Aberrant Functional Organization Within And Between Resting-state Networks In AD

Objective:This study combined multiple imaging modalities to investigate the functional and structural changes within each resting-state network (RSN) and between RSNs in patients with amnestic mild cognitive impairment (aMCI) or Alzheimer’s disease (AD).Subjects and Methods:A total of87older subjects were recruited in this study. Cognitive function was evaluated with the mini-mental state examination (MMSE), and the degree of dementia was determined by the clinical dementia rating scale (CDR). Thirteen subjects were excluded due to excessive head motion during MR scanning or poor image quality. In total,74subjects, including21NC,18aMCI, and35AD, were included in the further analyses. Resting-state functional magnetic resonance imaging (fMRI) scan and structural imaging were performed using a3.0T Signa MR scanner (Magnetom Trio, Siemens, Germany). During fMRI scans, all subjects were instructed to keep their eyes closed, to stay as motionless as possible, to think of nothing in particular, and not to fall asleep. The fMRI data were analyzed using the Data Processing Assistant for Resting-State fMRI (DPARSFA). Voxel-based morphometry (VBM) analysis was performed using Statistical Parametric Mapping (SPM8, http://www.fil.ion.ucl.ac.uk/spm/software/spm8) based on Matlab. The preprocessing of rs-fMRI data including slice timing, realignment, spatial normalization to the EPI template, resampling to2-mm3voxels, a band-pass frequency filter (0.01-0.08Hz) and smoothed with a4mm full-width at half-maximum (FWHM) Gaussian kernel. Finally, individual ALFF maps were generated by calculating the square root of the power spectrum at each frequency and averaged across0.01-0.08Hz for each voxel by the DPARSFA software. RSNs were identified from fMRI data using the group independent component analysis (GICA) of the fMRI toolbox (Stable and Consistent Group ICA of the fMRI Toolbox, version1.2; http://www.nitrc.org/projects/cogicat/). The preprocessing of high resolution structural data including segmentation of gray and white matter, spatial normalization and smooth with a FWHM kernel of4mm. The functional network connectivity (FNC) was computed for each pair of the8RSNs. For the intra-network analyses, we first compared the intra-network FCs among the three groups in a voxel-wise manner. brain regions whose intra-network FCs showed significant differences among the three groups were extracted as regions of interest (ROIs). The ALFF and GMV of each ROI were extracted and compared among the three groups. To exclude the atrophy effect on the FC and ALFF comparisons, we also repeated the ROI-based analyses of FC and ALFF by controlling for the GMV of each ROI. Finally, partial correlation analysis was performed to investigate the associations between the intra-network FC, ALFF/GMV of each ROI and the MMSE scores in AD and aMCI groups. Moreover, the same analyses for the functional measures were repeated after treating the GMV of each ROI as an additional covariate of no interest. Partial correlation analysis was also performed for the inter-network FC, with and without treating the mean GMV of the whole brain as an additional covariate of no interest.Results:The RSNs were the SMN, VN, AN, precuneus (Pcu), anterior (aDMN) and posterior (pDMN) DMN, and left (LFP) and right (RFP) FPN.Compared with the NC group, AD patients showed significantly decreased FNCs between the aDMN and pDMN and between the VN and LFP. The values of the aMCI group were numerically between the NC and AD groups. AD patients showed a trend toward decreased FNC between the aDMN and pDMN compared to the aMCI patients.For the intra-network FCs, AD patients showed significantly decreased FCs in the bilateral (B-Pcu) and right (R-Pcu) Pcu of the Pcu network, the posterior cingulate cortex (PCC) and left Pcu (L-Pcu) of the pDMN, and the left superior parietal lobule (L-SPL) of the left FPN compared with the NCs. Then, we extracted these significant brain regions as ROIs. All of the comparisons between NC and AD groups were significant. The FC values of these ROIs of aMCI patients were numerically between those of the NCs and AD patients. Compared with NCs, aMCI patients showed significantly decreased FC in the PCC. After GMV correction, compared with aMCI patients, AD patients showed only a trend toward decreased FCs in the ROIs of the pDMN, and the other results did not change.Compared with the NC, we found that the ALFF was significantly decreased in all five ROIs of the AD patients. Compared with aMCI patients, AD patients showed significantly decreased ALFF in the R-Pcu, B-Pcu, PCC and L-SPL and exhibited a trend toward decreased ALFF in the L-Pcu. The aMCI group showed a trend toward decreased ALFF in the PCC relative to the NC group. Notably, the ALFFs of the aMCI group were numerically between the NC and AD groups. After GMV correction, most of the between-group differences in ALFF remained.Compared with NCs, patients with AD showed significant GMV decreases in the PCC and L-Pcu, and a trend toward decreased GMV in the R-Pcu. Compared with aMCI patients, AD patients showed significantly decreased GMV in the PCC and L-Pcu. The aMCI group showed a trend toward decreased GMV in the PCC and L-Pcu compared with the NC group. The B-Pcu and L-SPL did not show any significant changes in GMV across the three groups.We found that most of the FCs, ALFFs, and GMVs of the ROIs were correlated with MMSE scores. After correction for GMV, most of the correlations of functional measures were significant.Conclusion:1. The Pcu, pDMN and LFP networks were selectively impaired in AD, and only the pDMN was impaired in MCI;2. Brain regions with impaired FC in AD also exhibited alterations in ALFF and GMV;3. Most of the functional changes were independent of the structural changes;4. FNCs between the aDMN and pDMN as well as between the VN and LFP were also changed in AD;5. Most of these MRI measures were associated with cognitive decline in older people.