes (see Fig. 1A, right side, Supplementary Table 1B). In healthy participants, the networks of increased and decreased connectivity in those with high compared to those with low CR were largely overlapping (see Fig. 1B and Supplementary Table 1C, D) indicating no real difference. Finally, no differences in network connectivity were detected between AD patients with high and low CR.

Global and local brain functional connectivity measures (Graph Theory)

The only between-CR levels difference in global measures was found in Clustering coefficient between AD-L_CR and AD-H_CR. As most of the variables considered showed a non-normal distribution, non-parametrical statistical tests were used to assess correlation with the general cognitive efficiency. We found significant positive correlations only in patients with aMCI-H_CR between MMSE score and Imv-small-worldness (R = 0.47, p = 0.01) and between MMSE and Imv-clustering coefficient (R = 0.43, p = 0.02).

Conversely, we found significant differences between CR groups in the connectivity of local measures. These results are graphically summarized in Fig. 2. AD-H_CR compared to AD-L_CR patients showed significantly increased connectivity in the frontal, temporal, and parietal lobes, and in the cerebellum; however, AD-H_CR showed also reduced connectivity in the left superior frontal and in the fusiform gyri (see detailed statistics in Supplementary Table 2).

A-MCI-H_CR patients compared to aMCI-L_CR showed both increased and decreased connectivity in several nodes. The increases were mainly observed in the frontal and in the parietal lobes; whereas reduced connectivity involved the betweenness-centrality of the right posterior cingulate cortex and of the cerebellum and also the nodal efficiency and degree of the right cuneus, of the left insula, of the left pallidum and of the occipital gyrus, and of the thalamus bilaterally (see detailed statistics in Supplementary Table 3).

Lastly, HS-H_CR subjects compared to HS-L_CR showed increased connectivity in the right inferior frontal gyrus, in the bilateral middle frontal gyrus, in the left anterior cingulate cortex, in the left amygdala, and in the cerebellum; they showed mainly a reduced nodal efficiency in the right hippocampus and bilateral parahippocampal gyrus (see Supplementary Table 4).

DISCUSSION

The aim of the present study was to assess the impact of different levels of CR on brain functional connectivity of older individuals spanning the AD continuum and including healthy subjects. As previously reported [4, 5], the levels of CR were assessed using the years of formal education. Indeed, education has been considered as an environmental factor potentially enhancing neural connectivity [9], and it is often used as a proxy of CR [2]. This is justified by the long-term consequences education usually has on lifestyle [9]. The main novelty of this study lies in the use of graph theory-based approaches to investigate the neural mechanism underlying the CR, with the specific aim of providing evidence for neural reserve, the biological substrate of the CR [2 –14].

As previously argued by Stern and co-workers [2 –14], CR mechanisms can increase resilience to the pathological insult only in the early stages of the disease, before reaching a inflection point, after which the slope of decline becomes very steep. Consistent with this classical view, at neuropsychological level, we found significant difference between high and low CR participants only for aMCI. Specifically, patients with aMCI and low CR compared to those with high CR showed worse performances in the phonological verbal fluency test. This cognitive task, that requires the selection of an appropriate item by using a strategic retrieval constrained by phonology [33], relies on the integrity of several brain structures including the inferior and middle frontal gyri and the postcentral gyrus [34].

Based on previous suggestions that functional connectivity might reflect the recruitment of alternative brain areas to preserve function in the presence of structural damage [4], we assessed both global and local measures of functional connectivity within the framework of graph theory [15, 16]. At global level, we did not find any difference between aMCI-H_CR and aMCI-L_CR. We speculate that the lack of significant difference in the global segregation/integration measures [35] is reflected by the absence of impairment in the general cognitive efficiency observed in the whole cohort of aMCI patients. However, we found a significant positive correlation between some global measures (Small-worldness and Normalized clustering coefficient, which measure both integration and segregation) and the level of general cognitive efficiency in aMCI with high CR. Comprehensively, these findings demonstrated that the optimal balance between segregation and integration in brain networks is linked to cognitive efficiency, supporting the hypothesis that higher order cognitive functions depend on a more integrative network topology [36].

Local topological properties, on the other hand, indicate the ability of nodes to act as “hubs”, and thus to facilitate functional integration. Damage to a hub can have a disproportionate impact on the network’sglobal efficiency of information processing [37, 38]. In terms of local measures, we found that patients with aMCI and low CR showed decreased connectivity compared to those with high CR in several regions including the inferior frontal gyrus, the anterior cingulate cortex bilaterally, and the left postcentral gyrus. Changes to the connectivity among these regions, which are those typically involved in the phonological verbal fluency task [33, 34], provide a neural basis for their poor performance at this task.

Beside the anatomical localization, it is interesting to speculate also about the nature of the changes we observe. In particular, nodes with reduced connectivity in aMCI-L_CR showed mainly reduced nodal degree, indicating loss of connections compared aMCI-H_CR. By contrast, nodes having lower connectivity in aMCI-H_CR, encompassing the cerebellum, the posterior cingulate cortex, the thalamus and the occipital regions, were mainly characterized by loss of nodal efficiency. Within the classic CR hypothesis, we could therefore speculate that the accrual of pathological damage causes this loss of efficiency in aMCI-H_CR, which is however compensated by the increased number of alternative connections, compared to aMCI-L_CR.

This hypothesis is also corroborated by an independent analysis performed using NBS. This complementary analysis aims at isolating sub-networks whose connectivity differs between high and low CR. We found evidence of the presence of networks with both, increased (fronto-parietal network) and decreased (fronto-temporo-cerebellar network) connectivity in aMCI-H_CR compared to aMCI-L_CR. The fronto-temporo-cerebellar network involved mainly the anterior prefrontal areas (BA10), the temporal pole and the cerebellar vermis and lobule 9, which are respectively involved in several high cognitive functions such as executive control [39], and in cognitive and emotional processing [40]. Conversely, the fronto-parietal network involved mainly the orbital and ventromedial-prefrontal (BA11/47) regions, the anterior cingulate, and the precuneus. Also these brain regions are involved in higher cognitive processes such as inhibition of behavior [41], reversal learning [41], planning [41], and decision making [41]. We hypothesized that high CR allowed aMCI patients to cope better with neurodegenerative process, strengthening a compensatory network in the superior portion of the brain (the fronto-parietal regions) that included areas specialized in similar cognitive tasks tapping from the disrupted inferior fronto-temporo-cerebellar network. We propose that these opposing networks can be considered as the biological evidence for the neural substrate of CR, namely neural reserve.

Moreover, as previously reported [6], here we reinforce the hypothesis that CR impacts on neurodegenerative process in a confined time-window that is in the early phase of AD only. Indeed, we did not find any robust indication of functional connectivity difference between high and low CR in AD or HS participants. This finding supports the hypothesis that CR can only compensate for pathology up to a certain stage, and therefore does not act in the advanced stage of AD. Interestingly, some differences at local level were found for HS, for whom reduced connectivity in high CR individuals was mainly measured by loss of efficiency (in the medial temporal lobe), while increased connectivity was primarily characterized by increased nodal degree (in the frontal cortex and cerebellum). Further studies are warranted to better understand the clinical relevance of these changes.

In conclusion, our data support the existence of a “neural reserve”, which is stimulated by lifestyle, and is characterized by specific neural networks. We speculate that cognitive functioning and its efficiency in the presence of AD pathology might reflect the efficiency of these networks. It remains to be demonstrated whether interventions later in life can modulate this neural reserve.

Footnotes

ACKNOWLEDGMENTS

This project has been supported by grants from the Italian Ministry of Health to Dr. Marco Bozzali (150/RF-2009-1491699; 047/RF-2010-2311484).

Authors’ disclosures available online ().

Appendix

References

Wimo

, Jönsson

, Bond

, Prince

, Winblad

, Alzheimer

Disease International

(2013) The worldwide economic impact of dementia 2010. Alzheimers Dement 9, 1–11.e3.

Stern

(2012) Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 11, 1006–1012.

Bozzali

, Padovani

, Caltagirone

, Borroni

(2011) Regional grey matter loss and brain disconnection across Alzheimer disease evolution. Curr Med Chem 18, 2452–2458.

Bozzali

, Dowling

, Serra

, Spanò

, Torso

, Marra

, Castelli

, Dowell

, Koch

, Caltagirone

, Cercignani

(2015) The impact of cognitive reserve on brain functional connectivity in Alzheimer’s disease. J Alzheimers Dis 44, 243–250.

Serra

, Cercignani

, Petrosini

, Basile

, Perri

, Fadda

, Spanò

, Marra

, Giubilei

, Carlesimo

, Caltagirone

, Bozzali

(2011) Neuroanatomical correlates of cognitive reserve in Alzheimer disease. Rejuvenation Res 14, 143–151.

Serra

, Musicco

, Cercignani

, Torso

, Spanò

, Mastropasqua

, Giulietti

, Marra

, Bruno

, Koch

, Caltagirone

, Bozzali

(2015) Cognitive reserve and the risk for Alzheimer’s disease: A longitudinal study. Neurobiol Aging 36, 592–600.

Liberati

, Raffone

, Olivetti Belardinelli

(2012) Cognitive reserve and its implications for rehabilitation and Alzheimer’s disease. Cogn Process 13, 1–12.

Gilleard

(1997) Education and Alzheimer’s disease: A review of recent international epidemiological studies. Aging Ment Health 1, 33–46.

Vance

, Crowe

(2006) A proposed model of neuroplasticity and cognitive reserve in older adults. Activ Adapt Aging 30, 61–79.

10.

Friedland

(1993) Epidemiology, education, and the ecology of Alzheimer’s disease. Neurology 43, 246–249.

11.

Ewers

, Insel

, Stern

, Weiner

, Alzheimer’s Disease Neuroimaging Initiative (ADNI) (2013) Cognitive reserve associated with FDG-PET in preclinical Alzheimer disease. Neurology 80, 1194–1201.

12.

Garibotto

, Tettamanti

, Marcone

, Florea

, Panzacchi

, Moresco

, Virta

, Rinne

, Cappa

, Perani

(2013) Cholinergic activity correlates with reserve proxies in Alzheimer’s disease. Neurobiol Aging 34, 2694.e13–8.

13.

Morbelli

, Perneczky

, Drzezga

, Frisoni

, Caroli

, van Berckel

, Ossenkoppele

, Guedj

, Didic

, Brugnolo

, Naseri

, Sambuceti

, Pagani

, Nobili

(2013) Metabolic networks underlying cognitive reserve in prodromal Alzheimer disease: A European Alzheimer disease consortium project. J Nucl Med 54, 894–902.

14.

Barulli

, Stern

(2013) Efficiency, capacity, compensation, maintenance, plasticity: Emerging concepts in cognitive reserve. Trends Cogn Sci 17, 502–509.

15.

Bullmore

, Sporns

(2009) Complex brain networks: Graph theoretical analysis of structural and functional systems. Nat Rev Neurosci 10, 186–198.

16.

Rubinov

, Sporns

(2010) Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069.

17.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–944.

18.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

19.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

20.

American Psychiatric Association (2013) Diagnostic and Statistical Manual of Mental Disorders, 5th edition. American Psychiatric Association, Washington, DC.

21.

Hughes

, Berg

, Danziger

, Coben

, Martin

(1982) A new clinical scale for the staging of dementia. Br J Psychiatry 140, 566–572.

22.

Hachinski

, Iliff

, Zilhka

, Du Boulay

, McAllister

, Marshall

, Russell

, Symon

(1975) Cerebral blood flow in dementia. Arch Neurol 3, 632–637.

23.

Büsch

, Hagemann

, Bender

(2010) The dimensionality of the Edinburgh Handedness Inventory: An analysis with models of the item response theory. Laterality 15, 610–628.

24.

Scheltens

, Launer

, Barkhof

, Weinstein

, van Gool

(1995) Visual assessment of medial temporal lobe atrophy on magnetic resonance imaging: Interobserver reliability. J Neurol 242, 557–560.

25.

Pereira

, Cavallin

, Spulber

, Aguilar

, Mecocci

, Vellas

, Tsolaki

, Kłoszewska

, Soininen

, Spenger

, Aarsland

, Lovestone

, Simmons

, Wahlund

, Westman

, AddNeuroMed consortium and for the Alzheimer’s Disease Neuroimaging Initiative (2014) Influence of age, disease onset and ApoE4 on visual medial temporal lobe atrophy cut-offs. J Intern Med 275, 317–330.

26.

Zalesky

, Fornito

, Bullmore

(2010) Network-based statistic: Identifying differences in brain networks. Neuroimage 53, 1197–1207.

27.

van Wijk

, Stam

, Daffertshofer

(2010) Comparing brain networks of different size and connectivity density using graph theory. PLoS One 5, e13701.

28.

Liu

, Zhang

, Yan

, Bai

, Dai

, Wei

, Zhong

, Xue

, Wang

, Feng

, You

, Zhang

, Tian

(2012) Altered topological patterns of brain networks in mild cognitive impairment and Alzheimer’s disease: A resting-state fMRI study. Psychiatry Res 202, 118–125.

29.

Yao

, Zhang

, Lin

, Zhou

, Xu

, Jiang

; Alzheimer’s Disease Neuroimaging Initiative (2010) Abnormal cortical networks in mild cognitive impairment and Alzheimer’s disease. PLoS Comput Biol 6, e1001006.

30.

Shim

, Kim

, Lee

, Im

(2014) Disruptions in small-world cortical functional connectivity network during an auditory oddball paradigm task in patients with schizophrenia. Schizophr Res 156, 197–203.

31.

Folstein

, Folstein

, McHugh

(1975) Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

32.

Measso

, Cavarlezan

, Zappalà

, Lebowitz

, Crook

, Pirozzolo

, Grigoletto

, Amaducci

, Massari

, Lebowitz

(1993) The Mini Mental State Examination: Normative study of a random sample of Italian population. Dev Neuropsychol 9, 77–95.

33.

Baldo

, Schwartz

, Wilkins

, Dronkers

(2006) Role of frontal versus temporal cortex in verbal fluency as revealed by voxel-based lesion symptom mapping. J Int Neuropsychol Soc 12, 896–900.

34.

Biesbroek

, van Zandvoort

, Kappelle

, Velthuis

, Biessels

, Postma

(2016) Shared and distinct anatomical correlates of semantic and phonemic fluency revealed by lesion-symptom mapping in patients with ischemic stroke. Brain Struct Funct 221, 2123–2134.

35.

Dai

, He

(2014) Disrupted structural and functional brain connectomes in mild cognitive impairment and Alzheimer’s disease. Neurosci Bull 30, 217–232.

36.

Dehaene

, Changeux

(2011) Experimental and theoretical approaches to conscious processing. Neuron 70, 200–227.

37.

Albert

, Barabasi

(2000) Topology of evolving networks: Local events and universality. Phys Rev Lett 85, 5234–5237.

38.

Crossley

, Mechelli

, Scott

, Carletti

, Fox

, McGuire

, Bullmore

(2014) The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain 137, 2382–2395.

39.

Mansouri

, Tanaka

, Buckley

(2009) Conflict-induced behavioural adjustment: A clue to the executive functions of the prefrontal cortex. Nat Rev Neurosci 10, 141–152.

40.

Schmahmann

(2010) The role of the cerebellum in cognition and emotion: Personal reflections since 1982 on the dysmetria of thought hypothesis, and its historical evolution from theory to therapy. Neuropsychol Rev 20, 236–260.

41.

Zald

, Andreotti

(2010) Neuropsychological assessment of the orbital and ventromedial prefrontal cortex. Neuropsychologia 48, 3377–3391.

42.

Xia

, Wang

, He

(2013) BrainNet Viewer: A network visualization tool for human brain connectomics. PLoS One 8, e68910.