Selective Vulnerability Related to Aging in Large-Scale Resting Brain Networks

Abstract

Normal aging is associated with cognitive decline. Evidence indicates that large-scale brain networks are affected by aging; however, it has not been established whether aging has equivalent effects on specific large-scale networks. In the present study, 40 healthy subjects including 22 older (aged 60–80 years) and 18 younger (aged 22–33 years) adults underwent resting-state functional MRI scanning. Four canonical resting-state networks, including the default mode network (DMN), executive control network (ECN), dorsal attention network (DAN) and salience network, were extracted, and the functional connectivities in these canonical networks were compared between the younger and older groups. We found distinct, disruptive alterations present in the large-scale aging-related resting brain networks: the ECN was affected the most, followed by the DAN. However, the DMN and salience networks showed limited functional connectivity disruption. The visual network served as a control and was similarly preserved in both groups. Our findings suggest that the aged brain is characterized by selective vulnerability in large-scale brain networks. These results could help improve our understanding of the mechanism of degeneration in the aging brain. Additional work is warranted to determine whether selective alterations in the intrinsic networks are related to impairments in behavioral performance.

About the Authors:

Hong-Ying Zhang

Affiliation: Department of Radiology, Subei People's Hospital of Jiangsu Province, Yangzhou University, Yangzhou, China

Wen-Xin Chen

Affiliation: Department of Radiology, Subei People's Hospital of Jiangsu Province, Yangzhou University, Yangzhou, China

Yun Jiao

Affiliation: Jiangsu Key Laboratory of Molecular and Functional Imaging, Southeast University, Nanjing, China

Yao Xu

Affiliation: Department of Neurology, Subei People's Hospital of Jiangsu Province, Yangzhou University, Yangzhou, China

Xiang-Rong Zhang

Affiliation: Department of Psychiatry, Zhongda Hospital, Southeast University, Nanjing, China

Jing-Tao Wu

* E-mail: [email protected]

Affiliation: Department of Radiology, Subei People's Hospital of Jiangsu Province, Yangzhou University, Yangzhou, China

Introduction

The decline of functions such as memory, attention, problem-solving and sensorimotor ability is commonly experienced in old age. Such age-related cognitive decline may involve the selective deterioration of specific brain systems. Many hypotheses have been proposed to describe the aging process of the brain [1], [2]. Studying the aged brain with respect to the networks that sustain brain functions may be helpful in understanding the complicated, age-related changes that occur in this organ.

Measuring synchronic low-frequency fluctuations (LFFs) among spatially distant brain regions using blood oxygen level-dependent (BOLD) functional magnetic resonance imaging (fMRI) signals is useful for mapping intrinsic neural networks in healthy and diseased subjects. Researchers have identified resting state neural network (RSN) architectures that are presumed to underlie the higher-order cognitive functions that are associated with memory, attention, execution and emotion processing [3]–[6] in addition to motor and sensory functions [7], [8]. The corresponding higher-order networks include the following: 1) a default mode network (DMN), 2) a dorsal attention network (DAN), 3) an executive control network (ECN) and 4) a salience network (SN). The DMN is involved in internal and external environment alerts as well as memory- and self-related functions, and it presents temporospatial anti-correlations with other canonical networks during externalized cognitive tasks [5], [9]–[11]. These canonical networks have been systemically measured in patients with neuropsychiatric disorders and have shown selective vulnerability in distinct neurodegenerative illnesses, such as Alzheimer's disease (AD) and frontotemporal dementia syndromes [12]–[15]. For instance, the DMN selectively disrupts the connectivity in patients with AD but enhances the connectivity in patients with behavioral variant frontotemporal dementia, whereas the SN shows enhanced activities in patients with AD and disrupted connectivity in patients with behavioral variant frontotemporal dementia [13]. These canonical networks are also differentially affected in schizophrenic patients compared to controls; schizophrenic patients demonstrate increased connectivity in the DMN, less connectivity in the ECN and DAN, and no difference in the salience network [16]. These findings indicate that the selective alterations in large-scale resting brain networks could represent a characteristic of neuropsychiatric disorders.

Aging, which could broadly affect multiple brain systems, is considered to be the primary risk factor for neurodegenerative disorders, such as sporadic AD. Numerous conventional task-based and resting-state fMRI investigations have been conducted to understand the effects of aging on brain function. In task-based studies, abnormally decreased or increased brain activation was found in normal elderly subjects during various cognitive processes, such as memory, attention and semantic judgments [17]–[19]. The alterations in the balance between DMN and task-related activity could account for the increased vulnerability to distraction by irrelevant information among the elderly subjects [17]. Additional investigations revealed that aging is associated with changes in LFF correlations between different cortical regions during task performance and in the resting state [20]–[25]. Stronger resting-state network connectivity in older adults is associated with better performance in tests of executive function and processing speed, such as the Trail Making Test, the word frequency task and face-name associative memory task [20], [24], [25]. Accumulating evidence from neuroimaging studies has suggested that the decline in cognitive functions is accompanied by focal changes in activity in several brain regions, such as the prefrontal cortex [26]–[28], hippocampal formation [28]–[30], and parietal regions [19], [27]. The evidence from graph theoretical analyses has indicated that older adults have decreased topological efficiency in small-world and modular organization of the entire brain [22], [31], [32]. In the present study, we investigated how the four canonical intrinsic networks underpinning cognitive functions change with aging.

Many studies have strongly indicated that resting-state functional connectivity provides relevant information regarding the effects of aging on brain functioning and cognition (for a review, see [33]). The most consistent age-related change is decreased resting-state connectivity in the DMN, which is crucial for memory [20]–[23], [32], [34]; another convergent finding is that the DAN is affected by the aging process [28], [32], [35]. Aging has a particular effect on the hub-related regions in the brain. The patterns of low-level anti-correlations between the DMN sub-networks and task-positive networks have also been demonstrated [23]. The findings noted from the topological analyses consistently suggest that the long-range connections mainly within the DMN and DAN regions are more vulnerable to aging effects compared with the short-range connections [31], [32], [36].

Age has a strong impact on the DMN throughout an individual's lifespan, and age-related changes in interregional functional connectivity exhibit spatially and temporally specific patterns [36], [37]. In a broad spectrum of populations, resting-state fMRI studies have shown reliable and replicable results [38]–[40]. Particularly in healthy older adults, such reliability has been supported by a recent test-retest study that evaluated different methods for data processing, including seed-based, independent component analyses and graph theoretical approaches [41].

A majority of the studies have focused on the aging-related DMN and DAN, simultaneously ignoring several specified high-order networks. The ECN, which includes the anterior prefrontal cortex, insular and frontal opercular cortices, and the dorsal anterior cingulated cortex [11], is modulated by dopamine and is functionally involved in cognitive control, attention shift control and decision-making [11], [42]. The behavioral theories of cognitive aging suggest that age-related decline in a range of cognitive tasks is mediated by selective impairment in executive processing functions [43]–[45]. Using neurobehavioral tests, such as an attention network test, previous studies have indicated that the executive effect is more significantly decreased with age than other functions, such as the alerting function and orienting attention function [46]. Although this observation indicates the existence of selective cognitive differences with advanced age, corresponding observations from neuroimaging are still required. Diffusion tensor MRI research provides a structural basis for the selective loss of executive functions by measuring the correlations between the changes in the frontal white matter integrity and the executive function performance assessed using the Trail Making Test [47]. A previous study demonstrated that moderate exercise attenuates the disruption of age-related resting brain connectivity between the regions in the ECN and in the DMN [48]. Although there have been a few examinations of age-related disruption in intrinsic functional connectivity in the executive control and salience networks [47]–[49], much of the literature supports age-related decline in the performance of tasks implanted in the ECN (for reviews, see [33], [45], [50]).

Several studies have measured topological whole-brain connectivity by applying the graph theory, which demonstrated significant decreases in intermodular connections to the frontal modular regions in older groups and revealed that normal aging is associated with the functional segregation of large-scale brain systems [22], [31], [37]. One study indicated that aging impacts not only connectivity within networks but also connectivity between different functional networks [51]. Indeed, a specific brain network may have more relevance to a specific function than to other functions. However, whether aging has equivalent effects or selective effects on specific large-scale networks has not been established. Thus, the purpose of our study was to investigate the effects of aging on four canonical RSNs including the DMN, DAN, ECN and SN by comparing two groups consisting of younger and older adults. We hypothesize that the alterations in the canonical networks would vary across the networks; therefore, we are especially concerned with the degree to which the large-scale intrinsic networks are affected. Because previous studies have shown that reduced functional connectivity in the motor network is related to aging [52], we used the visual system network as a control.

Materials and Methods

Subjects and procedure

This study was approved by the Yangzhou University Research Ethics Committee. All the participants provided their written informed consent.

A total of 40 right-handed healthy subjects were recruited, including 18 younger subjects (age, 22–33 years; mean age, 23.9±1.8 years; n = 9 males, 9 females) and 22 older adults (age, 60–80 years; mean age, 69.8±5.8 years; n = 10 males, 12 females). Both groups were education- and gender-matched. None of the subjects had a history of head trauma, neuropsychiatric disorders, hypertensive disease or metabolic disorders. Additionally, none of the subjects showed abnormal findings in their structural brain MRIs. Each subject had a Mini-Mental State Examination (MMSE) score ranging between 28 and 30.

Imaging was performed using a 1.5T GE Signa Excite MR scanner (GE Healthcare Systems, Milwaukee, WI, USA) and a standard head coil. Before the fMRI scan, the subjects were instructed to lie quietly with their eyes closed and to avoid thinking of anything in particular. The functional images were obtained using a gradient echo-planar imaging (EPI) sequence (TR, 3000 ms; TE, 40 ms; flip angle, 90°; slice thickness, 6 mm; slice gap, 0 mm; FOV, 240 mm; and matrix, 64×64), and each frame included 18 contiguous slices that covered the entire cerebral volume. The slices were obtained parallel to the anterior/posterior commissures. The EPI scan lasted 6 min and 24 sec. Finally, a T1-weighted 3D fast spoiled gradient echo sequence was acquired (TR, 70 ms; TE, 4.2 ms; FOV, 240 mm; matrix, 256×256; slice thickness, 1 mm).

Data preprocessing and analysis

The fMRI data preprocessing was performed using DPARSF software V2.3 (http://www.restfmri.net), which is based on SPM8 (http://www.fil.ion.ucl.ac.uk/spm) and the Resting-State fMRI Data Analysis Toolkit (Beijing Normal University, Beijing, http://www.restfmri.net). The first four volumes were discarded. The images were corrected for slice timing and realigned for head movement correction. One young subject and two elderly subjects who failed to meet the study criterion were excluded because their head translation exceeded 1.5 mm or their head rotation exceeded 1.5°. There was no statistical significance between the groups for the movement parameters (two sample t test; the p values were 0.92, 0.93, and 0.55 for maximum, squared, and framewise displacement of the translation, respectively, and 0.60, 0.21, 0.12 for maximum, squared, and framewise displacement of the rotation, respectively). The functional images were normalized using DARTEL. The normalized volumes were re-sampled to a voxel size of 3 mm × 3 mm × 3 mm in MNI space, and the EPI images were spatially smoothed using an isotropic Gaussian filter (8 mm FWHM). The movement parameters (Friston 24-parameter model [53]) and signals of white matter, cerebrospinal fluid and global mean were regressed out. Linear detrending and temporal bandpass filtering (0.01–0.08 Hz) were applied.

The subsequent data processing and statistical analyses were performed using the Resting state fMRI data Analysis Toolkit (REST V1.8) (http://www.restfmri.net). For connectivity analysis, we created eight seed regions of interest (ROIs) with 6-mm radius spheres, which have been used in previous studies to identify the corresponding RSNs [11], [16], [54]. Each RSN and the seed ROI (in MNI coordinates) were as follows: DMN (PCC: −2, −54, 27; ventral medial prefrontal cortex: −2, 55, 6); ECN (left and right dorsolateral prefrontal cortex (dLPFC): −42, 34, 20/44, 36, 20); DAN (left and right superior parietal lobule (SPL): −25, −53, 52/25, −57, 52); and SN (left and right frontal-insular cortex (FI): −32, 26, −14/38, 22, −10). The positive Pearson correlation coefficients between the time series of each ROI and the time series in other voxels in the brain were calculated. A Fisher's z-transform was applied to improve the normality of these correlation coefficients. For the DMN, connectivity maps derived from the anterior and posterior seeds were averaged to create a single connectivity map; for the other canonical networks with bilateral ROI seeds, the connectivity maps derived from left and right ROI were averaged to create a single connectivity map, consistent with a prior study [16].

For the visual network, we selected two allelic seeds located approximately in the bilateral V2 primary cortex, which had been used in previous studies for mapping visual networks [35], [55]. The two seeds are centered on MNI coordinates (19, −95, 2) and (−19, −95, 2) with a 6-mm radius. The mean time course was extracted and analyzed using correlation analysis.

One-sample t tests were performed on the individual z-maps to determine initial intra-group level network regions; the threshold was set at AlphaSim corrected p<0.01, cluster size>97. Considering the representation of other nodes within each RSN and the bias of the pre-defined seed approach, a dual regression process was performed (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/DualRegression). The dual regression process includes the following steps: 1) the initial group spatial maps are regressed into each subject's 4D dataset to give a set of time courses; 2) those time courses are regressed into the same 4D dataset to obtain a subject-specific set of spatial maps; 3) then, within-group and between-group analyses are performed based on the spatial maps acquired in the prior step. Dual regression has been applied in many brain network research studies [56]–[58]. Using dual regression, Wang et al. discovered that the network-wise temporal patterns for the majority of the RSNs (especially for the DMN) exhibited moderate-to-high test-retest reliability and reproducibility under different scan conditions [56]. For within-group statistics, a threshold adjustment method based on the Monte-Carlo simulation correction was used with voxel wise p<0.8×10⁻⁵, cluster size>102 (2754 mm³), and cluster connectivity criterion 4 rmm; this outcome yielded an AlphaSim correction threshold of p<0.005. Before the group comparison of each map, the within-group threshold maps were combined across the young and elderly groups to create unified masks, which were then used to restrict the between-group analysis. The between-group two sample t test analysis was performed at an AlphaSim correction threshold of p<0.01, cluster size>97 (2619 mm³), and cluster connectivity criterion 4 rmm based on Monte-Carlo simulation correction. The individual modulated gray matter volumes were entered as covariates to regress out the nuisance from the brain volume differences between groups.

Results

Within-group analyses of brain networks

As shown in Figure 1, the DMN consisted of the PCC, ventral medial prefrontal cortex (vmPFC), bilateral medial temporal cortex (MTC), hippocampus, pulvinars, precuneus and inferior parietal cortex. The regions of the bilateral insular cortex, dorsolateral prefrontal cortex (dlPFC), inferior parietal cortex and posterior middle temporal (area MT+) were involved in the ECN. The DAN consisted of the SPL/intraparietal sulcus (IPS), frontal eye fields (FEF), and extra-striate visual areas, and the SN included the orbital frontal-insular cortex, anterior cingulated cortex, superior temporal pole and subcortical paralimbic regions. The visual network spread across the bilateral primary and association visual cortex. A visual inspection of the RSNs indicated that the connectivity maps for both groups were similar and consistent with prior findings.

[Figure omitted. See PDF.]

Figure 1. Intra-group maps of canonical networks in the resting brains of younger and older groups.

DAN, dorsal attention network; DMN, default mode network; ECN, executive control network; SN: salience network; VN, visual network; R, right view; L, left view. The color bar denotes the T value. The statistical threshold was set at p<0.005 and was corrected with AlphaSim.

https://doi.org/10.1371/journal.pone.0108807.g001

Between-group analyses of brain networks

Distinct alterations were noted in the four canonical networks between the younger and older groups when compared at the same statistical threshold level. In other words, the ECN was disrupted mainly in the older group, followed by the DAN; the DMN and SN showed limited functional connectivity disruption relative to the ECN and DAN.

For the ECN, 10 clusters (total of 2,668 voxels, 72,036 mm³) presented significantly decreased connectivity in the older group, including regions of the bilateral dorsal frontal cortex, insular cortex, MT+ and right inferior parietal lobe, with a rightward hemisphere asymmetry. In the DAN, 6 clusters (total of 1,400 voxels, 37,800 mm³) displayed reduced connectivity in the areas of the parietal postcentral gyrus, supplementary motor area and left mid-occipital gyrus, with a leftward asymmetry. For the DMN, 4 clusters (total of 612 voxels, 16,524 mm³) involving the bilateral regions of vMPFC, PCC, pulvinars and bilateral dorsal MPFC presented significantly reduced connectivity. The SN showed the least connectivity reduction in the regions of the left frontal-insular cortex and right superior temporal cortex, with 3 clusters (310 voxels, 8,370 mm³) (see Figure 2, details in Table 1). Significantly increased connectivity was not detected in any of the canonical networks in the older group.

[Figure omitted. See PDF.]

Figure 2. Comparisons of canonical networks between the younger and older groups.

The visual network did not exhibit significant differences and is not shown on the map. The right and left lateralized disruptions are shown in the ECN and DAN, respectively. The threshold was set at p<0.01 and was corrected with AlphaSim. ECN, executive control network; DAN, dorsal attention network; DMN, default mode network; SN, salience network; R, right view; L, left view. The color bar denotes the T value.

https://doi.org/10.1371/journal.pone.0108807.g002

[Figure omitted. See PDF.]

Table 1. Primary clusters showing significant connectivity alterations in resting-state networks in younger and older groups.

https://doi.org/10.1371/journal.pone.0108807.t001

In the control analysis, the visual networks did not demonstrate significant connectivity differences between the two experimental groups.

Considering that the canonical networks have different scales, the volume of connectivity decrease for each network was compared with the volume of each combined mask of the paired networks for the two groups (ratio of disruptive network volume) (Figure 3). For the ECN, DAN, DMN and SN, each disruptive network volume accounted for 25.7%, 12.7%, 5.7% and 4.9%, respectively, in each unified mask volume. Although the DAN had the largest volume, it was the second most impaired network in the older group. The map for the ratio of disruptive network volume can reflect the varying degree to which the canonical networks were affected.

[Figure omitted. See PDF.]

Figure 3. Histogram of the ratio of the disruptive network volume.

The light blue bars represent each of the unified mask volumes in the canonical networks; the dark pink bars represent each of the disruptive network volumes in the canonical networks. In the older group, the network volumes were reduced by 25.7%, 12.7%, 5.7% and 4.9% compared with each unified mask for the DMN, DAN, ECN and SN, respectively. This histogram reflects the distinct influence of aging on the canonical networks.

https://doi.org/10.1371/journal.pone.0108807.g003

Because there are debates concerning global signal regression, which could impact the resting-state fMRI inferences [59] and might reduce the effects of age-related differences on the connectivity in the blood flow, we estimated the results without global signal regression. We observed that the areas involved in each of the canonical networks exceeded the traditional network regions with positively biased correlations for the intra-group analyses, which is consistent with a recent study [60], and we found that the nuisance from the cerebrospinal fluid was incorporated into the maps for the inter-group analyses. DMN contrasting maps with and without global signal regression were used to illustrate the findings (Figure 4). The network maps without global signal regression for the intragroup and intergroup analyses are shown in Figure S1 and Figure S2. We believe that the global signal regression procedure might be appropriate in our study.

[Figure omitted. See PDF.]

Figure 4. DMN contrasting maps with and without global signal regression for the two experimental groups.

“Young” indicates the intra-group analyses with one t test for the younger adults. “Older” indicates the intra-group analyses with one t test for the older adults. The threshold for the intragroup and intergroup statistical analyses was set at p<0.01 and was corrected by AlphaSim. Note the positively biased correlations and the non-neural noise in the basilar cistern without global signal regression.

https://doi.org/10.1371/journal.pone.0108807.g004

Discussion

Age-related functional brain changes have been extensively explored during tasks and during resting states; however, less is known regarding the effect of normal aging on large-scale brain networks. The present study analyzes the effects of aging on resting-state canonical networks. Our major finding is that selective vulnerability is present in large-scale resting brain networks in normal older adults. The functional connectivity in these canonical networks was selectively reduced as a function of age, even though these networks varied in size.

The most intensive connectivity disruption was demonstrated in the ECN, followed by the DAN and DMN. The salience network was minimally affected, and the visual network, which served as the control, was not significantly affected. These findings are similar to the common observation that age-related processes are characterized by the loss of neural specificity within distinct functional systems [61]. Reduced signal coordination within specific networks may imply less effective functional communication in that specific brain system.

Previous research has shown that clinical neurodegenerative syndromes involving AD, behavioral variant frontotemporal dementia, semantic dementia and progressive nonfluent aphasia each include distinct network vulnerability patterns. For example, the DMN is vulnerable in AD, and the behavioral variant frontotemporal dementia targets the SN [12]–[14]; schizophrenia-vulnerable networks include the ECN and DAN [16]. Our findings suggest that normal aging is also characterized by selective neuronal network vulnerability, with ECN as the first target.

Executive control function, which determines the manipulation and modulation of concrete information processing for many cognitive tasks, is thought to be a higher order cognitive activity that is critically dependent on the ECN. Cognitive tasks require segregated and integrated processing in these large-scale brain networks; for example, the ECN may flexibly couple with the DMN and DAN according to the task domain [5], [10], [62]. Connectivity disruption in the ECN implies that the communication signals between these higher order brain networks have been impaired. Neurobehavioral tests have suggested that normal aging is accompanied by a decline in a range of cognitive abilities that are thought to rely on executive functioning and that executive skills seem to be particularly vulnerable to the effects of aging [43]–[46], [50]. By demonstrating that the ECN is impaired most severely among the canonical brain networks, our findings could provide support for the behavioral theories of cognitive aging.

Several studies have demonstrated the influence of aging on large-scale networks during various task states; for instance, during a semantic classification task, the DMN and DAN are markedly disrupted in advanced aging [35]. Grady et al. measured the connectivity within the DMN and task-positive networks during multiple cognitive tasks [63]. These authors noted that the functional connectivity in the DMN was reduced in older adults, whereas the pattern of task-positive network connectivity was equivalent in younger and older groups [63]. Task-directed age-comparative fMRI investigations showed that older adults have decreased and increased brain activation patterns and that these patterns can shift depending on the degree of the task demand [64]–[66]. Although the aging brain preserves the adaptive function for supporting explicit tasks, our findings indicated that the underlying brain networks were disrupted to varying degrees. Therefore, studying the aging brain using resting-state network connectivity measurements may be advantageous.

The DMN participates in episodic memory, presenting nonspecific deactivation during externally directed tasks. The DMN is active when individuals are engaged in internally focused tasks, including retrieving autobiographical memory, envisioning the future, and conceiving the perspectives of others [3], [9]–[11], [67], [68]. Disruption of the DMN implies impairment in these associated functions. Based on our findings, the DMN was mildly disrupted relative to the ECN and DAN, even though the DMN was sensitive to aging, as previously documented.

Currently, there are two models of hemispheric asymmetry about aging brain: the right hemi-aging model and the hemispheric asymmetry reduction in older adults (HAROLD) model. The former model, which proposes that the right hemisphere shows greater age-related decline than the left hemisphere, is supported by behavioral studies in the domains of cognitive, affective, and sensorimotor processing; the latter model proposes that the prefrontal activity during cognitive performance, which tends to be less left-lateralized in older adults than in younger adults, is supported by evidence from task-directed fMRI [69], [70]. In terms of our present observations on the asymmetric distribution patterns of the ECN and DAN, the two models are not incompatible; e.g., the right lateralized disruption in the ECN may support the right hemi-aging model, and the left lateralized disruption in the DAN may support the HAROLD model. The aging brain models could depend on which network resource is more demanding.

The salience network was affected minimally in the older group, with decreased coordination between regions of the left FI and right superior temporal cortex. The salience network is involved in emotion processing, reward and interoceptive regulation [5], [71], [72]. Previous research has suggested that, on average, the emotional well-being of adults is stable and well maintained or even improved as they age, despite the fact that their physical health and cognitive abilities decline with age [73], [74]. Our findings regarding SN changes in the elderly group could support this proposal.

The visual cortex, belonging to a low-order brain system, has broad fiber tracts connecting to the frontal and temporal lobes. Although the higher-order networks exhibit broadly reduced functional cooperation, our results showed that the visual network was spared during the aging process, which is consistent with previous findings [35]. According to supportive anatomical evidence, the occipital visual areas have the highest neuronal density and the most differential cytoarchitectonic structures in all cortices of the brain [75], [76]. Although many studies indicated that the higher visual attentional function in older adults was impaired, our findings may only reflect the primary photic sense function in older adults.

Selective neuronal and molecular vulnerability could account for the selective network vulnerability in the aging brain. A pattern of selective loss of synapses and neurons in certain brain regions has been described during the aging process [77]–[79]. Genomic analyses have suggested that different molecular mechanisms exist in different neural subtypes that determine their survival or vulnerability to death, which is evidenced by several transgenic mouse model subsets [15].

According to our findings, the older group presented decreased resting functional connectivity within multiple networks rather than increased functional connectivity; therefore, we hypothesize that reduced brain functional cooperation could be a critical pattern as aging progresses.

We acknowledge that the present study has several limitations. First, the relationship between the alteration of networks and behavioral performance requires additional investigation. Second, age-related preclinical atherosclerosis risks may have several effects on the brain's functional connectivity; for example, changes in interactions between the amygdala and rostral ACC have been implicated in an increased risk of atherosclerosis [80]. Thus, atherosclerosis co-variations should be quantified in the future. Additionally, the DMN is inhomogeneous and was decomposed into two or three sub-networks in previous studies [81], [82]. The DMN was not further subdivided in our study.

In summary, our results suggest that the process of aging in the brain is characterized by selective vulnerability in large-scale brain systems. Specifically, the ECN, DAN and DMN are among the canonical networks that are sensitive to the effects of aging. Moreover, high-order brain networks are preferentially affected compared to low-order systems. These findings could provide insight to further our understanding of patterns in age-related decline in multiple cognitive functions.

Supporting Information

[Figure omitted. See PDF.]

Figure S1.

Intra-group maps of canonical networks without global signal regression in the resting brains of younger and older groups. DAN, dorsal attention network; DMN, default mode network; ECN, executive control network; SN: salience network; VN, visual network; R, right view; L, left view. The color bar denotes the T value. The statistical threshold was set at p<0.001 and was corrected with AlphaSim. The map shows that the areas involved in each of the canonical networks exceed the traditional network regions.

https://doi.org/10.1371/journal.pone.0108807.s001

(TIF)

Figure S2.

Comparison of the canonical networks between the younger and older groups without global signal regression procedure. DAN, dorsal attention network; DMN, default mode network; ECN, executive control network; SN: salience network. Left is left. The color bar denotes the T value. The statistical threshold was set at p<0.01 and was corrected with AlphaSim. The prominent nuisance from the cerebrospinal fluid, arteries and veins can be noted, particularly on maps of the SN and DMN.

https://doi.org/10.1371/journal.pone.0108807.s002

(TIF)

Author Contributions

Conceived and designed the experiments: HYZ JTW. Analyzed the data: HYZ YJ. Contributed reagents/materials/analysis tools: HYZ JTW. Wrote the paper: HYZ WXC YJ YX XRZ JTW. Contributed to data interpretation: HYZ WXC YJ YX XRZ JTW.

Citation: Zhang H-Y, Chen W-X, Jiao Y, Xu Y, Zhang X-R, Wu J-T (2014) Selective Vulnerability Related to Aging in Large-Scale Resting Brain Networks. PLoS ONE 9(10): e108807. https://doi.org/10.1371/journal.pone.0108807

References

1. Hedden T, Gabrieli JD (2005) Healthy and pathological processes in adult development: new evidence from neuroimaging of the aging brain. Curr Opin Neurol 18: 740–747.

2. Silver H, Goodman C, Bilker W (2009) Age in high-functioning healthy men is associated with nonlinear decline in some ‘executive’ functions in late middle age. Dement Geriatr Cogn Disord 27: 292–300.

3. Greicius MD, Krasnow B, Reiss AL, Menon V (2003) Functional connectivity in the resting brain: a network analysis of the default mode hypothesis. Proc Natl Acad Sci U S A 100: 253–258.

4. Fox MD, Corbetta M, Snyder AZ, Vincent JL, Raichle ME (2006) Spontaneous neuronal activity distinguishes human dorsal and ventral attention systems. Proc Natl Acad Sci U S A 103: 10046–10051.

5. Seeley WW, Menon V, Schatzberg AF, Keller J, Glover GH, et al. (2007) Dissociable intrinsic connectivity networks for salience processing and executive control. J Neurosci 27: 2349–2356.

6. Hampson M, Driesen NR, Skudlarski P, Gore JC, Constable RT (2006) Brain connectivity related to working memory performance. J Neurosci 26: 13338–13343.

7. Biswal B, Yetkin FZ, Haughton VM, Hyde JS (1995) Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn Reson Med 34: 537–541.

8. Li SJ, Biswal B, Li Z, Risinger R, Rainey C, et al. (2000) Cocaine administration decreases functional connectivity in human primary visual and motor cortex as detected by functional MRI. Magn Reson Med 43: 45–51.

9. Fox MD, Snyder AZ, Vincent JL, Corbetta M, Van Essen DC, et al. (2005) The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A 102: 9673–9678.

10. Spreng RN, Stevens WD, Chamberlain JP, Gilmore AW, Schacter DL (2010) Default network activity, coupled with the frontoparietal control network, supports goal-directed cognition. Neuroimage 53: 303–317.

11. Vincent JL, Kahn I, Snyder AZ, Raichle ME, Buckner RL (2008) Evidence for a frontoparietal control system revealed by intrinsic functional connectivity. J Neurophysiol 100: 3328–3342.

12. Seeley WW, Crawford RK, Zhou J, Miller BL, Greicius MD (2009) Neurodegenerative diseases target large-scale human brain networks. Neuron 62: 42–52.

13. Zhou J, Greicius MD, Gennatas ED, Growdon ME, Jang JY, et al. (2010) Divergent network connectivity changes in behavioural variant frontotemporal dementia and Alzheimer's disease. Brain 133: 1352–1367.

14. Seeley WW (2008) Selective functional, regional, and neuronal vulnerability in frontotemporal dementia. Curr Opin Neurol 21: 701–707.

15. Gotz J, Schonrock N, Vissel B, Ittner LM (2009) Alzheimer's disease selective vulnerability and modeling in transgenic mice. J Alzheimers Dis 18: 243–251.

16. Woodward ND, Rogers B, Heckers S (2011) Functional resting-state networks are differentially affected in schizophrenia. Schizophr Res 130: 86–93.

17. Grady CL, Springer MV, Hongwanishkul D, McIntosh AR, Winocur G (2006) Age-related changes in brain activity across the adult lifespan. J Cogn Neurosci 18: 227–241.

18. Ansado J, Monchi O, Ennabil N, Faure S, Joanette Y (2012) Load-dependent posterior-anterior shift in aging in complex visual selective attention situations. Brain Res.

19. Lustig C, Snyder AZ, Bhakta M, O'Brien KC, McAvoy M, et al. (2003) Functional deactivations: change with age and dementia of the Alzheimer type. Proc Natl Acad Sci U S A 100: 14504–14509.

20. Damoiseaux JS, Beckmann CF, Arigita EJ, Barkhof F, Scheltens P, et al. (2008) Reduced resting-state brain activity in the “default network” in normal aging. Cereb Cortex 18: 1856–1864.

21. Koch W, Teipel S, Mueller S, Buerger K, Bokde AL, et al. (2010) Effects of aging on default mode network activity in resting state fMRI: does the method of analysis matter? Neuroimage 51: 280–287.

22. Meunier D, Achard S, Morcom A, Bullmore E (2009) Age-related changes in modular organization of human brain functional networks. Neuroimage 44: 715–723.

23. Wu JT, Wu HZ, Yan CG, Chen WX, Zhang HY, et al. (2011) Aging-related changes in the default mode network and its anti-correlated networks: A resting-state fMRI study. Neurosci Lett 504: 62–67.

24. Goh JO (2011) Functional Dedifferentiation and Altered Connectivity in Older Adults: Neural Accounts of Cognitive Aging. Aging Dis 2: 30–48.

25. Wang L, Laviolette P, O'Keefe K, Putcha D, Bakkour A, et al. (2010) Intrinsic connectivity between the hippocampus and posteromedial cortex predicts memory performance in cognitively intact older individuals. Neuroimage 51: 910–917.

26. Logan JM, Sanders AL, Snyder AZ, Morris JC, Buckner RL (2002) Under-recruitment and nonselective recruitment: dissociable neural mechanisms associated with aging. Neuron 33: 827–840.

27. Madden DJ, Whiting WL, Provenzale JM, Huettel SA (2004) Age-related changes in neural activity during visual target detection measured by fMRI. Cereb Cortex 14: 143–155.

28. Cabeza R, Daselaar SM, Dolcos F, Prince SE, Budde M, et al. (2004) Task-independent and task-specific age effects on brain activity during working memory, visual attention and episodic retrieval. Cereb Cortex 14: 364–375.

29. Kukolja J, Thiel CM, Wilms M, Mirzazade S, Fink GR (2009) Ageing-related changes of neural activity associated with spatial contextual memory. Neurobiol Aging 30: 630–645.

30. Gutchess AH, Welsh RC, Hedden T, Bangert A, Minear M, et al. (2005) Aging and the neural correlates of successful picture encoding: frontal activations compensate for decreased medial-temporal activity. J Cogn Neurosci 17: 84–96.

31. Wang L, Li Y, Metzak P, He Y, Woodward TS (2010) Age-related changes in topological patterns of large-scale brain functional networks during memory encoding and recognition. Neuroimage 50: 862–872.

32. Tomasi D, Volkow ND (2012) Aging and functional brain networks. Mol Psychiatry 17: : 471, 549–458.

33. Ferreira LK, Busatto GF (2013) Resting-state functional connectivity in normal brain aging. Neurosci Biobehav Rev 37: 384–400.

34. Sambataro F, Murty VP, Callicott JH, Tan HY, Das S, et al. (2010) Age-related alterations in default mode network: impact on working memory performance. Neurobiol Aging 31: 839–852.

35. Andrews-Hanna JR, Snyder AZ, Vincent JL, Lustig C, Head D, et al. (2007) Disruption of large-scale brain systems in advanced aging. Neuron 56: 924–935.

36. Cao M, Wang JH, Dai ZJ, Cao XY, Jiang LL, et al. (2014) Topological organization of the human brain functional connectome across the lifespan. Dev Cogn Neurosci 7: 76–93.

37. Wang L, Su L, Shen H, Hu D (2012) Decoding lifespan changes of the human brain using resting-state functional connectivity MRI. PLoS One 7: e44530.

38. Buckner RL, Sepulcre J, Talukdar T, Krienen FM, Liu H, et al. (2009) Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of stability, and relation to Alzheimer's disease. J Neurosci 29: 1860–1873.

39. Zuo XN, Di Martino A, Kelly C, Shehzad ZE, Gee DG, et al. (2010) The oscillating brain: complex and reliable. Neuroimage 49: 1432–1445.

40. Meindl T, Teipel S, Elmouden R, Mueller S, Koch W, et al. (2010) Test-retest reproducibility of the default-mode network in healthy individuals. Hum Brain Mapp 31: 237–246.

41. Guo CC, Kurth F, Zhou J, Mayer EA, Eickhoff SB, et al. (2012) One-year test-retest reliability of intrinsic connectivity network fMRI in older adults. Neuroimage 61: 1471–1483.

42. Diamond A, Briand L, Fossella J, Gehlbach L (2004) Genetic and neurochemical modulation of prefrontal cognitive functions in children. Am J Psychiatry 161: 125–132.

43. Braver TS, Barch DM (2002) A theory of cognitive control, aging cognition, and neuromodulation. Neurosci Biobehav Rev 26: 809–817.

44. Goh JO, Park DC (2009) Neuroplasticity and cognitive aging: the scaffolding theory of aging and cognition. Restor Neurol Neurosci 27: 391–403.

45. Bryan J, Luszcz MA (2000) Measurement of executive function: considerations for detecting adult age differences. J Clin Exp Neuropsychol 22: 40–55.

46. Zhou SS, Fan J, Lee TM, Wang CQ, Wang K (2011) Age-related differences in attentional networks of alerting and executive control in young, middle-aged, and older Chinese adults. Brain Cogn 75: 205–210.

47. O'Sullivan M, Jones DK, Summers PE, Morris RG, Williams SC, et al. (2001) Evidence for cortical “disconnection” as a mechanism of age-related cognitive decline. Neurology 57: 632–638.

48. Voss MW, Prakash RS, Erickson KI, Basak C, Chaddock L, et al.. (2010) Plasticity of brain networks in a randomized intervention trial of exercise training in older adults. Front Aging Neurosci 2..

49. Onoda K, Ishihara M, Yamaguchi S (2012) Decreased functional connectivity by aging is associated with cognitive decline. J Cogn Neurosci 24: 2186–2198.

50. Park DC, Reuter-Lorenz P (2009) The adaptive brain: aging and neurocognitive scaffolding. Annu Rev Psychol 60: 173–196.

51. Geerligs L, Renken RJ, Saliasi E, Maurits NM, Lorist MM (2014) A Brain-Wide Study of Age-Related Changes in Functional Connectivity. Cereb Cortex.

52. Wu T, Zang Y, Wang L, Long X, Hallett M, et al. (2007) Aging influence on functional connectivity of the motor network in the resting state. Neurosci Lett 422: 164–168.

53. Friston KJ, Williams S, Howard R, Frackowiak RS, Turner R (1996) Movement-related effects in fMRI time-series. Magn Reson Med 35: 346–355.

54. Uddin LQ, Kelly AM, Biswal BB, Xavier Castellanos F, Milham MP (2009) Functional connectivity of default mode network components: correlation, anticorrelation, and causality. Hum Brain Mapp 30: 625–637.

55. Konishi S, Wheeler ME, Donaldson DI, Buckner RL (2000) Neural correlates of episodic retrieval success. Neuroimage 12: 276–286.

56. Wang X, Jiao Y, Tang T, Wang H, Lu Z (2013) Investigating univariate temporal patterns for intrinsic connectivity networks based on complexity and low-frequency oscillation: a test-retest reliability study. Neuroscience 254: 404–426.

57. Filippini N, MacIntosh BJ, Hough MG, Goodwin GM, Frisoni GB, et al. (2009) Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A 106: 7209–7214.

58. Washington SD, Gordon EM, Brar J, Warburton S, Sawyer AT, et al. (2014) Dysmaturation of the default mode network in autism. Hum Brain Mapp 35: 1284–1296.

59. Saad ZS, Gotts SJ, Murphy K, Chen G, Jo HJ, et al. (2012) Trouble at rest: how correlation patterns and group differences become distorted after global signal regression. Brain Connect 2: 25–32.

60. Hayasaka S (2013) Functional connectivity networks with and without global signal correction. Front Hum Neurosci 7: 880.

61. Reuter-Lorenz PA, Park DC (2010) Human neuroscience and the aging mind: a new look at old problems. J Gerontol B Psychol Sci Soc Sci 65: 405–415.

62. Bressler SL, Kelso JA (2001) Cortical coordination dynamics and cognition. Trends Cogn Sci 5: 26–36.

63. Grady CL, Protzner AB, Kovacevic N, Strother SC, Afshin-Pour B, et al. (2010) A multivariate analysis of age-related differences in default mode and task-positive networks across multiple cognitive domains. Cereb Cortex 20: 1432–1447.

64. Persson J, Nyberg L (2006) Altered brain activity in healthy seniors: what does it mean? Prog Brain Res 157: 45–56.

65. Smith EE, Geva A, Jonides J, Miller A, Reuter-Lorenz P, et al. (2001) The neural basis of task-switching in working memory: effects of performance and aging. Proc Natl Acad Sci U S A 98: 2095–2100.

66. Clapp WC, Rubens MT, Sabharwal J, Gazzaley A (2011) Deficit in switching between functional brain networks underlies the impact of multitasking on working memory in older adults. Proc Natl Acad Sci U S A 108: 7212–7217.

67. Buckner RL, Andrews-Hanna JR, Schacter DL (2008) The brain's default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci 1124: 1–38.

68. Fransson P (2005) Spontaneous low-frequency BOLD signal fluctuations: an fMRI investigation of the resting-state default mode of brain function hypothesis. Hum Brain Mapp 26: 15–29.

69. Dolcos F, Rice HJ, Cabeza R (2002) Hemispheric asymmetry and aging: right hemisphere decline or asymmetry reduction. Neurosci Biobehav Rev 26: 819–825.

70. Cabeza R (2002) Hemispheric asymmetry reduction in older adults: the HAROLD model. Psychol Aging 17: 85–100.

71. Ongur D, Price JL (2000) The organization of networks within the orbital and medial prefrontal cortex of rats, monkeys and humans. Cereb Cortex 10: 206–219.

72. Menon V, Levitin DJ (2005) The rewards of music listening: response and physiological connectivity of the mesolimbic system. Neuroimage 28: 175–184.

73. Mather M (2012) The emotion paradox in the aging brain. Ann N Y Acad Sci 1251: 33–49.

74. Nashiro K, Sakaki M, Mather M (2012) Age differences in brain activity during emotion processing: reflections of age-related decline or increased emotion regulation? Gerontology 58: 156–163.

75. Zezula J, Cortes R, Probst A, Palacios JM (1988) Benzodiazepine receptor sites in the human brain: autoradiographic mapping. Neuroscience 25: 771–795.

76. la Fougere C, Grant S, Kostikov A, Schirrmacher R, Gravel P, et al. (2011) Where in-vivo imaging meets cytoarchitectonics: the relationship between cortical thickness and neuronal density measured with high-resolution [18F]flumazenil-PET. Neuroimage 56: 951–960.

77. Morrison JH, Hof PR (2007) Life and death of neurons in the aging cerebral cortex. Int Rev Neurobiol 81: 41–57.

78. Morrison JH, Hof PR (2002) Selective vulnerability of corticocortical and hippocampal circuits in aging and Alzheimer's disease. Prog Brain Res 136: 467–486.

79. Wang X, Michaelis ML, Michaelis EK (2010) Functional genomics of brain aging and Alzheimer's disease: focus on selective neuronal vulnerability. Curr Genomics 11: 618–633.

80. Gianaros PJ, Hariri AR, Sheu LK, Muldoon MF, Sutton-Tyrrell K, et al. (2009) Preclinical atherosclerosis covaries with individual differences in reactivity and functional connectivity of the amygdala. Biol Psychiatry 65: 943–950.

81. Uddin LQ, Kelly AM, Biswal BB, Castellanos FX, Milham MP (2009) Functional connectivity of default mode network components: correlation, anticorrelation, and causality. Hum Brain Mapp 30: 625–637.

82. Andrews-Hanna JR, Reidler JS, Sepulcre J, Poulin R, Buckner RL (2010) Functional-anatomic fractionation of the brain's default network. Neuron 65: 550–562.