Convergent Multimodal Imaging Abnormalities in the Dorsal Precuneus in Subjective Cognitive Decline

Abstract

Background:

A range of imaging modalities have reported Alzheimer’s disease-related abnormalities in individuals experiencing subjective cognitive decline (SCD). However, there has been no consistent local abnormality identified across multiple neuroimaging modalities for SCD.

Objective:

We aimed to investigate the convergent local alterations in amyloid-β (Aβ) deposition, glucose metabolism, and resting-state functional MRI (RS-fMRI) metrics in SCD.

Methods:

Fifty SCD patients (66.4±5.7 years old, 19 men [38%]) and 15 normal controls (NC) (66.3±4.4 years old, 5 men [33.3%]) were scanned with both [¹⁸F]-florbetapir PET and [¹⁸F]-fluorodeoxyglucose PET, as well as simultaneous RS-fMRI from February 2018 to November 2018. Voxel-wise metrics were retrospectively analyzed, including Aβ deposition, glucose metabolism, amplitude of low frequency fluctuation (ALFF), regional homogeneity (ReHo), and degree centrality(DC).

Results:

The SCD group showed increased Aβ deposition and glucose metabolism (p < 0.05, corrected), as well as decreased ALFF, ReHo, and DC (p < 0.05, uncorrected) in the left dorsal precuneus (dPCu). Furthermore, the dPCu illustrated negative resting-state functional connectivity with the default mode network. Regarding global Aβ deposition positivity, the Aβ deposition in the left dPCu showed a gradient change, i.e., Aβ positive SCD > Aβ negative SCD > Aβ negative NC. Additionally, both Aβ positive SCD and Aβ negative SCD showed increased glucose metabolism and decreased RS-fMRI metrics in the dPCu.

Conclusions:

The dorsal precuneus, an area implicated in early AD, shows convergent neuroimaging alterations in SCD, and might be more related to other cognitive functions (e.g., unfocused attention) than episodic memory.

Keywords

Alzheimer’s disease functional magnetic resonance imaging positron emission tomography precuneus subjective cognitive decline

INTRODUCTION

Subjective cognitive decline (SCD) refers to self-perceived cognitive decline relative to a previous status in the absence of measurable objective cognitive impairment. 1 SCD may represent the first symptomatic manifestation of Alzheimer’s disease (AD) and indicate a higher risk for a future objective cognitive impairment. 1 Precise localization of abnormal activity in the SCD stage not only helps understand the early pathophysiology of dementia, but also benefits the identification of potential targets for future precision treatment, e.g., focused brain stimulation therapy. 2 Recently, functional neuroimaging techniques, including amyloid-β (Aβ) PET, tau protein PET, [¹⁸F]-fluorodeoxyglucose (FDG)-PET, and resting-state functional MRI (RS-fMRI), have been used in SCD studies. However, as summarized in Supplementary Table 1, there has been no consensus on where the abnormal activity is in SCD.

Aβ-PET serves as a diagnostic tool across the AD spectrum and is widely employed in SCD research. However, it should be noted that most previous studies have assessed Aβ burden at the whole-brain scale or large regional scale, which inherently limits the precise localization of the brain region affected by Aβ pathology. To date, only five voxel-level studies have been published but did not reach consistent conclusions regarding regions with significant Aβ deposition in SCD (as summarized in Supplementary Table 1).3 –7 Abnormal glucose metabolism from FDG-PET has not been consistent in terms of brain regions in previous SCD studies, e.g., hypometabolism in the precuneus and parietotemporal and parahippocampal gyrus, hypermetabolism in the medial temporal lobe, or no significant finding (see Supplementary Table 1 for details).

The RS-fMRI provides an effective and non-invasive method to measure blood oxygen level-dependent (BOLD) signal as a proxy for spontaneous brain activity.8,9, 8,9 Over the last decades, RS-fMRI has gained popularity for characterizing brain organization and function and their associations with various behavioral and clinical conditions.10 –14 A few RS-fMRI metrics depict the local or regional brain activity from multiple complementary dimensions and facilitate the precise localization of the abnormality, which is vital for precision treatment. These metrics include amplitude of low frequency fluctuation (ALFF), 15 regional homogeneity (ReHo), 16 and degree centrality (DC), 17 which depict the neuronal activity, short-range functional connectivity (FC), and whole brain FC of the given voxel, respectively. For example, a study reported reduced ALFF in the left precuneus, left paracentral lobule, and left posterior cingulate cortex (PCC) and reduced ReHo in the left inferior parietal gyrus, right PCC, and right superior occipital gyrus, when contrasting SCD individuals with normal controls (NC). 18

The current study collected data with a PET/MR hybrid scanner and employed data-driven approaches to examine the local activities with Aβ deposition, glucose metabolism, and three RS-fMRI local metrics, i.e., ALFF, ReHo, and DC in SCD and NC. Coordinate-based meta-analyses of previous single neuroimaging modality studies have suggested that the PCC and its adjacent precuneus (usually named PCC/precuneus in the literature) showed increased Aβ deposition, decreased glucose metabolism, decreased ALFF, and decreased ReHo in the mild cognitive impairment (MCI) or AD.19 –22 Here, we hypothesized that: 1) SCD would show concurrent abnormalities across multiple neuroimaging modalities in AD-affected brain regions (e.g., PCC/precuneus); 2) the concurrent abnormal region may belong to the default mode network; 3) local abnormalities may be more sensitive than global Aβ cortical burden.

METHODS

Participants

Participants were recruited through public advertisements and referrals from general physicians, memory clinics, and informants. All participants signed written informed consent before the experiment. The recruitment and assessment of participants have been described in the protocol of the Sino Longitudinal Study on Cognitive Decline (SILCODE) previously. 23 All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was registered on ClinicalTrials.gov and approved by the ethics committee of Xuanwu Hospital of Capital Medical University (Identifier No. [2017]046). The participants were recruited from February 2018 to November 2018, and finally 65 right-handed and Mandarin speaking participants (50 for SCD and 15 for NC) were included. The 15 individuals of the control group represented all available data in our dataset. Figure 1 illustrates the participant selectionprocess.

Fig. 1

The flowchart for participant selection. SCD, subjective cognitive decline; NC, normal control; AV45, [¹⁸F] florbetapir; FDG, [¹⁸F] fluorodeoxyglucose.

SCD was defined according to the research criteria for pre-MCI SCD, including 1) self-experienced persistent decline in cognitive capacity compared with a previously normal status and unrelated to an acute event; 2) normal age-, sex-, and education-adjusted performance on standardized cognitive tests and not meeting the criteria for MCI or dementia.1,23, 1,23 Individuals with normal performance on the standardized neuropsychological tests and without cognitive problems via the SCD-interview were included as the NC group. 24

The exclusion criteria were as follows: 1) a history of stroke; 2) current major psychiatric diagnoses, such as severe depression and anxiety (Hamilton Depression Scale (HAMD) score > 24 or Hamilton Anxiety Scale (HAMA) score > 29); 3) other neurological conditions that could cause cognitive decline (e.g., brain tumors, Parkinson’s disease, encephalitis, or epilepsy) other than AD spectrum disorders; 4) other diseases that could cause cognitive decline (e.g., thyroid dysfunction, severe anemia, syphilis, or HIV); 5) cognitive decline caused by traumatic brain injury; 6) history of psychosis or congenital mental developmental delay; 7) contraindications for MRI or inability to complete the study protocol.

Neuropsychological assessments

The neuropsychological tests included three cognitive domains: Auditory Verbal Learning Test-Huashan version (AVLT-H), Animal Fluency Test (AFT), 30-item Boston Naming Test (BNT), and Shape Trail Test (STT) parts A and B. In addition, all subjects complemented tests for global cognition, daily life ability, and neuropsychiatric assessments, including Montreal Cognitive Assessment-Basic (MoCA-B), Functional Activities Questionnaire (FAQ), HAMA, and HAMD.

Image data acquisition

For each participant, PET and MRI data were simultaneously acquired with an integrated 3.0 Tesla TOF PET/MR scanner (SIGNA PET/MR, GE Healthcare, Milwaukee, WI, USA). Participants were instructed to keep their eyes closed, relax but not fall asleep, and move as little as possible during the scanning. Foam pads were used to minimize head movements and earplugs were used to reduce machine noise.

PET data acquisition

Participants were scanned with [¹⁸F] florbetapir (AV45) and [¹⁸F]-fluorodeoxyglucose (FDG)-PET in a 3D acquisition mode about four weeks apart (24.4±23.8 days). For FDG-PET, participants were instructed to fast for at least 6 hours. The imaging parameters were listed in the Supplementary Methods.

MRI data acquisition

We obtained two fMRI datasets for each participant, hereafter referred to fMRI_1 (simultaneously collected with AV45-PET) and fMRI_2 (simultaneously collected with FDG-PET). Thus, the corresponding fMRI parameters derived from the fMRI_1 and fMRI_2 datasets are referred to as ALFF_1, ReHo_1, DC_1, and ALFF_2, ReHo_2, DC_2, respectively. The detailed scanning parameters for the 3D-T1 and RS-fMRI images were in the Supplementary Methods.

Data analysis

PET data processing

AV45-PET and FDG-PET images were analyzed by Statistical Parametric Mapping version 12 (SPM 12; https://www.fil.ion.ucl.ac.uk/spm/software/spm12) and in-house script based on MATLAB (https://www.mathworks.com). The 3D-T1 images were firstly registrated to PET images and segmented. 25 Then, the PET images were spatially normalized to Montreal Neurological Institute (MNI) standard space with the forward deformation field estimated during the unified segmentation, resampled to an isotropic voxel size of 3×3×3 mm³, and smoothed with a Gaussian kernel with a full width at half maximum (FWHM) of 10 mm. Finally, the voxel-wise standardized uptake value ratio (SUVR) was obtained by choosing the whole cerebellum (for AV45-PET) and pons and vermis (for FDG-PET) as the reference regions. 26 The global AV45-SUVR was computed by averaging the SUVR in a set of regions including the frontal, temporal, parietal and cingulate cortices. 27

RS-fMRI data preprocessing

RS-fMRI datasets were preprocessed using the Data Processing Assistant for Brain Imaging (DPABI_V4.1) and Resting-State fMRI Data Analysis Toolkit (RESTplus_1.22).28,29, 28,29 The first 10 volumes were discarded as the participants have not get used to the scanning noise. The remaining volumes were corrected for slice time differences and then realigned to the first volume for motion correction. Next, the 3D-T1 images were coregistered to the averaged functional image and segmented. The motion-corrected functional volumes were spatially normalized to MNI standard space with the forward deformation field from unified segmentation and resampled to 3×3×3 mm³. The resulting images were spatially smoothed with a Gaussian kernel of 6 mm FWHM. It should be noted that the spatial smoothing was performed before the calculation of ALFF and seed-based resting-sate functional connectivity (RSFC), but after the calculation of ReHo and DC. Then, the linear trend of the time courses was removed. Nuisance signals (including Friston-24 head motion parameters, white matter signal, and cerebrospinal fluid signal) were extracted and regressed out from the data. Finally, band-pass filtering (0.01 – 0.08 Hz) was performed for before ReHo and DC calculation, but not for ALFF calculation.

ALFF, ReHo, and DC calculations

Three RS-fMRI metrics (i.e., ALFF, ReHo, and DC) depicting the local brain activity from multiple complementary dimensions were calculated. The ALFF characterizes the amplitude of the fluctuation of each time series of the fMRI signal and reflects the brain activity strength. 30 The ReHo characterizes the local synchronization of the time series of neighboring voxels and high ReHo suggests high functional synchronization with its neighbors. 16 The DC characterizes the functional connectivity density of a voxel with all other voxels and high DC illustrates widespread and strong communication between this voxel and the rest of the brain. 17 It should be noted that, DC is actually a functional connectivity (FC) index, but often taken as a local metric in graph theory analysis.

ALFF analysis: Before performing ALFF analysis, spatial smoothing (Gaussian kernel with 6 mm FWHM) was performed. The time course of each voxel was transformed into the frequency domain with the fast Fourier transform. The square root of the power spectrum was calculated and averaged across 0.01–0.08 Hz. 15 This averaged square root was the ALFF value for a given voxel. For standardization, each voxel’s ALFF value was divided by the global mean ALFF resulting in an mALFF map.

ReHo analysis: Kendall’s coefficient of concordance (KCC) was computed on the ranked time course of a given voxel and its 26 nearest neighbors. 16 The obtained KCC was the ReHo value. Then, mReHo maps were obtained as mALFF maps.

DC analysis: DC represents the functional connection strength between a given voxel and all the voxels in the brain. 17 First, we computed the Pearson correlation between the time course of a given voxel and all the other voxels in the whole brain. Then, the Pearson correlation coefficients larger than 0.2 were summed and the then assigned to that given voxel as the DC value. Standardized mDC maps were created as mALFF. It should be noted that, DC is actually a functional connectivity (FC) index, but often taken as a local metric in graph theory analysis.

Seed-based resting-sate functional connectivity calculations

The seed-based RSFC was conducted post hoc. The RSFC quantifies the connection between the region of interest (ROI) and the whole brain, which is widely applied to identify the network of which the ROI belongs. For example, if the ROI has strong positive connectivity with brain regions in the default mode network (DMN), then the ROI may functionally cooperate with DMN. Inversely, if the ROI has strong negative connectivity with the network, suggesting antistatic function between them. The seed was defined based on the comparison results of image metrics between SCD and NC groups (Fig. 2). As shown in the Alteration of local activity section of the results (Fig. 2), an area in the left dorsal precuneus (6 voxels), which showed abnormal activity in 7 of the total 8 neuroimaging metrics in SCD subjects, was selected as the ROI (referred to as the dorsal precuneus ROI or dPCu-ROI hereafter). First, the global mean signal was extracted and regressed out from the data. Individual RSFC maps of the dPCu-ROI were generated by calculating Pearson’s correlation coefficients between the mean time course of the dPCu-ROI and the time course of each voxel in the whole brain. Subsequently, subject-level RSFC maps were converted into z-value maps (zRSFC) with Fisher’s transformation to improve the normality.

Fig. 2

Group differences on image metrics between subjective cognitive decline (SCD) and normal control (NC) groups with controlling for age, sex, and years of education. T maps of Aβ deposition and glucose metabolism (top row), RS-fMRI metrics (middle row), and the overlapping area of the PET and fMRI metrics (bottom row) are depicted. Overlapping area at the left dPCu means that seven of eight metrics showed group difference, i.e., AV45-SUVR, FDG-SUVR, ALFF_1, ReHo_1, DC_1, ALFF_2, ReHo_2, and DC_2. The fMRI_1 dataset was acquired simultaneously with AV45-PET; the fMRI_2 dataset was acquired simultaneously with FDG-PET. Red and yellow colors indicate increased metrics in SCD versus NC, and blue color indicates the opposite. SUVR, standardized uptake value ratio; FDR, false discovery rate; ALFF, amplitude of low frequency fluctuation; ReHo, regional homogeneity; DC, degree centrality; L and R, left and right in the brain; T, the T value of two-sample t-test on the PET and fMRI metrics maps.

Statistical analysis

SPSS (version 24.0, IBM) was used for comparisons of demographic and neuropsychological data. DPABI_V4.1 was used for statistical analysis of the local activities and FC between SCD and NC.

Demographic and neuropsychological data comparison

Group comparisons on demographic and neuropsychological data were performed with two-sample t-test and Chi-square analyses.

Local activity alteration

Two-sample t-tests were conducted between SCD group and NC group on eight local neuroimaging metrics (AV45-SUVR, FDG-SUVR, ALFF_1, ReHo_1, DC_1, ALFF_2, ReHo_2, and DC_2) with the age, gender, and years of education as covariates. False discovery rate (FDR) (p < 0.05) was used for multiple comparison correction. Subject numbers for each functional neuroimaging modality analysis were summarized in Supplementary Table 2.

Functional connectivity alteration

One-sample t-tests were performed on the zRSFC maps of fMRI_1 and fMRI_2 datasets for all subjects, respectively. Regions showing significant correlation (p < 0.05, FDR corrected) were then entered into two-sample t-tests between SCD group and NC group with the age, gender, and years of education as covariates. FDR (p < 0.05) was used for multiple comparison correction.

Correlation between neuroimaging metrics and behavioral measures

Partial correlation analysis was performed to investigate the relationship between the eight local neuroimaging metrics (mean values in the dPCu-ROI of AV45-SUVR, FDG-SUVR, ALFF_1, ReHo_1, DC_1, ALFF_2, ReHo_2, and DC_2) and neuropsychological measures (AVLT-H delayed recall, AVLT-H recognition, STT A and B, AFT, BNT, MoCA-B, HAMD, and HAMA) in the SCD group with the age, gender, and education level as covariates.

Sub-groups of positivity and negativity of Aβ deposition

Participants were categorized as Aβ positive if their global AV45-SUVR value was ≥1.11, 27 resulting in the identification of Aβ negative NC (NC^-, n = 11), Aβ positive NC (NC⁺, n = 4), Aβ negative SCD (SCD^-, n = 22), and Aβ positive SCD (SCD⁺, n = 28). The NC⁺ group was excluded due to the small sample size (n = 4). Group comparisons on demographic, neuropsychological data, global AV45-SUVR, and mean values of the eight neuroimaging metrics in the dPCu-ROI were performed using one-way ANOVA with least significant difference (LSD) post-hoc test. Mean values of fMRI metrics were further compared among the three groups using the analysis of covariance (ANCOVA), controlling for global AV45-SUVR and the left dPCu-ROI AV45-SUVR, respectively. Furthermore, we also compared the mean values of the eight metrics in the PCC ROI (available for download on the LONI website), 26 one of the most important abnormal regions indicative of pathological changes in MCI and AD.

RESULTS

Behavioral measures

No significant differences were found in the age, sex, or education level between the SCD group and NC group (all p > 0.05) (Table 1). The HAMA and HAMD scores were significantly higher in the SCD group than in the NC group, but no subjects met the diagnostic criteria for anxiety or depression.

Table 1

Demographic information and neuropsychological tests

	SCD (N = 50)	NC (N = 15)	Statistics
Age (y)	66.4±5.7	66.3±4.4	t ₍₆₃₎ = 0.04, p = 0.97
Gender (M/F)	19/31	5/10	χ ² ₍₁₎ = 0.11, p = 0.74
Education (y)	13.8±3.0	12.8±2.5	t ₍₆₃₎ = 1.16, p = 0.25
HAMD	3.8±3.7	1.1±1.3	t ₍₆₃₎ = 2.81, p = 0.007^*
HAMA	4.5±3.9	2.1±2.4	t ₍₆₃₎ = 2.25, p = 0.03^*
FAQ	0.2±0.5	0.0±0.0	t ₍₆₁₎ = 1.63, p = 0.11^†
AVLT-H delayed recall	7.7±2.0	7.3±1.9	t ₍₆₃₎ = 0.70, p = 0.48
AVLT-H recognition	22.3±1.4	22.5±1.6	t ₍₆₃₎=–0.30, p = 0.77
STT part A	55.5±17.7	53.2±14.5	t ₍₆₃₎ = 0.46, p = 0.65
STT part B	128.7±37.0	128.5±29.1	t ₍₆₃₎ = 0.03, p = 0.98
AFT	20.0±4.7	19.5±4.8	t ₍₆₃₎ = 0.40, p = 0.69
BNT	26.0±2.4	25.5±3.2	t ₍₆₃₎ = 0.74, p = 0.46
MoCA-B	26.3±2.1	26.1±1.9	t ₍₆₃₎ = 0.46, p = 0.65

Data were presented as mean±SD. SCD: subjective cognitive decline. NC, normal control; HAMD, Hamilton depression scale; HAMA, Hamilton anxiety scale; FAQ, functional activities questionnaire; AVLT-H, auditory verbal learning test-Huashan version; STT, shape trails test; AFT, animal fluency test; BNT, Boston naming test; MoCA-B, Montreal cognitive assessment-basic. ^* p < 0.05, Comparison between SCD and NC. ^†Missing data for 1 SCD and 1 NC subjects.

Alteration of local activity

The SCD group showed: 1) increased Aβ deposition in the left dPCu, left cuneus, left superior parietal cortex, and right medial superior frontal cortex (Supplementary Figure 1 and Fig. 2; p < 0.05, FDR corrected); 2) increased glucose metabolism in the left dPCu, left middle occipital cortex, and bilateral paracentral lobule (Supplementary Figure 1 and Fig. 2; p < 0.05, FDR corrected). The left dPCu was the only region showing overlapped abnormality in both Aβ deposition and glucosemetabolism.

The differences in RS-fMRI metrics (ALFF, ReHo, and DC) did not survive FDR correction. However, with an uncorrected p < 0.05, the SCD group showed decreased ALFF, ReHo, and DC in the left dPCu (Fig. 2) for both the fMRI_1 and fMRI_2 datasets.

Seed-based functional connectivity

One sample t-tests showed that the connectivity patterns of the left dPCu were very similar for fMRI_1 and fMRI_2 datasets (p < 0.05, FDR corrected, Fig. 3). Significantly negative functional connectivity with the left dPCu was mainly observed in the regions of the default mode network, including the bilateral medial prefrontal cortex, posterior cingulate cortex, and its adjacent ventral precuneus. Significantly positive functional connectivity was observed in the bilateral inferior frontal cortex, middle frontal cortex, lingual gyrus. No significant difference in RSFC was found between the SCD group and NC group (p < 0.05, FDRcorrected).

Fig. 3

Seed-based resting-state functional connectivity (RSFC) patterns of the left dorsal precuneus (dPCu) of all the subjects. Patterns is depicted in fMRI_1 dataset (A) and fMRI_2 dataset (B) (p < 0.05, FDR corrected). The fMRI_1 dataset was acquired simultaneously with AV45-PET; the fMRI_2 dataset was acquired simultaneously with FDG-PET. The yellow-red color indicates positive connections and the blue color indicates negative connections with the left dPCu. MPFC, medial prefrontal cortex; PCC, posterior cingulate cortex; vPCu, ventral precuneus; IPS, intraparietal sulcus; L and R, left and right in the brain; T, the T value of one-sample t-test on the RSFC maps.

Correlation between imaging metrics and behavioral measures in the SCD group

We observed significant correlation between among pairs of fMRI metrics (ALFF, ReHo, and DC; p < 0.001) within the dPCu-ROI. No significant correlation was found between AV45-SUVR and fMRI metrics, between FDG-SUVR and fMRI metrics, or between AV45-SUVR and FDG-SUVR (Supplementary Table 3).

Regarding the correlation between behavioral scales and imaging metrics, although a few correlations showed a p value between 0.05 – 0.01, none survived multiple comparison correction (Supplementary Table S4).

Aβ deposition positivity and neuroimaging metrics in the dPCu-ROI

The demographic and neuropsychological data of NC^-, SCD^- and SCD⁺ groups can be found in Supplementary Table 5. The SCD⁺ group showed significantly higher global AV45-SUVR than both NC^- and SCD^- groups, but no significant difference was found between NC^- and SCD^- (p = 0.43) (Fig. 4A). Both the SCD^- and SCD⁺ groups had significantly higher AV45-SUVR in left dPCu-ROI relative to the NC^- group, while no significant difference was found between SCD^- and SCD⁺ (p = 0.08) (Fig. 4B). Regarding the FDG-SUVR and RS-fMRI metrics, both the SCD⁺ and SCD^- groups showed hypermetabolism and decreased fMRI activity in the dPCu-ROI, except for the comparison of ReHo_2 between the SCD⁺ group and NC^- group (p = 0.11) (Fig. 4). Most of the significant group differences in the fMRI metrics remained after controlling for global AV45-SUVR or the left dPCu-ROI AV45-SUVR (Supplementary Tables 6 and 7).

Fig. 4

Comparisons of global AV45-SUVR and the eight metrics in the left dPCu-ROI among three groups. Global AV45-SUVR (A), AV45-SUVR (B), FDG-SUVR (C), ALFF_1 (D), ReHo_1 (E), DC_1 (F), ALFF_2 (G), ReHo_2 (H), and DC_2 (I). The fMRI_1 dataset was acquired simultaneously with AV45-PET; the fMRI_2 dataset was acquired simultaneously with FDG-PET. NC^-, Aβ negative NC; SCD–, Aβ negative SCD; SCD⁺, Aβ positive SCD. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001. All box and whisker plots: box range, 25–75%; whisker range, 5–95%.

Independent verification of data analyses

As the replicability of fMRI research has recently been called into question, 31 replicating the data analysis and results by another independent research team (i.e., independent verification) helps ensure the accuracy and replicability of the code and data. Specifically, we sent the raw data and code to a research team from Shanghai University (authors Chang-Chang Ding and Jie-Hui Jiang) to independently perform the analysis procedure described in the “Methods” section. The comparison between their results and ours illustrated substantial similarity (Supplementary Figures 2–4 and Supplementary Tables 8 and 9).

DISCUSSION

Convergent abnormal local activity in SCD

The SCD group exhibited increased Aβ deposition in regions including the left dPCu, left cuneus, left superior parietal cortex, and right medial superior frontal cortex, which are early affected by amyloid pathology. 32 We observed hypermetabolism in the left dPCu, left middle occipital cortex, and bilateral paracentral lobule in the SCD group, while evidence regarding alterations in glucose metabolism in SCD remains inconclusive. 33 Notably, the left dPCu displayed converging abnormal activities across three distinct neuroimaging modalities: Aβ-PET, FDG-PET, and fMRI. This contrasts with prior reports indicating that, in the PCC and its adjacent precuneus (sometimes named PCC/precuneus) of MCI and AD, meta-analyses of single neuroimaging modality studies have reported increased Aβ deposition and decreased glucose metabolism,19,34, 19,34 decreased RS-fMRI ALFF, 21 and ReHo, 22 separately. Furthermore, the seed-based RS-fMRI FC showed that the dPCu was negatively connected with the default mode network (including PCC and its adjacent ventral precuneus), which demonstrated that the dPCu is not a region in the default mode network. Previous RS-fMRI studies have segmented the precuneus into six or eight sub-regions,35,36, 35,36 suggesting that the precuneus may be functionally heterogeneous. Moreover, we searched Neurosynth (http://neurosynth.org) (coordinates: –6, –68, 62; distance: 4 mm) and found that the dPCu was reported to be involved in more cognitive than memory tasks, e.g., unfocused attention, 37 in healthy people. Several task fMRI studies have been conducted on SCD (Supplementary Table 10). Notably, studies by Billette et al. and Corriveau-Lecavalier et al. provide evidence of elevated memory task-related fMRI activity in the precuneus (though without further parcellation) among individuals with SCD relative to NC.38,39, 38,39 Furthermore, their findings indicate that brain activation in the precuneus follows an inverted U-shaped pattern across the clinical spectrum of AD. Future studies are necessary to investigate the cognitive function of the dPCu to expand the current understanding of its role in SCD, and hence develop more specific cognitive tests for SCD.

Regarding the seed-based FC of the dPCu from RS-fMRI, no significant difference was found between the SCD group and NC group, illustrating the preservation of neuronal connectivity of the dPCu with other brain regions in SCD. Thus, the alterations in other local measurements, such as Aβ deposition, glucose metabolism, and fMRI local activity, in the dPCu, might not have caused measurable loss of the neuronal connectivity at the stage of SCD. However, this is hypothetical and should be investigated in future studies.

Increased Aβ deposition versus increased glucose metabolism in SCD

Meta-analysis has reported that increased Aβ deposition was associated with decreased glucose metabolism in the PCC in MCI/AD, 19 indicating decreased neuronal activity. However, we found an association of increased Aβ deposition with increased glucose metabolism in the dPCu, indicating increased activity. Our results are consistent with previous findings that suggest a trend towards increased glucose metabolism in APOE ɛ4 + SCD participants with significant Aβ deposition compared with APOE ɛ4- SCD participant. 5 Recent studies provide compelling evidences that soluble Aβ dimers can induce hyperexcitation in sensitive neurons by blocking glutamate reuptake and form a vicious cycle of hyperactivation to drive the disease process.40 –42 Our findings support the hypothesis that Aβ-related neuronal hypermetabolism may be a potential key feature of the early stage of AD.

Previous studies have shown inconsistent findings regarding glucose metabolism among SCD subjects (Supplementary Table 1), which may be attributed to following reasons: 1) SCD is a heterogenous population, tau-related or other pathologies can cause dysfunction. 4 2) SCD participants in different studies may be in different stages of Aβ deposition. Since the Aβ-dependent neuronal activity may initially increase and then gradually decreased,40,41,43 , 40,41,43 FDG findings in SCD subjects could show hypermetabolism, hypometabolism, or even no significant difference. Thus, longitudinal studies monitoring the evolution of Aβ deposition and neural dysfunction may help clarify this issue.

Decreased value of RS-fMRI local metrics vs. increased glucose metabolism in SCD

As aforementioned, for MCI and AD, both decreased glucose metabolism and decreased value of RS-fMRI local metrics (ALFF and ReHo) in the PCC were interpreted as indicating decreased neuronal activity. However, we found seemingly discrepant results, namely increased glucose metabolism and decreased value of the local RS-fMRI metrics in the dPCu. The following notations may help explain this observation.

Mathematically, the glucose metabolism measured by FDG-PET reflects the integrated or mean value over a period of time, while the RS-fMRI ALFF reflects the amplitude of fluctuation or standard deviation within a specific frequency band. The simulation result has shown that the mean value has no linear correlation with the standard deviation (Supplementary Figure 5). It has been commonly suggested that the fMRI signal is more coupled with the local field potential than the action potential. 44 Mathematically, both the RS-fMRI signal and the local field potential are analyzed in a frequency-dependent manner, however, their frequency-dependent correlation of the amplitude, a popular metric for both RS-fMRI and local field potential, is far from clear.

Several PET-fMRI studies have indicated that, although both the glucose metabolism and RS-fMRI local metrics show similar higher activity in the default mode network regions (e.g., PCC) than other brain regions, 45 voxel-wise across-subject correlation did not reveal robust significant linear correlation (Supplementary Table 11).46 –48

Brain disorders may exhibit opposite association between glucose metabolism and RS-fMRI local metrics, e.g., increased glucose metabolism but decreased ALFF in the putamen of patients with Parkinson’s disease (Supplementary Table 12).49,50, 49,50 Similarly, our study also found an opposite association in the dPCu. However, as aforementioned, MCI/AD shows a positive association, i.e., decreased glucose metabolism and decreased RS-fMRI metrics.19,21,22 , 19,21,22

Considering the non-invasiveness, low-cost, and easy-accessing advantages of RS-fMRI, it represents a promising technique for precisely localizing the AD-related abnormal activity and could be used to guide precise brain stimulation, e.g., transcranial magnetic stimulation.

The dorsal precuneus observed in the present study may belong to parietal memory network

Based on the location and RSFC pattern analysis, the region vulnerable to AD pathology observed in the present study may be more likely to belong to parietal memory network (PMN) rather than default mode network (DMN). Recent studies indicate that the precuneus is a key node of the PMN, involved in memory encoding and retrieval tasks.51,52, 51,52 However, due to the effects of spatial smoothing and parametric settings for group independent component analysis, PMN can be easily mislabeled as a posterior part of DMN. 52 Moreover, in a cross-lifespan fMRI dataset (7–85 years old), PMN may by separable from the DMN in youth, and its functional role tends to diminish with increasing age. 53 Together with previous reports that PMN is disrupted in AD, 54 our findings further emphasize the need to consider the PMN and the DMN separately in future AD neuroimaging studies.

Amyloid-negative SCD showed regional amyloid accumulation, hypermetabolism and fMRI abnormalities

Binary classification of Aβ deposition into positive and negative is a popular way for clinical diagnosis in both healthy aging and individuals along the AD continuum. 55 However, autopsy and PET evidence reveal that initial Aβ deposits aggregate in a more focal manner compared to the more diffuse patterns observed at later stages. 56 As a result, such early Aβ deposition might not meet the study-specific thresholds based on global Aβ cortical burden and are considered “amyloid negative”. The current study classified all participants into SCD⁺, SCD^-, NC⁺, and NC^- with the criteria of the cortical summary AV45-SUVR value≥1.11. 27 The NC⁺ group was excluded from the further analysis due to the limited sample size (n = 4). Both the SCD^- and SCD⁺ groups showed increased Aβ deposition and glucose metabolism and decreased RS-fMRI metrics in the dPCu compared with the NC^- group. In other words, like the SCD⁺ group, the SCD^- group also exhibited elevated Aβ pathology and hypermetabolism, and abnormal fMRI activities. Although the Aβ+/– categorization is often used to define preclinical AD, it does not imply that globally Aβ negative subjects have no pathology in the brain. 55 Consistent with our findings, previous literatures have also indicated that local regions (including the precuneus and posterior cingulate, etc.) may be more sensitive in detecting early Aβ deposition than global cortical summary regions.56,57, 56,57 Furthermore, we compared the mean values of eight metrics in the PCC (Supplementary Figure 6), and the SCD⁺ group showed significantly higher AV45-SUVR than the SCD^- and NC^- group. However, we did not find significant difference in the FDG-SUVR and fMRI metrics of the PCC between the NC^- group and the two SCD groups. Together, it is likely that the increased Aβ deposition may occur concurrently with hypermetabolism and abnormal fMRI activities in the dPCu (an early Aβ-affected region) in the early stage of AD.

Limitations

This study has several limitations. Firstly, it is a cross-sectional study, further longitudinal studies would be helpful to investigate the changes in Aβ deposition, neuronal dysfunction, and cognitive function as the disease progresses. Secondly, tau PET imaging was not conducted here to evaluate tau pathology. Thirdly, although multimodal metrics and two fMRI datasets were used to increase the reliability of our results, data from multiple research centers are needed to validate the findings of this study. Fourthly, cognitive tests related to unfocused attention were not included, which might provide more specific insights into the functional roles of the dPCu. Fifthly, due to the requirement for high subject compliance for complete data collection, the resulting dataset is relatively small, particularly in the control group where participants did not have concerns about their cognitive status. The current sample size may impact the statistical analysis and generalizability of our findings. Lastly, ALFF might also reflect changes or differences in cerebrovascular reactivity, not only neuronal activity, 58 which requires caution when interpreting the results.

Conclusions

Our study highlights that the left dorsal precuneus as a region where SCD subjects, at increased risk for development of AD dementia, show convergent neuroimaging alterations compared to normal controls, including increased local Aβ deposition, glucose hypermetabolism, and altered intrinsic functional activity. The left dorsal precuneus negatively connects with the default mode network indicating that the dPCu is not a region within the default mode network. Our findings suggests the dPCu as a promising potential target for therapeutic interventions and furthers our understanding of its role in cognitive processing.

AUTHOR CONTRIBUTIONS

Xuan-Yu Li (Conceptualization; Data curation; Formal analysis; Investigation; Visualization; Writing – original draft); Li-Xia Yuan (Conceptualization; Data curation; Formal analysis; Methodology; Software; Visualization; Writing – original draft); Chang-Chang Ding (Formal analysis; Validation); Teng-Fei Guo (Methodology; Writing – review & editing); Wen-Ying Du (Data curation; Investigation; Writing – review & editing); Jie-Hui Jiang (Methodology; Supervision; Validation; Writing – review & editing); Frank Jessen (Conceptualization; Methodology; Supervision; Writing – review & editing); Yu-Feng Zang (Conceptualization; Funding acquisition; Methodology; Software; Visualization; Writing – review & editing); Ying Han (Conceptualization; Funding acquisition; Methodology; Project administration; Resources; Supervision; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

The authors would like to thank Professor Fei-Fan Zhou for her advice and guidance regarding the structure and presentation of the article.

FUNDING

This work was supported by the National Natural Science Foundation of China (grant numbers 82020108013, 82327809 and 81661148045), STI2030-Major Projects (2022ZD0211800), and Sino-German Cooperation Grant (M-0759), Shenzhen Bay Scholars Program and Tianchi Scholars Program.

CONFLICT OF INTEREST

The authors have no conflicts of interest to declare that are relevant to the content of this article. Ying Han is an Editorial Board Member of this journal but was not involved in the peer-review process of this article nor had access to any information regarding its peer-review.

DATA AVAILABILITY

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

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

References

Jessen

, Amariglio

, van Boxtel

, et al. A conceptual framework for research on subjective cognitive decline in preclinical Alzheimer’s disease. Alzheimers Dement 2014; 10: 844–852.

Chen

, Ma

, Hu

, et al. rTMS modulates precuneus-hippocampal subregion circuit in patients with subjective cognitive decline. Aging (Albany NY) 2020; 13: 1314–1331.

Yasuno

, Kazui

, Yamamoto

, et al. Resting-state synchrony between the retrosplenial cortex and anterior medial cortical structures relates to memory complaints in subjective cognitive impairment. Neurobiol Aging 2015; 36: 2145–2152.

, Luo

, Zeng

, et al. Aberrant functional connectivity network in subjective memory complaint individuals relates to pathological biomarkers. Transl Neurodegener 2018; 7: 27.

Risacher

, Kim

, Nho

, et al. APOE effect on Alzheimer’s disease biomarkers in older adults with significant memory concern. Alzheimers Dement 2015; 11: 1417–1429.

Snitz

, Lopez

, McDade

, et al. Amyloid-beta imaging in older adults presenting to a memory clinic with subjective cognitive decline: A pilot study. J Alzheimers Dis 2015; 48 Suppl 1: S151–159.

Ren

, Li

, Huang

, et al. Brain functional alterations and association with cognition in people with preclinical subjective cognitive decline and objective subtle cognitive difficulties. Neuroscience 2023; 513: 137–144.

Biswal

, Mennes

, Zuo

, et al. Toward discovery science of human brain function. Proc Natl Acad Sci U S A 2010; 107: 4734–4739.

Biswal

, Yetkin

, Haughton

, et al. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn Reson Med 1995; 34: 537–541.

10.

Cole

, Phillips

, Bentzley

, et al. Stanford Neuromodulation Therapy (SNT): A double-blind randomized controlled trial. Am J Psychiatry 2022; 179: 132–141.

11.

Genon

, Eickhoff

and Kharabian

. Linking interindividual variability in brain structure to behaviour. Nat Rev Neurosci 2022; 23: 307–318.

12.

Gordon

, Chauvin

, Van

, et al. A somato-cognitive action network alternates with effector regions in motor cortex. Nature 2023; 617: 351–359.

13.

Marek

, Tervo-Clemmens

, Calabro

, et al. Reproducible brain-wide association studies require thousands of individuals. Nature 2022; 603: 654–660.

14.

Zhang

and Raichle

. Disease and the brain’s dark energy. Nat Rev Neurol 2010; 6: 15–28.

15.

Zang

, He

, Zhu

, et al. Altered baseline brain activity in children with ADHD revealed by resting-state functional MRI. Brain Dev 2007; 29: 83–91.

16.

Zang

, Jiang

, Lu

, et al. Regional homogeneity approach to fMRI data analysis. Neuroimage 2004; 22: 394–400.

17.

Zuo

, Ehmke

, Mennes

, et al. Network centrality in the human functional connectome. Cereb Cortex 2012; 22: 1862–1875.

18.

Zhang

, Wang

, He

, et al. Altered regional cerebral blood flow and brain function across the Alzheimer’s disease spectrum: A potential biomarker. Front Aging Neurosci 2021; 13: 630382.

19.

, Liu

, Radua

, et al. Meta-analytic comparison between PIB-PET and FDG-PET results in Alzheimer’s disease and MCI. Cell Biochem Biophys 2015; 71: 17–26.

20.

Reiman

, Caselli

, Yun

, et al. Preclinical evidence of Alzheimer’s disease in persons homozygous for the epsilon 4 allele for apolipoprotein E. N Engl J Med 1996; 334: 752–758.

21.

Pan

, Zhu

, Yu

, et al. Aberrant spontaneous low-frequency brain activity in amnestic mild cognitive impairment: A meta-analysis of resting-state fMRI studies. Ageing Res Rev 2017; 35: 12–21.

22.

Zhen

, Xia

, Yi

, et al. Alterations of brain local functional connectivity in amnestic mild cognitive impairment. Transl Neurodegener 2018; 7: 26.

23.

, Wang

, Su

, et al. Sino Longitudinal Study on Cognitive Decline (SILCODE): Protocol for a Chinese longitudinal observational study to develop risk prediction models of conversion to mild cognitive impairment in individuals with subjective cognitive decline. BMJ Open 2019; 9: e028188.

24.

Jessen

, Spottke

, Boecker

, et al. Design and first baseline data of the DZNE multicenter observational study on predementia Alzheimer’s disease (DELCODE). Alzheimers Res Ther 2018; 10: 15.

25.

Ashburner

and Friston

. Unified segmentation. Neuroimage 2005; 26: 839–851.

26.

Landau

, Mintun

, Joshi

, et al. Amyloid deposition, hypometabolism, and longitudinal cognitive decline. Ann Neurol 2012; 72: 578–586.

27.

Landau

, Breault

, Joshi

, et al. Amyloid-beta imaging with Pittsburgh compound B and florbetapir: Comparing radiotracers and quantification methods. J Nucl Med 2013; 54: 70–77.

28.

Yan

, Wang

, Zuo

, et al. DPABI: Data processing & analysis for (resting-state) brain imaging. Neuroinformatics 2016; 14: 339–351.

29.

Jia

, Wang

, Sun

, et al. RESTplus: An improved toolkit for resting-state functional magnetic resonance imaging data processing. Sci Bull (Beijing) 2019; 64: 953–954.

30.

Yang

, Long

, Yang

, et al. Amplitude of low frequency fluctuation within visual areas revealed by resting-state functional MRI. Neuroimage 2007; 36: 144–152.

31.

Dubois

and Adolphs

. Building a science of individual differences from fMRI. Trends Cogn Sci 2016; 20: 425–443.

32.

Grothe

, Barthel

, Sepulcre

, et al. In vivo staging of regional amyloid deposition. Neurology 2017; 89: 2031–2038.

33.

Drzezga

, Altomare

, Festari

, et al. Diagnostic utility of 18F-Fluorodeoxyglucose positron emission tomography (FDG-PET) in asymptomatic subjects at increased risk for Alzheimer’s disease. Eur J Nucl Med Mol Imaging 2018; 45: 1487–1496.

34.

Franzmeier

, Duering

, Weiner

, et al. Left frontal cortex connectivity underlies cognitive reserve in prodromal Alzheimer disease. Neurology 2017; 88: 1054–1061.

35.

Luo

, Zeng

, Qin

, et al. Functional parcellation of human brain precuneus using density-based clustering. Cereb Cortex 2020; 30: 269–282.

36.

Zhang

and Li

. Functional connectivity mapping of the human precuneus by resting state fMRI. Neuroimage 2012; 59: 3548–3562.

37.

Dickenson

, Berkman

, Arch

, et al. Neural correlates of focused attention during a brief mindfulness induction. Soc Cogn Affect Neurosci 2013; 8: 40–47.

38.

Billette

, Ziegler

, Aruci

, et al. Novelty-related fMRI responses of precuneus and medial temporal regions in individuals at risk for Alzheimer disease. Neurology 2022; 99: e775–e788.

39.

Corriveau-Lecavalier

, Duchesne

, Gauthier

, et al. A quadratic function of activation in individuals at risk of Alzheimer’s disease. Alzheimers Dement (Amst) 2020; 12: e12139.

40.

Zott

, Simon

, Hong

, et al. A vicious cycle of beta amyloid-dependent neuronal hyperactivation. Science 2019; 365: 559–565.

41.

Busche

and Konnerth

. Neuronal hyperactivity–A key defect in Alzheimer’s disease? Bioessays 2015; 37: 624–632.

42.

, Madison

, Baker

, et al. Dynamic relationships between age, amyloid-beta deposition, and glucose metabolism link to the regional vulnerability to Alzheimer’s disease. Brain 2016; 139: 2275–2289.

43.

Zott

, Busche

, Sperling

, et al. What happens with the circuit in Alzheimer’s disease in mice and humans? Annu Rev Neurosci 2018; 41: 277–297.

44.

Logothetis

, Pauls

, Augath

, et al. Neurophysiological investigation of the basis of the fMRI signal. Nature 2001; 412: 150–157.

45.

Raichle

, MacLeod

, Snyder

, et al. A default mode of brain function. Proc Natl Acad Sci U S A 2001; 98: 676–682.

46.

Aiello

, Salvatore

, Cachia

, et al. Relationship between simultaneously acquired resting-state regional cerebral glucose metabolism and functional MRI: A PET/MR hybrid scanner study. Neuroimage 2015; 113: 111–121.

47.

Bernier

, Croteau

, Castellano

, et al. Spatial distribution of resting-state BOLD regional homogeneity as a predictor of brain glucose uptake: A study in healthy aging. Neuroimage 2017; 150: 14–22.

48.

Jiao

, Gao

, Shi

, et al. Frequency-dependent relationship between resting-state fMRI and glucose metabolism in the elderly. Front Neurol 2019; 10: 566.

49.

Tomse

, Jensterle

, Grmek

, et al. Abnormal metabolic brain network associated with Parkinson’s disease: Replication on a new European sample. Neuroradiology 2017; 59: 507–515.

50.

Wang

, Zhang

, Zang

, et al. Consistent decreased activity in the putamen in Parkinson’s disease: A meta-analysis and an independent validation of resting-state fMRI. Gigascience 2018; 7: giy071.

51.

Gilmore

, Nelson

and McDermott

. A parietal memory network revealed by multiple MRI methods. Trends Cogn Sci 2015; 19: 534–543.

52.

, Wang

, Li

, et al. Segregation between the parietal memory network and the default mode network: Effects of spatial smoothing and model order in ICA. Sci Bull (Beijing) 2016; 61: 1844–1854.

53.

Yang

, Chang

, Xu

, et al. Connectivity trajectory across lifespan differentiates the precuneus from the default network. Neuroimage 2014; 89: 45–56.

54.

, Du

, Zhang

, et al. Loss of parietal memory network integrity in Alzheimer’s disease. Front Aging Neurosci 2019; 11: 67.

55.

Jack

Jr , Bennett

, Blennow

, et al. A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology 2016; 87: 539–547.

56.

Farrell

, Chen

, Rundle

, et al. Regional amyloid accumulation and cognitive decline in initially amyloid-negative adults. Neurology 2018; 91: e1809–e1821.

57.

Guo

, Landau

, Jagust

, et al. Detecting earlier stages of amyloid deposition using PET in cognitively normal elderly adults. Neurology 2020; 94: e1512–e1524.

58.

Pinto

, Bright

, Bulte

, et al. Cerebrovascular reactivity mapping without gas challenges: A methodological guide. Front Physiol 2020; 11: 608475.