Temporal Dynamic Changes of Intrinsic Brain Activity Associated with Cognitive Reserve in Prodromal Alzheimer’s Disease

Abstract

Background:

Cognitive reserve (CR) is an important protective factor for Alzheimer’s disease (AD), yet its mechanism has not been fully elucidated.

Objective:

To explore the effect of CR on resting and dynamic brain intrinsic activity in patients with mild cognitive impairment (MCI).

Methods:

65 amyloid-β PET-negative (Aβ-) normal controls (NC) and 30 amyloid-β PET-positive (Aβ+) MCI patients underwent resting-state functional magnetic resonance imaging were included from Alzheimer’s Disease Neuroimaging Initiative. According to the years of education, the subjects were divided into high education group and low education group. A two-way analysis of variance was employed for the fractional amplitude of low-frequency fluctuation (fALFF) and dynamic fALFF (dfALFF) comparisons among the four groups. Moreover, the interaction effect of neuroimaging×pathology on clinical cognitive function was tested with linear regression analysis.

Results:

The value of fALFF in the left prefrontal lobe was increased in Aβ+ MCI patients compared to Aβ- NC. The significant interactive effect between disease state and education (binary factor) was observed in the right parahippocampal gyrus (PHG) for fALFF, the right PHG and the right inferior parietal lobule for dfALFF. While no significant results between education (continuous factor) and brain activity was found in voxel-by-voxel analysis. For MCI patients, a significant fluorodeoxyglucose hypometabolic convergence index×right PHG dfALFF interaction was found, indicating the maintenance of executive function at higher levels of dfALFF in the right PHG.

Conclusion:

High CR can alleviate the impairment of hypometabolism on executive function in MCI patients, which is partially achieved by regulating the dynamic brain activity in the right PHG.

Keywords

Cognitive reserve dynamic intrinsic brain activity mild cognitive impairment right parahippocampal gyrus

INTRODUCTION

Alzheimer’s disease (AD) is the most common cause of dementia characterized by progressive and irreversible cognitive dysfunction. The development of biomarker research makes the diagnosis of AD increasingly dependent on pathology [1]. Extracellular amyloid plaques and intracellular neurofibrillary tangles are the pathological hallmarks of AD. However, the degree of cognitive decline in AD patients with the same pathological changes varies greatly. The discontinuity between pathology and clinical performance is defined as cognitive reserve (CR), which is mainly evaluated by epidemiological indicators such as years of education, occupational attainment, or leisure activities in later life [2]. High CR can reduce the risk of developing dementia [3, 4], delay later compromised cognitive health [5], thus reducing social burden. Yet, the neural basis of CR is unclear. Exploring the mechanism of CR can make the intervention more targeted and maximize the protective effect of CR.

Functional magnetic resonance imaging (fMRI) is widely used in neural basis research. Previous studies have shown associations between multiple brain regions with CR during cognitive tasks, such as the posterior cingulate gyrus, parahippocampal gyrus, fusiform gyrus, superior and middle temporal gyrus, hippocampal, frontal lobe, and central anterior gyrus [6]. Due to differences in task paradigms, studies using task-state fMRI to explore the neural basis of CR have inevitably led to mixed results. Research on resting-state functional magnetic resonance imaging (rsfMRI) found that CR was positively correlated with the functional connectivity of the posterior cingulate gyrus [7], left frontal lobe [8, 9], and right middle temporal pole [10]. Higher CR could promote the formation of segregated functional groups [11] and was associated with higher network efficiency in the frontal region [12], right middle temporal pole [10], and parietal and occipital regions [11], whereas, most rsfMRI studies have focused on static functional connectivity which mainly targets the spatial dimension, and has not considered dynamic effects of CR in the temporal dimension.

Besides functional connectivity, the most common intrinsic brain activity (IBA) measures include the amplitude of low-frequency fluctuation (ALFF) and fractional ALFF (fALFF) focused on regional fluctuation properties, and regional homogeneity focused on local synchronization [13]. Compared with regional homogeneity and global brain connectivity, ALFF and fALFF were found to be more sensitive in distinguishing between aMCI and AD patients [14]. ALFF calculates the square root of the power spectrum in low-frequency (0.01–0.08 Hz) fluctuations and directly reflects the intensity of regional spontaneous neural activity [15]. ALFF is susceptible to physiological noise around the ventricle, and fALFF, which measures the ratio of the low-frequency power spectrum to the entire frequency range, has been reported to have higher sensitivity and specificity than ALFF in detecting spontaneous brain activity [15, 16]. Brain activity is dynamic, and short time scales measurements of rsfMRI remain relevant to potential neurological and metabolic activities [17]. However, traditional rsfMRI research assumed that brain activity was temporally stationary throughout the scan, which might underestimate the complex and dynamic interaction patterns of IBA [18]. Many studies have demonstrated that dynamic rsfMRI can better distinguish disease states compared to static rsfMRI [17].

In this study, we explore the neural basis of CR by fALFF and dynamic fractional ALFF (dfALFF), which reflect the intensity and stability of IBA respectively. Furthermore, we evaluate whether the neural basis we found could mitigate the impact of pathology on clinical cognitive function.

MATERIALS AND METHODS

Study population

Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://www.loni.ucla.edu/ADNI). The ADNI was launched in 2003 by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies, and non-profit organizations, as a $60 million, 5-year public–private partnership. The primary goal of ADNI has been to test whether serial MRI, positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early AD. Determination of sensitive and specific markers of very early AD progression is intended to aid researchers and clinicians in developing new treatments and monitor their effectiveness, as well as lessen the time and cost of clinical trials.

All participants were recruited from the Alzheimer’s Disease Neuroimaging Initiative (ADNI2/3). We included subjects (both MCI and normal controls (NC)) who underwent 3T high-resolution structural MRI, resting-state functional MRI (Open eyes), and Amyloid PET by using tracer AV-45. According to the cut-off of 1.11, only amyloid-positive MCI and amyloid-negative NC were included in this study. Exclusion criteria: 1) Progress to AD during scanning; 2) MRI scan parameters are inconsistent with most subjects and cannot be analyzed together; 3) The scanning interval between the structural MRI and rsfMRI exceeds 1 year. Finally, 65 Aβ- NC and 30 Aβ+ MCI patients were included and divided into low cognition reserve group (LCR) and high cognition reserve group (HCR) according to the years of education (more or less than median: 16 years). Ethics approval and informed consent were obtained by the ADNI investigators.

Neuropsychological assessment

The neuropsychological assessments we selected include Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scales to measures global cognitive performance, ADNI composite scores for memory (ADNI_Mem) to assess memory, and ADNI composite scores for executive functioning (ADNI_EF) to assess executive function [19].

MRI acquisition

The structural MRI and rsfMRI were all from the ADNI2/3 and were downloaded on May 27, 2019. The structural MRI were acquired using a rapid gradient-echo sequence with the following parameters: Field Strength = 3 tesla; Flip Angle = 9 degree; Slice Thickness = 1.0 mm; TE = 2.98 ms; TI = 900 ms; TR = 2300 ms; Field of view = 256 mm×240 mm; Voxel Size = 1.1 mm×1.1 mm×1.2 mm, Bandwidth = 240 Hz/pix. Resting-state fMRI images (Eyes Open) were acquired using a echo-planar imaging sequence with the following parameters: Field Strength = 3 tesla; Flip Angle = 80 degree; Slices = 48; Slice Thickness = 3.3 mm; TE = 30 ms; TR = 3000 ms; Voxel Resolution = 3.31 mm×3.31 mm×3.31 mm; Matrix Size = 64×64.

Imaging pre-processing

The preprocessing of rsfMRI was performed using the Data Processing Assistant for Resting-state fMRI (DPARSF) [20], which was based on Statistical Parametric Mapping (SPM8). The first 10 time points were removed to obtain a stable magnetic field and correct the differences in the adaptability of the subjects to the scanning noise. Subsequent preprocessing steps were as follows: slice timing corrected, realigned (cut off < 2 mm or 2°), normalized to the Montreal Neurologic Institute (MNI) space (voxel size was resampled to 3 mm³). Smoothing was performed with a 6 mm full width at half maximum (FWHM) Gaussian kernel. Several covariates (Friston-24 head motion parameters, ventricular signals, white matter signals, and global signals) were regressed out. Next, functional images were detrended and filtered by a 0.01–0.08 Hz band.

Static fALFF and dynamic fALFF calculation

The filtered time series was transformed into the frequency domain based on a fast Fourier transform. And the square root at each frequency of the power spectrum was averaged across 0.01–0.08 Hz at each voxel to obtain the ALFF. The fALFF was calculated by taking the ratio of the low frequency (0.01–0.08 Hz) to the total frequency range. Then, based on the sliding window analysis, we calculated the dynamic fALFF using an open-source MATLAB toolbox DynamicBC [21]. The sliding window length was 13 TR (39 s) and the shifting step size was 1 TR (2 s). To better quantify the dynamic variation of brain activity, we defined the coefficient of variation (CV), which was computed by dividing the standard deviation of dfALFF by the mean.

Hypometabolic convergence index calculation

Each participating site acquired and reconstructed FDG-PET data according to a standardized protocol, and the preprocessing was performed by ADNI PET Coordinating Center investigators. First, the FDG-PET scans in AD patients were compared with those of NCs to obtain the AD hypometabolic map. Similarly, the FDG-PET scans of each subject were compared with those of NCs to obtain the subject’s hypometabolic map. The Hypometabolic Convergence Index (HCI) of each subject was defined as the sum of the product of all voxels with negative z-scores in the subject’s hypometabolic map and AD hypometabolic map. The HCI reflects the extent to which the pattern of cerebral hypometabolism in an individual’s FDG-PET image corresponds to that in AD patients. The greater the HCI value, the more severe the individual’s metabolic damage [22]. The time interval between FDG-PET scan and neuropsychological test/MRI scan is less than one year.

Statistical analysis

Continuous data such as age, years of education, and neuropsychological scores were compared among groups with a two-way analysis of variance (ANOVA). Sex was compared with a Chi-square test. A 2×2 factorial design was employed for fALFF and dfALFF to analyze the interaction effects of education (binary factor)×disease status, correlation analyses (Pearson correlation coefficient) were performed to detect the cerebral region associated with education (continuous factor) in the whole sample, and in MCI and NC subgroups (voxel-wise p < 0.01, cluster-wise p < 0.05, Gaussian Random Field), controlling for age, sex, and gray matter gray matter volume. We verified the interaction effect of education (continuous factor)×disease status (NC versus MCI) by a general linear model, the fALFF and dfALFF of the significant regions were extracted as the dependent variables. A two-sample t-test was used for post-hoc comparisons to compare the fALFF and dfALFF of these regions. After eliminating outliers, a linear regression analysis was used to test the interaction effect of neuroimaging×FDG HCI on neuropsychological scores, controlling for age and sex in Aβ+MCI patients. The neuroimaging variable was defined as the average cluster value of the fALFF and dfALFF in interaction region. The 2×2 factorial analysis for neuroimaging was performed using DPARSF, the linear regression analysis was performed using RStudio, and the rest of the data was analyzed using SPSS 22.0.

RESULTS

Sample characteristics

Demographics and neuropsychological characteristics of the included population were displayed in Table 1. There was no significant difference in age among the four groups. The MMSE scores of MCI patients were lower than NC, and there was no significant difference between high and low education groups in MCI patients. The percentage of males differs among the four groups.

Table 1

Demographics characteristics and neuropsychological assessment

	NC		MCI		F (χ²)	p
	LCR (n = 31)	HCR (n = 34)	LCR (n = 17)	HCR (n = 13)
Age (y)	71.16±8.42	72.86±7.74	77.71±6.61	73.85±8.04	2.62	0.06
Education (y)	14.97±1.35	18.65±0.92	14.53±1.55	18.92±1.12	81.81	<0.001
Gender (M/F)	13/18	16/18	10/7	10/3	9.09	0.01
MMSE	29.13±1.06	29.18±0.94	28.18±1.42	28.54±1.85	3.30	0.02

Data presented as mean±SD. NC, normal controls; MCI, mild cognitive impairment; LCR, low cognitive reserve; HCR, high cognitive reserve; MMSE, Mini-Mental State Examination.

Main effects

There was a significant difference in the fALFF of the left inferior frontal gyrus (peak MNI: –51, 39, 3) between the MCI and NC group, and the brain activity signal of the MCI group was higher than that of the NC group (Fig. 1B). No difference was found between the two groups using dynamic fALFF analysis. No main effects were found for CR.

Fig. 1

The difference of fALFF between MCI and NC group. A) The blue indicates the difference brain region between MCI and NC, the left inferior frontal gyrus. B) The box plot of fALFF in the left inferior frontal gyrus between NC and MCI groups. fALFF, fractional amplitude of low-frequency fluctuation; MCI, mild cognitive impairment; NC, normal controls.

Interaction effects

The fALFF analysis showed that the interaction effect of disease state×education (binary factor) was in the right parahippocampal gyrus (PHG) (peak MNI: 30, –18, -24). Post-hoc analysis showed that in NC, the fALFF of the right PHG in the LCR group was higher than that in the HCR group, and it was the opposite in MCI patients (Fig. 2B).

Fig. 2

The interaction effect of disease state × cognitive reserve on fALFF. A) The red indicates the significant interaction effect in the right parahippocampal gyrus. B) The box plot of fALFF in the right parahippocampal gyrus among four groups. fALFF, fractional amplitude of low-frequency fluctuation; MCI, mild cognitive impairment; NC, normal controls; LCR, low cognitive reserve; HCR, high cognitive reserve.

Dynamic fALFF analysis showed that the interaction effects of disease state×education (binary factor) were in the right inferior parietal lobule (IPL) (peak MNI: 42, 48, 57) and the right PHG (peak MNI: 24, –15, -30). Post-hoc analysis of the right IPL showed significantly decreased CVs of dfALFF in the LCR NC group compared to the HCR NC group, and it was the opposite in MCI patients (Fig. 3B). Post-hoc analysis of the right PHG showed significantly increased CVs of dfALFF in the LCR MCI patients compared to the HCR MCI patients, and no significant difference between LCR NC and HCR NC was observed (Fig. 3D).

Fig. 3

The interaction effect of disease state×cognitive reserve on dynamic fALFF. A) The red indicates the significant interaction effect in the right inferior parietal lobule. B) The box plot of dfALFF in the right inferior parietal lobule among four groups. C) The red indicates the significant interaction effect in the right parahippocampal gyrus. D) The box plot of dfALFF in the right parahippocampal gyrus among four groups. CV, coefficient of variation; dfALFF, dynamic fractional amplitude of low-frequency fluctuation; MCI, mild cognitive impairment; NC, normal controls; LCR, low cognitive reserve; HCR, high cognitive reserve.

Validation analysis showed that education (continuous factors) and disease status (NC versus MCI) had significant interaction effects on fALFF and/or dfALFF in the PHG and IPL (see Supplementary Table 1). However, when using education as a continuous factor in the first-level analysis, we did not get any significant results after GRF correction (voxel-wise p < 0.01, cluster-wise p < 0.05).

Association among neuroimaging, FDG HCI and cognitive function in Aβ+MCI

In Aβ+MCI patients, the linear regression analysis found that only the CVs of dfALFF in the right PHG×FDG HCI had a significant interaction effect on ADNI_EF score (β/SE = 2.400/0.959, t = 2.502, p = 0.022), and the remaining items had no significant interaction effect (see Supplementary Table 2). Visual inspection of Fig. 4 revealed that at low levels of dfALFF in the right PHG, there was a trend of negative correlation between FDG HCI and executive function performance, whereas at higher levels of dfALFF in the right PHG, the same trend between FDG HCI and executive function performance was not observed. The main result was consistent when outlier remained (β/SE = 1.489/0.701, t = 2.124, p = 0.046), and at low levels of dfALFF in the right PHG, the higher FDG HCI was associated with worse executive function performance (p = 0.007), but not at higher levels of dfALFF in the right PHG.

Fig. 4

Scatterplots of the interaction dfALFF in the right parahippocampal gyrus×pathology on executive function. For illustrational purposes, the data were divided into high dfALFF group (H dfALFF) and low dfALFF group (L dfALFF) via median split. Dashed lines indicate 95% confidence intervals. dfALFF, dynamic fractional amplitude of low-frequency fluctuation; ADNI_EF, ADNI composite scores for executive functioning; FDG HCI, fluorodeoxyglucose hypometabolic convergence index.

DISCUSSION

In this study, we found that the interaction effect of education (binary factor)×AD disease state was in the right PHG using the resting state fALFF analysis, and in the right PHG, right IPL using the dynamic fALFF analysis. Further analysis revealed that a higher CVs of dfALFF in the right PHG could alleviate the impairment of low metabolic on executive function in MCI patients. These results suggested that the dfALFF of the right PHG might be a reliable neuroimaging proxy of CR, and the role of education in maintaining executive function was partly achieved by regulating the dynamic brain activity in the right PHG.

Healthy elderly people with high CR had higher neural efficiency. Post-hoc analysis found that the brain activity of the right PHG in the LCR group was significantly higher than that in the HCR group in normal controls. Previous studies have demonstrated that healthy seniors with high CR showed reduced brain activity and connectivity during cognitive tasks, which indicated higher neural efficiency [23, 24]. These were consistent with the results of our study.

The right parahippocampal gyrus in MCI patients with high CR was presumably associated with compensatory recruitment. The neural implementation of CR was often divided into neural reserve and neural compensation, the former was associated with the resilience of pre-existing cognitive networks, while the latter referred to recruit compensatory resources [2]. The parahippocampal gyrus is a cortical region of the medial temporal lobe mainly associated with episodic memory and visuospatial processing [25, 26]. However, we did not find that the intensity or variability of brain activity in the right PHG changed the relationship between pathology and memory (or global cognitive function). Instead, our results indicated that dynamic brain activity in the right PHG could regulate the decline of executive function. Previous studies have suggested that when facing short-term memory tasks, older people not only expressed the same primary network as young people but also recruited additional compensation network, which mainly involves the right parahippocampal cortex [27, 28]. People with higher CR tolerated more pathology when they showed the same clinical cognitive status [2 , 5]. In MCI or AD patients, CR was positively correlated with brain activity during cognitive tasks [24], and individuals with higher CR usually had greater functional connectivity and efficiency [7 , 12]. In summary, the higher brain activity intensity and variability in the right parahippocampal gyrus might be the neural compensation of MCI patients to cope with more severe neuropathological damage.

We found that the brain activity of the left inferior frontal gyrus in the MCI group was significantly higher than that in the normal control group, but we did not find any difference between the two CR groups. Several studies have reported the frontal lobe was related to the CR in healthy elderly, MCI, or AD patients [6]. Franzmeier et al. showed higher left frontal cortex connectivity was associated with a higher education level in aMCI (Aβ+) patients, but not in healthy elderly [8]. However, no significant correlation between the left frontal lobe and the CR was observed in this study. One possible explanation was that the effect of education on the left frontal lobe was already present before the symptoms arise [29]. The subjects included in this study had a relatively high education level, and generally had a higher left frontal lobe activity before the onset, so there might be a ceiling effect.

Another interactive brain region discovered by dynamic fALFF analysis was the right inferior parietal lobule. The inferior parietal lobule located at the junction of the parietal and temporal lobe was a multifunctional region that receives, integrates, and transfers sensory, visual, acoustic information. Wang et al. found that compared with healthy elderly people, AD patients demonstrated distinct disruption of IPL subregional connectivity with various functional networks, but the connectivity between the IPL subregions and the posterior part of the DMN increased, which suggested functional disconnection and compensation of the IPL coexisted in AD patients [30]. This might be one of the reasons why we failed to find that the dynamic brain activity variability of the IPL has a regulatory effect on the relationship between low metabolism and clinical cognition performance in MCI patients. Besides, studies have showed compared with normal controls, the connectivity between the medial prefrontal lobe and the IPL in aMCI patients increased [31], indicating the role of the IPL in aMCI patients seemed to be associated with the prefrontal lobe. We speculated that the right parietal inferior lobule might involve in neural reserve or neural compensation in the early stage, similar to the left frontal lobe, and the exacerbated pathology promoted the neural substrates of CR shifted to the right parahippocampal gyrus. However, there were few studies about the effect of the right IPL in the CR of MCI patients, our hypothesis needed to be treated with caution. More sophisticated experiments to clarify the meaning of IPL for CR would be necessary.

We did not find the effect of continuous education on brain activity. This might be due to the non-linear relationship between education and brain function. Menardi et al. proposed that in healthy and pathological individuals, the slope of CR function had opposite changes, showing mirror-like features, as did brain function [32]. Our results were similar to other studies from ADNI. Nicolas et al. observed no significant results between education (continuous factor) and cerebral metabolism but found the interaction between education (binary factor) and clinical stage in basal forebrain metabolism [33].

Some limitations of this study had to be noted. First, all data come from the ADNI database. Subjects with complete clinical and imaging data mostly had relatively high levels of education. Given the inclusion bias, we might miss some broadly recognized neural basis of the CR due to the ceiling effect. Second, this study only included MCI patients, did not verify whether the dynamic brain activity variability in the right parahippocampal gyrus was still a marker of CR across the entire AD spectrum. The role of the CR relied on brain tissue. As the brain atrophy and network destruction caused by AD progressed, the effect of CR would weaken. In the future, it is necessary to explore the neural substrates of CR along a broad AD spectrum.

Footnotes

ACKNOWLEDGMENTS

Data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.;Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.;Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California.

This research was also supported by NIH grants P30 AG010129 and K01 AG030514.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Knopman

, Haeberlein

, Carrillo

, Hendrix

, Kerchner

, Margolin

, Maruff

, Miller

, Tong

, Tome

, Murray

, Nelson

, Sano

, Mattsson

, Sultzer

, Montine

, Jack

Jr. , Kolb

, Petersen

, Vemuri

, Canniere

, Schneider

, Resnick

, Romano

, van Harten

, Wolk

, Bain

, Siemers

(2018) The National Institute on Aging and the Alzheimer’s Association Research Framework for Alzheimer’s disease: Perspectives from the Research Roundtable. Alzheimers Dement 14, 563–575.

Stern

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

, Yang

, Qi

, Dintica

, Song

, Bennett

, Xu

(2019) Association of lifespan cognitive reserve indicator with dementia risk in the presence of brain pathologies. JAMA Neurol 76, 1184–1191.

Brayne

, Ince

, Keage

, McKeith

, Matthews

, Polvikoski

, Sulkava

(2010) Education, the brain and dementia: Neuroprotection or compensation? Brain 133, 2210–2216.

Robitaille

, van den Hout

, Machado

RJM

, Bennett

, Cukic

, Deary

, Hofer

, Hoogendijk

, Huisman

, Johansson

, Koval

, van der Noordt

, Piccinin

, Rijnhart

JJM

, Singh-Manoux

, Skoog

, Starr

, Vermunt

, Clouston

, Muniz Terrera

(2018) Transitions across cognitive states and death among older adults in relation to education: A multistate survival model using data from six longitudinal studies. Alzheimers Dement 14, 462–472.

Anthony

, Lin

(2018) A systematic review for functional neuroimaging studies of cognitive reserve across the cognitive aging spectrum. Arch Clin Neuropsychol 33, 937–948.

Bozzali

, Dowling

, Serra

, Spano

, Torso

, Marra

, Castelli

, G Dowell

, Koch

, Caltagirone

, Cercignani

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

Franzmeier

, Duering

, Weiner

, Dichgans

, Ewers

(2017) Left frontal cortex connectivity underlies cognitive reserve in prodromal Alzheimer disease. Neurology 88, 1054–1061.

Franzmeier

, Duzel

, Jessen

, Buerger

, Levin

, Duering

, Dichgans

, Haass

, Suarez-Calvet

, Fagan

, Paumier

, Benzinger

, Masters

, Morris

, Perneczky

, Janowitz

, Catak

, Wolfsgruber

, Wagner

, Teipel

, Kilimann

, Ramirez

, Rossor

, Jucker

, Chhatwal

, Spottke

, Boecker

, Brosseron

, Falkai

, Fliessbach

, Heneka

, Laske

, Nestor

, Peters

, Fuentes

, Menne

, Priller

, Spruth

, Franke

, Schneider

, Kofler

, Westerteicher

, Speck

, Wiltfang

, Bartels

, Araque Caballero

, Metzger

, Bittner

, Weiner

, Lee

, Salloway

, Danek

, Goate

, Schofield

, Bateman

, Ewers

(2018) Left frontal hub connectivity delays cognitive impairment in autosomal-dominant and sporadic Alzheimer’s disease. Brain 141, 1186–1200.

10.

Lee

, Lee

, Seo

, Roh

, Oh

, Kim

, Jeong

(2019) Neural substrates of cognitive reserve in Alzheimer’s disease spectrum and normal aging. Neuroimage 186, 690–702.

11.

Marques

, Moreira

, Magalhães

, Costa

, Santos

, Zihl

, Soares

, Sousa

(2016) The functional connectome of cognitive reserve. Hum Brain Mapp 37, 3310–3322.

12.

Weiler

, Casseb

, de Campos

, de Ligo Teixeira

, Carletti-Cassani

AFMK

, Vicentini

, Magalhães

TNC

, de Almeira

, Talib

, Forlenza

, Balthazar

MLF

, Castellano

(2018) Cognitive reserve relates to functional network efficiency in Alzheimer’s disease. Front Aging Neurosci 10, 255–255.

13.

Yan

C-G

, Yang

, Colcombe

, Zuo

X-N

, Milham

(2017) Concordance among indices of intrinsic brain function: Insights from inter-individual variation and temporal dynamics. Sci Bull 62, 1572.

14.

Cha

, Hwang

J-M

, Jo

, Seo

, Na

, Lee

J-M

(2015) Assessment of functional characteristics of amnestic mild cognitive impairment and Alzheimer’s disease using various methods of resting-state fMRI analysis. Biomed Res Int 2015, 907464–907464.

15.

Zou

Q-H

, Zhu

C-Z

, Yang

, Zuo

X-N

, Long

X-Y

, Cao

Q-J

, Wang

Y-F

, Zang

Y-F

(2008) An improved approach to detection of amplitude of low-frequency fluctuation (ALFF) for resting-state fMRI: Fractional ALFF. J Neurosci Methods 172, 137–141.

16.

Zuo

, Di Martino

, Kelly

, Shehzad

, Gee

, Klein

, Castellanos

, Biswal

, Milham

(2010) The oscillating brain: Complex and reliable. Neuroimage 49, 1432–1445.

17.

Thompson

(2018) Neural and metabolic basis of dynamic resting state fMRI. Neuroimage 180, 448–462.

18.

Chen

, Zhang

, Shen

, Lee

S-W

, Shen

(2017) Extraction of dynamic functional connectivity from brain grey matter and white matter for MCI classification. Hum Brain Mapp 38, 5019–5034.

19.

Gibbons

, Carle

, Mackin

, Harvey

, Mukherjee

, Insel

, Curtis

, Mungas

, Crane

, Alzheimer’s Disease Neuroimaging Initiative (2012) A composite score for executive functioning, validated in Alzheimer’s Disease Neuroimaging Initiative (ADNI) participants with baseline mild cognitive impairment. Brain Imaging Behav 6, 517–527.

20.

Chao-Gan

, Yu-Feng

(2010) DPARSF: A MATLAB toolbox for “pipeline” data analysis of resting-state fMRI. Front Systems Neurosci 4, 13.

21.

Liao

, Wu

, Xu

, Ji

, Zhang

, Zang

, Lu

(2014) DynamicBC: A MATLAB toolbox for dynamic brain connectome analysis. Brain Connect 4, 780–790.

22.

Chen

, Ayutyanont

, Langbaum

JBS

, Fleisher

, Reschke

, Lee

, Liu

, Bandy

, Alexander

, Thompson

, Shaw

, Trojanowski

, Jack

Jr. , Landau

, Foster

, Harvey

, Weiner

, Koeppe

, Jagust

, Reiman

, Alzheimer’s Disease Neuroimaging Initiative (2011) Characterizing Alzheimer’s disease using a hypometabolic convergence index. Neuroimage 56, 52–60.

23.

López

, Aurtenetxe

, Pereda

, Cuesta

, Castellanos

, Bruña

, Niso

, Maestú

, Bajo

(2014) Cognitive reserve is associated with the functional organization of the brain in healthy aging: A MEG study. Front Aging Neurosci 6, 125.

24.

Bartrés-Faz

, Arenaza-Urquijo

(2011) Structural and functional imaging correlates of cognitive and brain reserve hypotheses in healthy and pathological aging. Brain Topogr 24, 340.

25.

Aminoff

, Kveraga

, Bar

(2013) The role of the parahippocampal cortex in cognition. Trends Cogn Sci 17, 379–390.

26.

Bohbot

, Allen

JJB

, Dagher

, Dumoulin

, Evans

, Petrides

, Kalina

, Stepankova

, Nadel

(2015) Role of the parahippocampal cortex in memory for the configuration but not the identity of objects: Converging evidence from patients with selective thermal lesions and fMRI. Front Hum Neurosci 9, 431.

27.

Zarahn

, Rakitin

, Abela

, Flynn

, Stern

(2007) Age-related changes in brain activation during a delayed item recognition task. Neurobiol Aging 28, 784–798.

28.

Steffener

, Brickman

, Rakitin

, Gazes

, Stern

(2009) The impact of age-related changes on working memory functional activity. Brain Imaging Behav 3, 142–153.

29.

Barulli

, Stern

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

30.

Wang

, Xia

, Dai

, Liang

, Song

, He

, Li

(2015) Differentially disrupted functional connectivity of the subregions of the inferior parietal lobule in Alzheimer’s disease. Brain Struct Funct 220, 745–762.

31.

Cai

, Chong

, Peng

, Shen

, Li

, von Deneen

, Huang

, Alzheimer’s Disease Neuroimaging Initiative (2017) Altered functional brain networks in amnestic mild cognitive impairment: A resting-state fMRI study. Brain Imaging Behav 11, 619–631.

32.

Menardi

, Pascual-Leone

, Fried

, Santarnecchi

(2018) The role of cognitive reserve in Alzheimer’s disease and aging: A multi-modal imaging review. J Alzheimers Dis 66, 1341–1362.

33.

Nicolas

, Alessandra

, Daniela

, Osman

, Sara

, Giovanni

, Valentina

(2020) Basal forebrain metabolism in Alzheimer’s disease continuum: Relationship with education. Neurobiol Aging 87, 70–77.