Table 1.
Demographics and clinical characteristics of the control subjects and RRMS patients.
Figure 1.
The union of the DMN mask identified and extracted with independent component analysis from RRMS and healthy control groups (LH: left hemisphere; RH: right hemisphere).
Figure 2.
A flowchart of calculating the correlation between structural connectivity and functional connectivity within the default mode network (DMN) subregions.
(1) DMN extraction was performed by ICA. (2) The subregions were extracted as regions of interest (ROIs) within the DMN. (3) The time series in each ROI was extracted with the DMN. (4) The temporal correlation coefficients between each ROI pair were quantified within the DMN. (5) Preprocessed motion and eddy current distortion correction of the model distributions of the relevant parameter (Monte Carlo sampling) was performed using FMRIB's diffusion toolbox (FDT v2.0). (6) Pair-wise ROI (AND logic) probabilistic tractography was used to calculated the distribution of fiber orientations. (7) Features of the long white matter tracts (above threshold 0.2) remaining fiber bundles connecting each pair of ROIs were compared between the two groups, including the strength of a pathway and the above-threshold standard DTI parameters.
Figure 3.
A set of matrices of functional and structural analyses of pair-wise default mode network subregions in the RRMS patient and healthy control (HC) groups.
The figure (a) is a correlation matrix of average time series of subregions (ROIs) in the network (subregion labels are on the left and bottom). The figures (b–s) show the strength of the structural connectivity matrix [of average log (N track) and volumes] of each pair-wise connection. The figures (d–g) are above-threshold (0.2) standard DTI parameters: including the fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) of each pair-wise connection. The black arrow indicates increased and the white arrow decreased alteration in the patients with RRMS. Notably, structural connectivity was only compared without the dashed boxes.
Figure 4.
An example of the long white matter fibers of pair-wise subregions within the default mode network detected by probabilistic tractography in one control subject (a) and one RRMS patient (c) in the native space.
Figures (b) and (d) correspond to 3-dimensional probabilistic tractography. (red: fiber tract, blue: mask of subregions, Sup = Superior, Inf = Inferior, Ant = Anterior, Pos = Posterior).
Figure 5.
The Pearson correlation between structural connectivity (SC) indices (y axis) and functional connectivity (FC) indices (x axis) of pair-wise subregions within the DMN of RRMS patients.
The FC value positively correlated with the MD value (a) and FC value (b) of the tract between the PCC/PCUN and the MPFC in RRMS patients (P<0.05, Bonferroni corrected). The FC values was positively correlated with the volume (c) of the tract between the PCC/PCUN and the left mTL in RRMS patients (P<0.05, Bonferroni corrected). The FC value positively correlated with the MD value (d) of the tract between the PCC/PCUN and right mTL (P<0.05, Bonferroni corrected). AD = axial diffusivity, IPL = inferior parietal lobule, MD = mean diffusivity, mTL = medial temporal lobes, MPFC = medial prefrontal cortex, PASAT = paced auditory serial addition test, PCC/PCUN = posterior cingulate cortex/precuneus, RRMS = relapsing–remitting multiple sclerosis.
Figure 6.
The Pearson correlation between structural/functional connectivity indices (x axis) and clinical markers (y axis) of pair-wise subregions within the DMN of RRMS patients.
The functional connectivity (FC) value (a) between the PCC/PCUN and the right mTL negatively correlated with the PASAT in RRMS patients (P<0.05, Bonferroni corrected). The EDSS score negatively correlated with the log (N track) of the tract between the PCC/PCUN and the left IPL (b) and negatively correlated with the FA value of the tract between the PCC/PCUN and the left mTL (c) in RRMS patients (P<0.05, Bonferroni corrected). The EDSS score positively correlated with the MD value of the tract between the MPFC and the left IPL (d) in RRMS patients (P<0.05, Bonferroni corrected). The PASAT score positively correlated with the FA value of the tract between the PCC/PCUN and the right IPL (e) in RRMS patients (P<0.05, Bonferroni corrected). The MFIS score negatively correlated with the MD values of the tract between the MPFC and the left IPL (f) in RRMS patients (P<0.05, Bonferroni corrected). The BPF score positively correlated with the FA value of the tract between the MPFC and the left IPL (g) in RRMS patients (P<0.05, Bonferroni corrected). The BPF score positively correlated with the volume of the tract between the PCC/PCUN and the MPFC (h) in RRMS patients (P<0.05, Bonferroni corrected). The BPF score positively correlated with the log (N track) value between the PCC/PCUN and the left mTL (i) in RRMS patients (P<0.05, Bonferroni corrected). AD = axial diffusivity, EDSS = expanded disability status scale, FA = fractional anisotropy, IPL = inferior parietal lobule, MD = mean diffusivity, mTL = medial temporal lobes, MPFC = medial prefrontal cortex, PASAT = paced auditory serial addition test, PCC/PCUN = posterior cingulate cortex/precuneus, RRMS = relapsing–remitting multiple sclerosis, TWMLL = total white matter lesion loads.