Graph theory is used to study brain connectivity, i.e. connectomes, estimated with diffusion magnetic resonance imaging (dMRI). Previous studies have already investigated the correlation between some network measures and the Expansion Disability Status Scale (EDSS), which assesses the clinical worsening of multiple sclerosis (MS) patients.
We investigated connectivity changes between healthy controls (HC) and relapsing remitting (RR) patients and tested whether such differences correlate with EDSS, comparing the effectiveness of various definitions of “connection strength” using different microstructural models.
dMRI was acquired for 67 HC (39F, 37±7yrs) and 49 RR (33F, 37±4yrs). Connectomes were created with deterministic tractography and weighting the connections by 1) number of streamlines (NOS) between grey-matter regions and, 2) mean value of quantitative scalar maps, estimated using state-of-the-art microstructural models, along the streamlines, notably: fractional anisotropy, FA; axial AD, radial RD and mean diffusivity MD; Intra Neurite and Isotropic Volume Fractions, ICVF and ISOVF; orientation dispersion, OD; Neurite volume fraction, INTRA; Extra-neurite transverse and mean diffusivity EXTRATRANS and EXTRAMD. We computed 5 network measures from each connectome: Density (ratio between actual and possible connections); Efficiency (capability of transferring and processing information); Modularity (network segregation); Clustering Coefficient (degree to which nodes tend to cluster together); Mean Strength (average of the sum of the edge weights connected to a node).
The network measures that significantly differ between the 2 groups were: Efficiency for ICVF p=0.031, AD p<0.01, RD p<0.01, EXTRATRANS p=0.019 and MD p<0.01 connectomes; Clustering Coefficient for AD p=0.015, RD p=0.013, EXTRATRANS p=0.021 and MD p<0.01 connectomes; Mean Strength for ICVF p=0.019, INTRA p=0.037, AD p=0.011, RD p<0.01, EXTRATRANS p=0.014 and MD p<0.01 connectomes. Only Modularity significantly correlate with EDSS for NOS p=0.047, FA p=0.049, ICVF p=0.041 and INTRA p=0.030 connectomes. All tests accounted for age, sex and density as confounding factor.
The maps discriminating more HC from MS patients were AD, RD, MD and EXTRATRANS. The microstructure features along the tracts with the highest correlation to EDSS were those investigating axonal integrity (FA, ICVF and INTRA). Modularity was the metric most correlated with EDSS.