pISSN 1738-6586   eISSN 2005-5013
Open Access, Peer-reviewed
    J Clin Neurol. 2024 May;20(3):265-275. English.
    Published online Feb 05, 2024.
    https://doi.org/10.3988/jcn.2023.0092
    Copyright © 2024 Korean Neurological Association
    Original Article

    Alterations of Structural Network Efficiency in Early-Onset and Late-Onset Alzheimer’s Disease

    Suyeon Heo,a,* Cindy W Yoon,b,* Sang-Young Kim,c,d Woo-Ram Kim,c Duk L. Na,e,f and Young Nohc,g
      • aGachon University, College of Medicine, Incheon, Korea.
      • bDepartment of Neurology, Inha University School of Medicine, Incheon, Korea.
      • cNeuroscience Research Institute, Gachon University, Incheon, Korea.
      • dMR Clinical Science, Health Systems, Philips Healthcare, Seoul, Korea.
      • eDepartment of Neurology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea.
      • fHappymind Clinic, Seoul, Korea.
      • gDepartment of Neurology, Gil Medical Center, Gachon University College of Medicine, Incheon, Korea.
    • Correspondence: Young Noh, MD, PhD. Department of Neurology, Gil Medical Center, Gachon University College of Medicine, 21 Namdong-daero 774beon-gil, Namdong-gu, Incheon 21565, Korea. Tel +82-32-460-3346, Fax +82-32-460-3344, Email: ynoh@gilhospital.com

      *These authors contributed equally to this work.
    Received March 08, 2023; Revised August 17, 2023; Accepted October 05, 2023.

    This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

    Abstract

    Background and Purpose

    Early- and late-onset Alzheimer’s disease (EOAD and LOAD, respectively) share the same neuropathological hallmarks of amyloid and neurofibrillary tangles but have distinct cognitive features. We compared structural brain connectivity between the EOAD and LOAD groups using structural network efficiency and evaluated the association of structural network efficiency with the cognitive profile and pathological markers of Alzheimer’s disease (AD).

    Methods

    The structural brain connectivity networks of 80 AD patients (47 with EOAD and 33 with LOAD) and 57 healthy controls were reconstructed using diffusion-tensor imaging. Graph-theoretic indices were calculated and intergroup differences were evaluated. Correlations between network parameters and neuropsychological test results were analyzed. The correlations of the amyloid and tau burdens with network parameters were evaluated for the patients and controls.

    Results

    Compared with the age-matched control group, the EOAD patients had increased global path length and decreased global efficiency, averaged local efficiency, and averaged clustering coefficient. In contrast, no significant differences were found in the LOAD patients. Locally, the EOAD patients showed decreases in local efficiency and the clustering coefficient over a wide area compared with the control group, whereas LOAD patients showed such decreases only within a limited area. Changes in network parameters were significantly correlated with multiple cognitive domains in EOAD patients, but only with Clinical Dementia Rating Sum-of-Boxes scores in LOAD patients. Finally, the tau burden was correlated with changes in network parameters in AD signature areas in both patient groups, while there was no correlation with the amyloid burden.

    Conclusions

    The impairment of structural network efficiency and its effects on cognition may differ between EOAD and LOAD.

    Graphical Abstract

    Keywords
    Alzheimer disease; early-onset Alzheimer’s disease; late-onset Alzheimer’s disease; white-matter connectivity; diffusion-tensor imaging

    INTRODUCTION

    Alzheimer’s disease (AD) is the most common type of dementia. Increasing age is the main risk factor for developing AD, and most AD patients are aged ≥65 years, which is called late-onset AD (LOAD). People who experience the clinical onset of AD at <65 years of age are diagnosed with early-onset AD (EOAD). There are obvious differences in the pattern of cognitive impairment and rate of cognitive decline between LOAD and EOAD patients. The most common and prominent clinical presentation is memory loss in LOAD patients, whereas nonamnestic clinical presentations are more frequent in EOAD patients.1, 2 Cognitive decline is more rapid in EOAD than LOAD patients.3

    There is evidence that not only gray matter but also white matter (WM) is damaged in AD.4, 5, 6 Diffusion tensor imaging (DTI) is a magnetic resonance imaging (MRI)-based technique for detecting changes in the WM microstructure. A previous DTI study found that WM integrity was disrupted in AD.7 A more recent DTI study suggested that differences in the pattern of WM connectivity damage contribute to the distinct clinical profiles in EOAD and LOAD patients.8, 9

    WM connectivity networks can be constructed from DTI data. Connectivity analysis incorporating graph-theoretic approaches can be applied to characterize the local and global efficiencies of the structural networks. A previous study found reduced network efficiency in AD patients that was associated with cognitive dysfunction.10 We recently utilized a fractional anisotropy-weighted connectivity matrix to show how structural connectivity across the brain differs between EOAD and LOAD patients.9

    In the present study, we computed the network efficiency by measuring both local connectivity parameters (local efficiency and clustering coefficient) and global connectivity parameters (path length and global efficiency). We aimed to compare the structural network efficiency between EOAD and LOAD patients. In addition, we evaluated the associations of structural network efficiency with the cognitive profile and pathological markers of AD (amyloid and tau) in both AD groups.

    METHODS

    Participants

    Eighty AD patients, comprising 47 with EOAD (age 60.11±5.36 years [mean±standard deviation], 16 males) and 33 with LOAD (age 77.33±6.57 years, 9 males), were recruited through the Gachon University Gil Medical Center (Table 1). The study was conducted from October 2015 to June 2017. All AD patients satisfied the criteria for probable AD proposed by the National Institute of Neurological and Communicative Disorders and Stroke, and the Alzheimer’s Disease and Related Disorders Association.11 All patients underwent the clinical interview. None of the patients had a familial AD history of autosomal dominant inheritance. Patients with other structural lesions on brain MRI such as territorial infarction, intracranial hemorrhage, traumatic brain injury, hydrocephalus, multiple sclerosis, or severe WM hyperintensities (WMH, defined as a cap or a band with a maximum diameter ≥10 mm or deep ≥25 mm, according to modified Fazekas ischemia criteria)12, 13 were excluded from the study.

    Table 1
    Demographic and clinical characteristics of the study population

    Mini-Mental State Examination (MMSE), Clinical Dementia Rating (CDR), and Clinical Dementia Rating Sum-of-Boxes (CDR-SOB) scores were obtained to measure the severity of dementia. Neuropsychological functioning including attention, language function, visuospatial function, memory function, and frontoexecutive function was evaluated using the Seoul Neuropsychological Screening Battery (SNSB).14 The SNSB score of each participant was converted into the age- and education-corrected standard value (z-score). Detailed information about the test items is provided in Table 2.

    Table 2
    Neuropsychological test results of the study population

    Fifty-seven age-matched healthy controls were recruited for comparisons with the AD patients. The controls were recruited from among the spouses of the patients and through the local community. Participants were eligible for inclusion as healthy controls if they had a CDR score of 0 and normal cognitive function, defined as neuropsychological test scores within 1.5 standard deviations of the age- and education-corrected normative mean. We excluded individuals with neurological or psychiatric abnormalities identified in the neuropsychological test as well as those with structural brain lesions such as cerebral infarction, intracranial hemorrhage, traumatic injury, hydrocephalus, or severe WMH. For the comparisons with the EOAD and LOAD groups, the cognitively normal subjects were divided into age-matched control groups comprising 31 young controls (YC) (age 57.61±7.26 years, 19 males) and 26 old controls (OC) (age 74.42±4.38 years, 12 males) (Table 1).

    All participants gave written informed consent, and the study protocol was approved by the Institutional Review Board of the Gachon University Gil Medical Center (GDIRB2015-272).

    MRI data acquisition and preprocessing

    The structural (T1-weighted, T1w; T2-weighted, T2w) and DTI images were acquired using a 3-T MRI scanner (Verio, Siemens, Erlangen, Germany) at Gil Medical Center. 3D T1w images were acquired using the following parameters: repetition time (TR)=1,900 ms, echo time (TE)=2.93 ms, field of view (FOV)=256 mm, pixel bandwidth=170 Hz/pixel, flip angle=8°, voxel resolution=0.5×0.5×1.0 mm3, and total acquisition time=250 s. 2D T2w images were acquired using turbo spin-echo sequences with the following parameters: TR=9,650 ms, TE=88 ms, flip angle=120°, in-plain resolution=0.5 mm×0.5 mm, slice thickness=4 mm, turbo factor= 21, and number of averages=3. DTI images were obtained using single-shot spin-echo echo-planar imaging sequences with the following parameters: 30 diffusion-gradient directions with b=900 s/mm2 and 1 with b=0 s/mm2, TR=12,000 ms, TE=78 ms, FOV=256 mm, pixel bandwidth=1,502 Hz/pixel, flip angle=90°, voxel resolution=2×2×2 mm3, and total acquisition time=806 s.

    The MR images were preprocessed in order to accurately reconstruct the brain structural connectivity. Briefly, T1w and T2w MR images were roughly aligned to the Montreal Neurological Institute (MNI) template with a 1-mm isotropic resolution using rigid-body transformation, which aligns the anterior–posterior commissure (AC-PC) line while maintaining the original size and shape of the brain. Initial brain extraction was performed using linear and nonlinear registration of the image to the MNI template. This alignment allows the MNI-space brain mask to be brought back to the T1w space. Additionally, T1w and T2w MR images were cross-modally registered using the FLIRT with a boundary-based registration cost function in the FSL (FMRIB’s Software Library, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki), and bias-field correction was performed by estimating the bias field from the square root of the product of the T1w and T2w images. The WM and pial surfaces were reconstructed using FreeSurfer (version 5.3.0-HCP, https://surfer.nmr.mgh.harvard.edu/). For accurate pial surface placement, we used an improved algorithm that uses both the T1w and T2w images in order to exclude large proportions of the dura and blood vessels.

    The DTI data were firstly preprocessed with brain extraction followed by correcting eddy-current-induced distortions and subject movements using the FSL eddy tool (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/eddy). However, we were not able to correct susceptibility-induced geometric distortion since reversed phase-encoded DTI data were not available from routine clinical trials. Eddy-current-corrected DTI data were then registered to AC-PC-aligned T1w images that were resampled to an isotropic resolution of 1.5 mm. Finally, six elements of the diffusion-tensor matrix were obtained by fitting DTI data using FMRIB’s Diffusion Toolbox in the FSL (FDT).

    Positron emission tomography data acquisition and preprocessing

    Tau and amyloid positron emission tomography (PET) scans were obtained using a Siemens Biograph 6 Truepoint PET/CT scanner (Siemens) with LIST-mode emission acquisition and [18F]THK5351 (THK) and [18F]flutemetamol (FLUTE), respectively. Tau PET images were acquired at 50–70 minutes after a bolus injection of 185 MBq of THK, while amyloid PET images were acquired at 90–110 minutes after the intravenous injection of 186 MBq of FLUTE. Low-dose CT scans were performed prior to PET scans for attention correction. The average interval between the two PET scans was 10 days. PET images were corrected for physical effects and reconstructed using a 2D ordered-subset expectation maximization algorithm with the following parameters: 8 iterations and 16 subsets, 256×256×109 matrix, and voxel size=1.3×1.3×1.5 mm3. FLUTE and THK PET images were processed using PETSurfer pipelines (https://surfer.nmr.mgh.harvard.edu/fswiki/PetSurfer), which included MRI–PET coregistration and partial volume (PV) correction using the Muller-Gartner method. The standardized uptake value ratio (SUVR) was calculated using cerebellar gray matter as a reference for THK images and the pons for FLUTE images. The PV-corrected SUVR images were also mapped onto the MNI template. The SUVR maps for amyloid and tau deposition were smoothed using a Gaussian kernel with an FWHM of 8 mm. The global retention ratio was generated based on AD-related regions including the prefrontal, superior and inferior parietal, lateral temporal, and anterior and posterior cingulate cortices for FLUTE images.15 Patients were classified as amyloid-positive based on visually interpreting images of the frontal, temporal, parietal cortices, striatum, and precuneus (PCUN).

    Network construction

    The nodes of the network were defined using the automated anatomical labeling (AAL) atlas,16 which consisted of 45 cortical and subcortical regions of the brain for each hemisphere. The AAL masks were transformed from MNI space to native diffusion space using the matrix derived by inverting the warp field for the registration of diffusion to MNI space.

    The edges of the network were estimated by the streamlines between each pair of node. WM structural fiber streamlines were reconstructed by the Fiber Assignment by Continuous Tracking (FACT) algorithm using the Diffusion Toolkit.17 The termination criteria for tracking was when the fractional anisotrophy (FA) of the voxel was less than 0.2 or when an excessive turning angle (>35°) was detected between two connected vectors,18 based on the assumption that they were no longer on the same pathway. Fibers shorter than 20 mm were removed from the streamlines. Two nodes were defined as linked by an edge when they were connected by at least three fibers.19 The weight of the edge was scaled by the average FA along all of the fibers composing the connection. The FA value is suggested to represent the integrity of the connection, implying that it may incorporate WM damage20 and is known to be correlated with the conductance of brain signals.21 FA-weighted 90×90 symmetric matrices were estimated for each individual using the above-described procedure.

    Network topology measures

    Nodal characteristics of the network were computed using the Brain Connectivity Toolbox of MATLAB.22 Parameters that have previously been suggested to be damaged in AD10, 23 were selected for inclusion in the study, including global connectivity parameters (characteristic path length and global efficiency) and local connectivity parameters (local efficiency and clustering coefficient).

    The characteristic path length is the measure of the network integrity, computed as the average shortest path length between any two nodes in the network. Global efficiency is the harmonic mean of the inverse of the characteristic path length between the pairs of the node in the network. A shorter characteristic path length indicates higher global efficiency of parallel information flow.24

    The local efficiency is a measure of the average efficiency of the local clusters. The clustering coefficient quantifies the number of connections among the directly connected neighbors as a proportion of all of the possible connections. The local efficiency and clustering coefficient are indicators of the robustness and efficiency of communication between the nearest neighbors of a node.

    Statistical analysis

    Demographic and cognitive features were compared using the nonparametric Mann–Whitney U test. Group comparisons of nominal variables were conducted using the chi-square test. All neuropsychological test results were transformed into z-scores that took age and education duration into consideration. To reveal the statistical group difference of the network parameters, analysis of variance method was used while controlling the age, sex, and education duration. Relationships between two continuous variables were calculated using Spearman correlation analysis, except for that between MMSE scores and the network parameters, for which partial Spearman correlation analysis was used to correct for the effects of age, sex, and education duration. The false-discovery-rate (FDR) method was used for multiple comparisons. All statistical analyses were conducted using MINITAB software (version 14, MINITAB, State College, PA, USA) with a significance of p<0.05 (two-way).

    RESULTS

    Demographic, clinical, and cognitive features

    The demographic and clinical characteristics are presented in Table 1. Age did not differ significantly between AD patients and the age-matched control group. Compared with the corresponding age-matched control group, global THK and FLUTE retentions were significantly greater in the AD patients (p<0.001). The values of global FLUTE retention were higher for EOAD than LOAD patients (p<0.001). The THK and FLUTE retentions are compared between EOAD and LOAD patients in Supplementary Fig. 1 (in the online-only Data Supplement). Detailed neuropsychological test results are presented in Table 2.

    Comparison of WM connectivity

    To examine the AD-related disruptions of the brain network, network measures were compared between the AD and control groups. The EOAD group had a significantly higher global path length (F1,75=10.18, p=0.002) and a lower global efficiency (F1,75=9.34, p=0.003), averaged local efficiency (F1,75=13.37, p<0.001), and averaged clustering coefficient (F1,75=12.47, p=0.001) compared with YC. In contrast, no significant intergroup difference was observed between the LOAD group and OC (Fig. 1).

    Fig. 1
    Comparison of network measures. Group comparisons of global path length (A), global efficiency (B), averaged local efficiency (C), and averaged clustering coefficient (D). Error bars indicate standard deviations. Analysis of covariance (ANCOVA) with the covariates of age, sex, and education duration was used to identify significant differences between each Alzheimer’s disease group and its age-matched control group. ANCOVA with covariates of sex and education duration was used to identify significant differences between early-onset Alzheimer’s disease (EOAD) and late-onset Alzheimer’s disease (LOAD) patients. *Indicates significant difference at the 0.05 level. OC, old controls; YC, young controls.

    AD-related alterations in specific brain regions were identified by comparing local measures between groups. Local efficiency was lower in the EOAD group than in YC in the right angular gyrus (ANG), right calcarine (CAL), bilateral cuneus (CUN), right median cingulate and paracingulate gyrus, right hippocampus, bilateral inferior parietal lobule (IPL), right inferior temporal gyrus (ITG), bilateral lingual gyrus (LING), bilateral middle occipital gyrus (MOG), bilateral middle temporal gyrus (MTG), left olfactory cortex, left medial orbital of the superior frontal gyrus, bilateral PCUN, right parahippocampal gyrus, right postcentral gyrus, right gyrus rectus, right Rolandic operculum (ROL), right supplementary motor area (SMA), right supramarginal gyrus (SMG), bilateral superior occipital gyrus (SOG), left superior parietal gyrus (SPG), right middle temporal gyrus of the temporal pole (TPOmid), and left superior temporal gyrus of the temporal pole (TPOsup) (Fig. 2A). Meanwhile, local efficiency was lower in the LOAD group than in OC in the right MOG, right SOG, right SMA, right putamen, and right ITG (Fig. 2B). Local efficiency was lower in the EOAD than the LOAD group in the left ROL, right Heschl gyrus, and right TPOsup, and lower in the LOAD group in the left ITG, right superior temporal gyrus (STG), and TPOmid (Fig. 2C).

    Fig. 2
    Topography of the brain regions with significant group effects of the network parameters. ANCOVA with covariates of age, sex, and education duration was used. Colored nodes indicate significant group difference at the false-discovery-rate-corrected p<0.05 level. Nodes colored orange indicate significant decreases in network parameters in the AD group relative to the age-matched controls (A, B, and D). None of the nodes shows significant group differences in the opposite direction. Nodes showing higher network parameters in LOAD and EOAD patients are colored red and blue, respectively (C). AD, Alzheimer’s disease; ANCOVA, analysis of covariance; ANG, angular gyrus; CAL, calcarine; CUN, cuneus; DCG, median cingulate and paracingulate gyrus; EOAD, early-onset AD; HES, Heschl gyrus; HIP, hippocampus; IPL, inferior parietal lobule; ITG, inferior temporal gyrus; LING, lingual gyrus; LOAD, late-onset AD; MOG, middle occipital gyrus; MTG, middle temporal gyrus; OC, old controls; OLF, olfactory cortex; ORBsupmed, superior frontal gyrus, medial orbital; PCUN, precuneus; PHG, parahippocampal gyrus; PoCG, postcentral gyrus; PUT, putamen; REC, gyrus rectus; ROL, Rolandic operculum; SMA, supplementary motor area; SMG, supramarginal gyrus; SOG, superior occipital gyrus; SPG, superior parietal gyrus; STG, superior temporal gyrus; TPOmid, Temporal pole, middle temporal gyrus; TPOsup, Temporal pole, superior temporal gyrus; YC, young controls.

    The clustering coefficients in the right CAL, left CUN, bilateral IPL, right ITG, right MOG, right SMG, and left TPOsup were significantly lower in the EOAD group than in YC (Fig. 2D). No significant AD-associated changes were found between the LOAD group and OC.

    Correlation between cognitive function and WM connectivity

    The relationships between cognitive function and brain-network alterations in EOAD and LOAD patients throughout were explored by calculating the correlations between cognitive function scores and network parameters. In the EOAD group, MMSE scores were negatively correlated with the global path length (Spearman’s rho=-0.508, p<0.001) and positively correlated with the global efficiency (Spearman’s rho=0.548, p<0.001), averaged local efficiency (Spearman’s rho=0.602, p<0.001), and averaged clustering coefficient (Spearman’s rho=0.586, p<0.001). Moreover, the CDR-SOB scores in the EOAD group were positively correlated with the global path length (Spearman’s rho=0.455, p=0.001) and negatively correlated with the global efficiency (Spearman’s rho=-0.475, p<0.001), averaged local efficiency (Spearman’s rho=-0.537, p<0.001), and averaged clustering coefficient (Spearman’s rho=-0.533, p<0.001) (Table 3).

    Table 3
    Relationships between global cognitive function and network parameters in Alzheimer’s disease (AD)

    In the LOAD group, no significant relationships were found between MMSE scores and the network parameters. In contrast, the CDR-SOB scores were positively correlated with the global path length (Spearman’s rho=0.393, p=0.024) and negatively correlated with the global efficiency (Spearman’s rho=-0.426, p=0.013), averaged local efficiency (Spearman’s rho=-0.512, p=0.002), and averaged clustering coefficient (Spearman’s rho=-0.497, p=0.003) (Table 3).

    The relationships between detailed neuropsychological test results and the network parameters are presented in Supplementary Tables 1, 2, 3, 4 (in the online-only Data Supplement). In the EOAD group there were significant negative associations between the global path length and the cognitive scores in the domains of attention (Digit Span Test, backward), language (Korean version of the Boston Naming Test [K-BNT]), verbal memory (Seoul Verbal Learning Test [SVLT], immediate recall; SVLT, recognition), and frontoexecutive function (Controlled Oral Word Association Test [COWAT], animal names; COWAT, supermarket; Stroop Test, reading colored words) (Supplementary Table 1 in the online-only Data Supplement). Global efficiency was positively correlated with these parameters (Supplementary Table 2 in the online-only Data Supplement). Averaged local efficiency was positively correlated with attention (Digit Span Test, backward), language (K-BNT), visuospatial (Rey-Osterrieth Complex Figure Test [RCFT], copying), memory (SVLT, immediate recall; SVLT, recognition; RCFT, recognition), and frontoexecutive function (COWAT, animal names; COWAT, supermarket; COWAT, phonemic items; Stroop Test, reading colored words) (Supplementary Table 3 in the online-only Data Supplement). Finally, the averaged clustering coefficient was significantly positively correlated with the same parameters as for averaged local efficiency, except for the RCFT, recognition (Supplementary Table 4 in the online-only Data Supplement). None of the specific cognitive results were correlated with the network features in the LOAD group.

    Correlations of tau and amyloid burdens with WM connectivity

    To reveal the distinct network disruption patterns in EOAD and LOAD patients, the correlations between pathological markers and the network parameters were computed in each group. The local efficiency in the EOAD group was inversely proportional to the THK retention in the right ANG, right triangular part of the inferior frontal gyrus, left IPL, right ITG, right MTG, bilateral posterior cingulate gyrus (PCG), right dorsolateral superior frontal gyrus, and bilateral SMG (Fig. 3A). The local efficiency in the LOAD group was negatively correlated with the THK retention in the right ITG, left inferior occipital gyrus, right TPOmid, right MTG, right PCUN, and left SPG (Fig. 3A).

    Fig. 3
    Relationships between network parameters and [18F]THK5351 (THK) retention in EOAD and LOAD patients. Spearman correlation analysis was used to identify significant relationships of THK retention with local efficiency (A) and clustering coefficient (B) in EOAD and LOAD patients. Colored nodes indicate significant differences at the false-discovery-rate-corrected p<0.05 level. ANG, angular gyrus; EOAD, early-onset Alzheimer’s disease; IFGtriang, triangular part of the inferior frontal gyrus; IOG, inferior occipital gyrus; IPL, inferior parietal lobule; ITG, inferior temporal gyrus; LOAD, late-onset Alzheimer’s disease; MTG, middle temporal gyrus; PCG, posterior cingulate gyrus; PCUN, precuneus; SFGdor, dorsolateral superior frontal gyrus; SMG, supramarginal gyrus; SPG, superior parietal gyrus; TPOmid, Temporal pole, middle temporal gyrus.

    The clustering coefficient decreased significantly as THK retention increased in the left IPL, right PCG, right MTG, and right ITG in EOAD patients (Fig. 3B). Significant correlations between clustering coefficient and THK retention were found in the left SPG, left SMG, and right ITG in LOAD patients (Fig. 3B).

    The amyloid burden showed no significant correlations with network disruption in either AD group.

    DISCUSSION

    This study applied graph-based network analysis to investigate changes in the organization of the brain network from an efficiency perspective. A graph-theoretic analysis demonstrated that the EOAD group showed significant deterioration in both global and local WM network parameters (longer global path length, and lower global efficiency, local efficiency, and clustering coefficient) compared with the age-matched control group. In contrast, the LOAD group did not show significant differences relative to the age-matched control group, although there was a trend similar to that in EOAD patients. Aging itself has been shown to affect WM networks, including in reducing global and local efficiencies in several previous studies.25, 26, 27, 28 This aging effect may be one of the reasons why the difference between the LOAD group and OC was less pronounced than that between the EOAD group and YC in the present study. However, some previous studies have also found significant differences.10, 29 It is possible that statistically significant differences could not be identified in our study due to the small sample size. We included subjects with normal cognitive function as normal controls, but this was not restricted to subjects who were negative for amyloid or tau. Two of the 26 OC (7.7%) were amyloid-positive. In addition, since we included clinically diagnosed probable AD, 2 of the 33 LOAD patients (6%) had amyloid negative results. It is possible that they had non-AD pathology, which could have also affected the results.

    When compared with each age-matched control, WM network disruption was more widespread in EOAD than LOAD patients. This widespread WM network disruption might be associated with a wider range of cognitive dysfunction (including in nonmemory domains) in EOAD than LOAD patients.2 In EOAD patients there were strong correlations between local WM network parameters (averaged local efficiency and clustering coefficient) and cognitive test results in all domains including attention, language, visuospatial, memory, and frontal/executive function as well as MMSE and CDR-SOB scores; these correlations were not present in LOAD patients.

    The local network disruption was limited to a smaller area (left ITG and the right SMA, putamen, and occipital gyrus) in the LOAD group. The ITG was one of the most-affected areas in the LOAD group and is associated with clinical impairment in AD.30, 31 Local WM network disruption in the SMA and putamen has also been reported previously in AD.32 The SMA is one of the neural correlates of verbal short-term memory in AD,33 and the putamen has also been reported to be associated with cognitive impairment in AD.34 A previous WM network study found decreased local efficiency in the right occipital area,35 similarly to our results. The occipital gyrus is a part of visual network, and visual network disruption in AD has been observed in some previous studies.36, 37, 38 Compared with EOAD patients, LOAD patients showed significantly lower local efficiency in the temporal area (left ITG and right STG and TPOmid), which matched the predominant temporal involvement in LOAD patients.39 There seemed to be a strong association between WM disruption and cognitive function in EOAD patients but not in LOAD patients. This might be due to the high likelihood that various factors in LOAD patients such as the cognitive reservoir, vascular changes, and glial senescence are involved in cognitive function in addition to structural brain changes.40

    In the present study, only tau, but not amyloid, was associated with reduced efficiency of WM connectivity networks in both AD groups. Local efficiency in several parietotemporal areas was associated with the local THK retention in the corresponding areas in both AD groups. This correlation was also observed in some parts of the frontal lobe and the PCG in EOAD patients. The greater extent of reduced local efficiency corresponding to THK retention in EOAD patients might have been mainly due to the pattern of THK retention itself. Previous studies have found that tau pathology in neocortical regions is more widespread in EOAD patients than in LOAD patients even at similar disease stages, and this pattern of tau pathology has been considered a key factor of distinct clinical features of EOAD patients.41, 42, 43 Previous postmortem AD studies found that tau pathology was related to the Wallerian degeneration protease calpain and alteration in WM microstructural integrity, suggesting an association between tau and WM degeneration in AD.44, 45 Axonal degeneration associated with tau pathology might also be responsible for disruption of WM integrity.46 Recent in vivo studies using [18F]AV-1451 (flortaucipir) have also produced evidence of tau-related WM alterations.47, 48, 49 In these recent studies, tau was associated with DTI measures of WM integrity, including an increase in the mean diffusivity, whereas amyloid was not associated with any measures of WM integrity. In line with these previous studies, our results also suggest that WM degeneration is more strongly associated with tau than with amyloid.

    The present results should be interpreted with caution. First, the sample in each group was relatively small. Further studies with larger samples are warranted to verify our findings. Second, the acquisition scheme of DTI employed in this work was a single b-value diffusion acquisition (i.e., a single shell) due to restricted scanning times. Although this method is valuable and has been widely used in clinical practice, the tensor model oversimplified the diffusion process for a complex WM fiber organization. More-advanced DTI acquisition methods (i.e., multiple shells and multiple b values) with a diffusion model can take the complexity of WM into account. This approach would also provide more-detailed features of the cellular environment, thus producing more-accurate data on structural connectivity in the brain. Third, we should acknowledge that the THK PET tracer is known to trace not only neurofibrillary tangles but also a combination of neurofibrillary tangles and astrocytosis.14, 50 Further studies utilizing tau PET tracers that are more specific to neurofibrillary tangles such as the [18F]MK-6240 PET tracer would be helpful for verifying our findings.

    In conclusion, EOAD patients demonstrated more-pronounced and widespread impairment of structural network efficiency compared with the age-matched controls, whereas LOAD patients did not. The structural network efficiency was strongly correlated with the profile of cognitive impairment only in EOAD patients. WM connectivity disruption as evaluated by the network efficiency was probably affected by tau but not by amyloid.

    Supplementary Materials

    The online-only Data Supplement is available with this article at https://doi.org/10.3988/jcn.2023.0092.

    Supplementary Table 1

    Relationship between neuropsychological test results and the network parameters: global path length

    Click here to view.(23K, pdf)

    Supplementary Table 2

    Relationship between neuropsychological test results and the network parameters: global efficiency

    Click here to view.(23K, pdf)

    Supplementary Table 3

    Relationship between neuropsychological test results and the network parameters: averaged local efficiency

    Click here to view.(23K, pdf)

    Supplementary Table 4

    Relationship between neuropsychological test results and the network parameters: averaged clustering coefficient

    Click here to view.(23K, pdf)

    Supplementary Fig. 1

    Comparison of the [18F]THK5351 and [18F]flutemetamol retentions between early-onset Alzheimer’s disease (EOAD) and late-onset Alzheimer’s disease (LOAD) patients. OC, old controls; YC, young controls.

    Click here to view.(586K, pdf)

    Notes

    Author Contributions:

    • Conceptualization: Young Noh.

    • Data curation: Sang-Young Kim, Woo-Ram Kim.

    • Formal analysis: Suyeon Heo, Sang-Young Kim.

    • Funding acquisition: Young Noh.

    • Investigation: Suyeon Heo, Sang-Young Kim.

    • Methodology: Suyeon Heo, Sang-Young Kim, Young Noh.

    • Project administration: Young Noh.

    • Resources: Cindy W Yoon, Woo-Ram Kim, Young Noh.

    • Software: Sang-Young Kim, Woo-Ram Kim.

    • Supervision: Duk L. Na, Young Noh.

    • Validation: Woo-Ram Kim.

    • Visualization: Suyeon Heo.

    • Writing—original draft: Suyeon Heo, Cindy W Yoon.

    • Writing—review & editing: Sang-Young Kim, Woo-Ram Kim, Duk L. Na, Young Noh.

    Conflicts of Interest:The authors have no potential conflicts of interest to disclose.

    Funding Statement:This study was supported by a grant from the Korea Healthcare Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare of the Republic of Korea (grant No: HI14C1135), a grant from the Basic Science Research Program through the National Research Foundation of Korea (NRF) that is funded by the Ministry of Education (2021R1A6A1A03038996) and a grant supported by Gachon University Gil Medical Center (Grant number: FRD2022-16).

    Availability of Data and Material

    The data that support the findings of this study are available from the corresponding author, upon reasonable request.

    References

      1. Koedam EL, Lauffer V, van der Vlies AE, van der Flier WM, Scheltens P, Pijnenburg YA. Early-versus late-onset Alzheimer’s disease: more than age alone. J Alzheimers Dis 2010;19:1401–1408.
      1. Smits LL, Pijnenburg YA, Koedam EL, van der Vlies AE, Reuling IE, Koene T, et al. Early onset Alzheimer’s disease is associated with a distinct neuropsychological profile. J Alzheimers Dis 2012;30:101–108.
      1. Jacobs D, Sano M, Marder K, Bell K, Bylsma F, Lafleche G, et al. Age at onset of Alzheimer’s disease: relation to pattern of cognitive dysfunction and rate of decline. Neurology 1994;44:1215–1220.
      1. Bronge L, Bogdanovic N, Wahlund LO. Postmortem MRI and histopathology of white matter changes in Alzheimer brains. A quantitative, comparative study. Dement Geriatr Cogn Disord 2002;13:205–212.
      1. Stahl R, Dietrich O, Teipel SJ, Hampel H, Reiser MF, Schoenberg SO. White matter damage in Alzheimer disease and mild cognitive impairment: assessment with diffusion-tensor MR imaging and parallel imaging techniques. Radiology 2007;243:483–492.
      1. Brun A, Englund E. A white matter disorder in dementia of the Alzheimer type: a pathoanatomical study. Ann Neurol 1986;19:253–262.
      1. Mayo CD, Garcia-Barrera MA, Mazerolle EL, Ritchie LJ, Fisk JD, Gawryluk JR, et al. Relationship between DTI metrics and cognitive function in Alzheimer’s disease. Front Aging Neurosci 2019;10:436
      1. Li KC, Luo X, Zeng QZ, Xu XJ, Huang PY, Shen ZJ, et al. Distinct patterns of interhemispheric connectivity in patients with early-and late-onset Alzheimer’s disease. Front Aging Neurosci 2018;10:261
      1. Lee WJ, Yoon CW, Kim SW, Jeong HJ, Seo S, Na DL, et al. Effects of Alzheimer’s and vascular pathologies on structural connectivity in early- and late-onset Alzheimer’s disease. Front Neurosci 2021;15:606600
      1. Reijmer YD, Leemans A, Caeyenberghs K, Heringa SM, Koek HL, Biessels GJ, et al. Disruption of cerebral networks and cognitive impairment in Alzheimer disease. Neurology 2013;80:1370–1377.
      1. Tierney MC, Fisher RH, Lewis AJ, Zorzitto ML, Snow WG, Reid DW, et al. The NINCDS-ADRDA Work Group criteria for the clinical diagnosis of probable Alzheimer’s disease: a clinicopathologic study of 57 cases. Neurology 1988;38:359–364.
      1. Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 1987;149:351–356.
      1. Seo SW, Cho SS, Park A, Chin J, Na DL. Subcortical vascular versus amnestic mild cognitive impairment: comparison of cerebral glucose metabolism. J Neuroimaging 2009;19:213–219.
      1. Ahn HJ, Chin J, Park A, Lee BH, Suh MK, Seo SW, et al. Seoul Neuropsychological Screening Battery-dementia version (SNSB-D): a useful tool for assessing and monitoring cognitive impairments in dementia patients. J Korean Med Sci 2010;25:1071–1076.
      1. Thurfjell L, Lilja J, Lundqvist R, Buckley C, Smith A, Vandenberghe R, et al. Automated quantification of 18F-flutemetamol PET activity for categorizing scans as negative or positive for brain amyloid: concordance with visual image reads. J Nucl Med 2014;55:1623–1628.
      1. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002;15:273–289.
      1. Wang R, Benner T, Sorensen AG, Wedeen VJ. Diffusion toolkit: a software package for diffusion imaging data processing and tractography. Proc Intl Soc Mag Reson Med 2007;15:3720
      1. Torgerson CM, Irimia A, Leow AD, Bartzokis G, Moody TD, Jennings RG, et al. DTI tractography and white matter fiber tract characteristics in euthymic bipolar I patients and healthy control subjects. Brain Imaging Behav 2013;7:129–139.
      1. Kim HJ, Im K, Kwon H, Lee JM, Kim C, Kim YJ, et al. Clinical effect of white matter network disruption related to amyloid and small vessel disease. Neurology 2015;85:63–70.
      1. Cho EB, Han CE, Seo SW, Chin J, Shin JH, Cho HJ, et al. White matter network disruption and cognitive dysfunction in neuromyelitis optica spectrum disorder. Front Neurol 2018;9:1104
      1. Tuch DS, Wedeen VJ, Dale AM, George JS, Belliveau JW. Conductivity tensor mapping of the human brain using diffusion tensor MRI. Proc Natl Acad Sci U S A 2001;98:11697–11701.
      1. Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations. Neuroimage 2010;52:1059–1069.
      1. Lo CY, Wang PN, Chou KH, Wang J, He Y, Lin CP. Diffusion tensor tractography reveals abnormal topological organization in structural cortical networks in Alzheimer’s disease. J Neurosci 2010;30:16876–16885.
      1. Bullmore E, Sporns O. Complex brain networks: graph theoretical analysis of structural and functional systems. Nat Rev Neurosci 2009;10:186–198.
      1. Bi Q, Wang W, Niu N, Li H, Wang Y, Huang W, et al. Relationship between the disrupted topological efficiency of the structural brain connectome and glucose hypometabolism in normal aging. Neuroimage 2020;226:117591
      1. Gong G, Rosa-Neto P, Carbonell F, Chen ZJ, He Y, Evans AC. Ageand gender-related differences in the cortical anatomical network. J Neurosci 2009;29:15684–15693.
      1. Wen W, Zhu W, He Y, Kochan NA, Reppermund S, Slavin MJ, et al. Discrete neuroanatomical networks are associated with specific cognitive abilities in old age. J Neurosci 2011;31:1204–1212.
      1. Zhao T, Cao M, Niu H, Zuo XN, Evans A, He Y, et al. Age-related changes in the topological organization of the white matter structural connectome across the human lifespan. Hum Brain Mapp 2015;36:3777–3792.
      1. Reid AT, Evans AC. Structural networks in Alzheimer’s disease. Eur Neuropsychopharmacol 2013;23:63–77.
      1. Scheff SW, Price DA, Schmitt FA, Scheff MA, Mufson EJ. Synaptic loss in the inferior temporal gyrus in mild cognitive impairment and Alzheimer’s disease. J Alzheimers Dis 2011;24:547–557.
      1. Halawa OA, Gatchel JR, Amariglio RE, Rentz DM, Sperling RA, Johnson KA, et al. Inferior and medial temporal tau and cortical amyloid are associated with daily functional impairment in Alzheimer’s disease. Alzheimers Res Ther 2019;11:14
      1. Lama RK, Lee SW. White matter network alterations in Alzheimer’s disease patients. Appl Sci 2020;10:919
      1. Peters F, Collette F, Degueldre C, Sterpenich V, Majerus S, Salmon E. The neural correlates of verbal short-term memory in Alzheimer’s disease: an fMRI study. Brain 2009;132(Pt 7):1833–1846.
      1. de Jong LW, van der Hiele K, Veer IM, Houwing JJ, Westendorp RG, Bollen EL, et al. Strongly reduced volumes of putamen and thalamus in Alzheimer’s disease: an MRI study. Brain 2008;131(Pt 12):3277–3285.
      1. Wang T, Shi F, Jin Y, Yap PT, Wee CY, Zhang J, et al. Multilevel deficiency of white matter connectivity networks in Alzheimer’s disease: a diffusion MRI study with DTI and HARDI models. Neural Plast 2016;2016:2947136
      1. Li Y, Yao Z, Yu Y, Zou Y, Fu Y, Hu B, et al. Brain network alterations in individuals with and without mild cognitive impairment: parallel independent component analysis of AV1451 and AV45 positron emission tomography. BMC psychiatry 2019;19:165
      1. Li Y, Wang X, Li Y, Sun Y, Sheng C, Li H, et al. Abnormal restingstate functional connectivity strength in mild cognitive impairment and its conversion to Alzheimer’s disease. Neural Plast 2016;2016:4680972
      1. Franciotti R, Falasca NW, Arnaldi D, Famà F, Babiloni C, Onofrj M, et al. Cortical network topology in prodromal and mild dementia due to Alzheimer’s disease: graph theory applied to resting state EEG. Brain Topogr 2019;32:127–141.
      1. Cavedo E, Pievani M, Boccardi M, Galluzzi S, Bocchetta M, Bonetti M, et al. Medial temporal atrophy in early and late-onset Alzheimer’s disease. Neurobiol Aging 2014;35:2004–2012.
      1. Lau V, Ramer L, Tremblay MÈ. An aging, pathology burden, and glial senescence build-up hypothesis for late onset Alzheimer’s disease. Nat Commun 2023;14:1670
      1. Schöll M, Ossenkoppele R, Strandberg O, Palmqvist S, Jögi J, et al. Swedish Bio-FINDER . Distinct 18F-AV-1451 tau PET retention patterns in early-and late-onset Alzheimer’s disease. Brain 2017;140:2286–2294.
      1. Cho H, Choi JY, Lee SH, Lee JH, Choi YC, Ryu YH, et al. Excessive tau accumulation in the parieto-occipital cortex characterizes early-onset Alzheimer’s disease. Neurobiol Aging 2017;53:103–111.
      1. Noh Y, Na HK, Seo S, Lee SY, Jeong HJ, Okamura N, et al. [P3–338] In vivo tau PET imaging in early-onset Alzheimer’s disease and late-onset Alzheimer’s disease. Alzheimers Dement 2017;13(7S_Part_22):1083–1083.
      1. McAleese KE, Walker L, Graham S, Moya ELJ, Johnson M, Erskine D, et al. Parietal white matter lesions in Alzheimer’s disease are associated with cortical neurodegenerative pathology, but not with small vessel disease. Acta Neuropathol 2017;134:459–473.
      1. Kantarci K, Murray ME, Schwarz CG, Reid RI, Przybelski SA, Lesnick T, et al. White-matter integrity on DTI and the pathologic staging of Alzheimer’s disease. Neurobiol Aging 2017;56:172–179.
      1. Terwel D, Dewachter I, Van Leuven F. Axonal transport, tau protein, and neurodegeneration in Alzheimer’s disease. Neuromolecular Med 2002;2:151–165.
      1. Wen Q, Risacher SL, Xie L, Li J, Harezlak J, Farlow MR, et al. Tau-related white-matter alterations along spatially selective pathways. Neuroimage 2021;226:117560
      1. Strain JF, Smith RX, Beaumont H, Roe CM, Gordon BA, Mishra S, et al. Loss of white matter integrity reflects tau accumulation in Alzheimer disease defined regions. Neurology 2018;91:e313–e318.
      1. Pereira JB, Ossenkoppele R, Palmqvist S, Strandberg TO, Smith R, Westman E, et al. Amyloid and tau accumulate across distinct spatial networks and are differentially associated with brain connectivity. Elife 2019;8:e50830
      1. Harada R, Ishiki A, Kai H, Sato N, Furukawa K, Furumoto S, et al. Correlations of 18F-THK5351 PET with postmortem burden of tau and astrogliosis in Alzheimer disease. J Nucl Med 2018;59:671–674.
    MeSH Terms
    Alzheimer Disease
    Amyloid
    Brain
    Cognition
    Humans
    Mental Status and Dementia Tests
    Neurofibrillary Tangles
    Neuropsychological Tests
    Metrics
    Share
    Figures
    slide
    slide
    slide

    1 / 3

    Tables
    slide
    slide
    slide

    1 / 3

    ORCID IDs
    Funding Information
    PERMALINK